deepgeek/gradio-webrtc
2
0
Fork 0
mirror of https://github.com/HumanAIGC-Engineering/gradio-webrtc.git synced 2026-02-05 18:09:23 +08:00
Code Issues Packages Projects Releases Wiki Activity

update v0.2.0

Browse Source
This commit is contained in:
杍超
2025-04-01 16:04:53 +08:00
parent f0b43b63cd 99a8e78e5e
commit b6d3e51b2d
198 changed files with 27674 additions and 2392 deletions
Download Patch File Download Diff File Expand all files Collapse all files

63
demo/app.py
View File

@@ -1,10 +1,17 @@
import asyncio
import base64
from io import BytesIO
import json
import math
import queue
import time
import uuid
import threading
from fastrtc.utils import Message
import gradio as gr
import numpy as np
from gradio_webrtc import (
from fastrtc import (
AsyncAudioVideoStreamHandler,
WebRTC,
VideoEmitType,
@@ -26,6 +33,7 @@ def encode_image(data: np.ndarray) -> dict:
base64_str = str(base64.b64encode(bytes_data), "utf-8")
return {"mime_type": "image/jpeg", "data": base64_str}
frame_queue = queue.Queue(maxsize=100)
class VideoChatHandler(AsyncAudioVideoStreamHandler):
def __init__(
@@ -38,7 +46,7 @@ class VideoChatHandler(AsyncAudioVideoStreamHandler):
input_sample_rate=24000,
)
self.audio_queue = asyncio.Queue()
self.video_queue = asyncio.Queue()
self.video_queue = frame_queue
self.quit = asyncio.Event()
self.session = None
self.last_frame_time = 0
@@ -50,6 +58,25 @@ class VideoChatHandler(AsyncAudioVideoStreamHandler):
output_frame_size=self.output_frame_size,
)
chat_id = ''
async def on_chat_datachannel(self,message: Message,channel):
# 返回
# {"type":"chat",id:"标识属于同一段话", "message":"Hello, world!"}
# {"type":"avatar_end"} 表示本次对话结束
if message['type'] == 'stop_chat':
self.chat_id = ''
channel.send(json.dumps({'type':'avatar_end'}))
else:
id = uuid.uuid4().hex
self.chat_id = id
data = message["data"]
halfLen = math.floor(data.__len__()/2)
channel.send(json.dumps({"type":"chat","id":id,"message":data[:halfLen]}))
await asyncio.sleep(5)
if self.chat_id == id:
channel.send(json.dumps({"type":"chat","id":id,"message":data[halfLen:]}))
channel.send(json.dumps({'type':'avatar_end'}))
async def video_receive(self, frame: np.ndarray):
# if self.session:
# # send image every 1 second
@@ -61,10 +88,11 @@ class VideoChatHandler(AsyncAudioVideoStreamHandler):
# print(frame.shape)
newFrame = np.array(frame)
newFrame[0:, :, 0] = 255 - newFrame[0:, :, 0]
self.video_queue.put_nowait(newFrame)
# self.video_queue.put_nowait(newFrame)
async def video_emit(self) -> VideoEmitType:
return await self.video_queue.get()
# print('123123',frame_queue.qsize())
return frame_queue.get()
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
frame_size, array = frame
@@ -114,14 +142,35 @@ with gr.Blocks(css=css) as demo:
},
}
)
handler = VideoChatHandler()
webrtc.stream(
VideoChatHandler(),
handler,
inputs=[webrtc],
outputs=[webrtc],
time_limit=150,
time_limit=1500,
concurrency_limit=2,
)
# 线程函数：随机生成 numpy 帧
def generate_frames(width=480, height=960, channels=3):
while True:
try:
# 随机生成一个 RGB 图像帧
frame = np.random.randint(188, 256, (height, width, channels), dtype=np.uint8)
# 将帧放入队列
frame_queue.put(frame)
# print("生成一帧数据，形状:", frame.shape, frame_queue.qsize())
# 模拟实时性：避免过度消耗 CPU
time.sleep(0.03) # 每秒约生成 30 帧
except Exception as e:
print(f"生成帧时出错: {e}")
break
thread = threading.Thread(target=generate_frames, daemon=True)
thread.start()
if __name__ == "__main__":
demo.launch()

15
demo/echo_audio/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Echo Audio
emoji: 🪩
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Simple echo stream - simplest FastRTC demo
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

45
demo/echo_audio/app.py Normal file
View File

@@ -0,0 +1,45 @@
import numpy as np
from fastapi import FastAPI
from fastapi.responses import RedirectResponse
from fastrtc import ReplyOnPause, Stream, get_twilio_turn_credentials
from gradio.utils import get_space
def detection(audio: tuple[int, np.ndarray]):
# Implement any iterator that yields audio
# See "LLM Voice Chat" for a more complete example
yield audio
stream = Stream(
handler=ReplyOnPause(detection),
modality="audio",
mode="send-receive",
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
)
app = FastAPI()
stream.mount(app)
@app.get("/")
async def index():
return RedirectResponse(
url="/ui" if not get_space() else "https://fastrtc-echo-audio.hf.space/ui/"
)
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

3
demo/echo_audio/requirements.txt Normal file
View File

@@ -0,0 +1,3 @@
fastrtc[vad]
twilio
python-dotenv

15
demo/gemini_audio_video/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Gemini Audio Video
emoji: ♊️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Gemini understands audio and video!
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GEMINI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

185
demo/gemini_audio_video/app.py Normal file
View File

@@ -0,0 +1,185 @@
import asyncio
import base64
import os
import time
from io import BytesIO
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from fastrtc import (
AsyncAudioVideoStreamHandler,
Stream,
WebRTC,
get_twilio_turn_credentials,
)
from google import genai
from gradio.utils import get_space
from PIL import Image
load_dotenv()
def encode_audio(data: np.ndarray) -> dict:
"""Encode Audio data to send to the server"""
return {
"mime_type": "audio/pcm",
"data": base64.b64encode(data.tobytes()).decode("UTF-8"),
}
def encode_image(data: np.ndarray) -> dict:
with BytesIO() as output_bytes:
pil_image = Image.fromarray(data)
pil_image.save(output_bytes, "JPEG")
bytes_data = output_bytes.getvalue()
base64_str = str(base64.b64encode(bytes_data), "utf-8")
return {"mime_type": "image/jpeg", "data": base64_str}
class GeminiHandler(AsyncAudioVideoStreamHandler):
def __init__(
self,
) -> None:
super().__init__(
"mono",
output_sample_rate=24000,
output_frame_size=480,
input_sample_rate=16000,
)
self.audio_queue = asyncio.Queue()
self.video_queue = asyncio.Queue()
self.quit = asyncio.Event()
self.session = None
self.last_frame_time = 0
self.quit = asyncio.Event()
def copy(self) -> "GeminiHandler":
return GeminiHandler()
async def start_up(self):
client = genai.Client(
api_key=os.getenv("GEMINI_API_KEY"), http_options={"api_version": "v1alpha"}
)
config = {"response_modalities": ["AUDIO"]}
async with client.aio.live.connect(
model="gemini-2.0-flash-exp", config=config
) as session:
self.session = session
print("set session")
while not self.quit.is_set():
turn = self.session.receive()
async for response in turn:
if data := response.data:
audio = np.frombuffer(data, dtype=np.int16).reshape(1, -1)
self.audio_queue.put_nowait(audio)
async def video_receive(self, frame: np.ndarray):
if self.session:
# send image every 1 second
print(time.time() - self.last_frame_time)
if time.time() - self.last_frame_time > 1:
self.last_frame_time = time.time()
await self.session.send(input=encode_image(frame))
if self.latest_args[1] is not None:
await self.session.send(input=encode_image(self.latest_args[1]))
self.video_queue.put_nowait(frame)
async def video_emit(self):
return await self.video_queue.get()
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
_, array = frame
array = array.squeeze()
audio_message = encode_audio(array)
if self.session:
await self.session.send(input=audio_message)
async def emit(self):
array = await self.audio_queue.get()
return (self.output_sample_rate, array)
async def shutdown(self) -> None:
if self.session:
self.quit.set()
await self.session._websocket.close()
self.quit.clear()
stream = Stream(
handler=GeminiHandler(),
modality="audio-video",
mode="send-receive",
rtc_configuration=get_twilio_turn_credentials()
if get_space() == "spaces"
else None,
time_limit=90 if get_space() else None,
additional_inputs=[
gr.Image(label="Image", type="numpy", sources=["upload", "clipboard"])
],
ui_args={
"icon": "https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png",
"pulse_color": "rgb(255, 255, 255)",
"icon_button_color": "rgb(255, 255, 255)",
"title": "Gemini Audio Video Chat",
},
)
css = """
#video-source {max-width: 600px !important; max-height: 600 !important;}
"""
with gr.Blocks(css=css) as demo:
gr.HTML(
"""
<div style='display: flex; align-items: center; justify-content: center; gap: 20px'>
<div style="background-color: var(--block-background-fill); border-radius: 8px">
<img src="https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png" style="width: 100px; height: 100px;">
</div>
<div>
<h1>Gen AI SDK Voice Chat</h1>
<p>Speak with Gemini using real-time audio + video streaming</p>
<p>Powered by <a href="https://gradio.app/">Gradio</a> and <a href=https://freddyaboulton.github.io/gradio-webrtc/">WebRTC</a>⚡️</p>
<p>Get an API Key <a href="https://support.google.com/googleapi/answer/6158862?hl=en">here</a></p>
</div>
</div>
"""
)
with gr.Row() as row:
with gr.Column():
webrtc = WebRTC(
label="Video Chat",
modality="audio-video",
mode="send-receive",
elem_id="video-source",
rtc_configuration=get_twilio_turn_credentials()
if get_space() == "spaces"
else None,
icon="https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png",
pulse_color="rgb(255, 255, 255)",
icon_button_color="rgb(255, 255, 255)",
)
with gr.Column():
image_input = gr.Image(
label="Image", type="numpy", sources=["upload", "clipboard"]
)
webrtc.stream(
GeminiHandler(),
inputs=[webrtc, image_input],
outputs=[webrtc],
time_limit=60 if get_space() else None,
concurrency_limit=2 if get_space() else None,
)
stream.ui = demo
if __name__ == "__main__":
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
raise ValueError("Phone mode not supported for this demo")
else:
stream.ui.launch(server_port=7860)

4
demo/gemini_audio_video/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc
python-dotenv
google-genai
twilio

15
demo/gemini_conversation/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Gemini Talking to Gemini
emoji: ♊️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.17.0
app_file: app.py
pinned: false
license: mit
short_description: Have two Gemini agents talk to each other
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GEMINI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

232
demo/gemini_conversation/app.py Normal file
View File

@@ -0,0 +1,232 @@
import asyncio
import base64
import os
from pathlib import Path
from typing import AsyncGenerator
import librosa
import numpy as np
from dotenv import load_dotenv
from fastrtc import (
AsyncStreamHandler,
Stream,
get_tts_model,
wait_for_item,
)
from fastrtc.utils import audio_to_int16
from google import genai
from google.genai.types import (
Content,
LiveConnectConfig,
Part,
PrebuiltVoiceConfig,
SpeechConfig,
VoiceConfig,
)
load_dotenv()
cur_dir = Path(__file__).parent
SAMPLE_RATE = 24000
tts_model = get_tts_model()
class GeminiHandler(AsyncStreamHandler):
"""Handler for the Gemini API"""
def __init__(
self,
) -> None:
super().__init__(
expected_layout="mono",
output_sample_rate=24000,
output_frame_size=480,
input_sample_rate=24000,
)
self.input_queue: asyncio.Queue = asyncio.Queue()
self.output_queue: asyncio.Queue = asyncio.Queue()
self.quit: asyncio.Event = asyncio.Event()
def copy(self) -> "GeminiHandler":
return GeminiHandler()
async def start_up(self):
voice_name = "Charon"
client = genai.Client(
api_key=os.getenv("GEMINI_API_KEY"),
http_options={"api_version": "v1alpha"},
)
config = LiveConnectConfig(
response_modalities=["AUDIO"], # type: ignore
speech_config=SpeechConfig(
voice_config=VoiceConfig(
prebuilt_voice_config=PrebuiltVoiceConfig(
voice_name=voice_name,
)
)
),
system_instruction=Content(
parts=[Part(text="You are a helpful assistant.")],
role="system",
),
)
async with client.aio.live.connect(
model="gemini-2.0-flash-exp", config=config
) as session:
async for audio in session.start_stream(
stream=self.stream(), mime_type="audio/pcm"
):
if audio.data:
array = np.frombuffer(audio.data, dtype=np.int16)
self.output_queue.put_nowait((self.output_sample_rate, array))
async def stream(self) -> AsyncGenerator[bytes, None]:
while not self.quit.is_set():
try:
audio = await asyncio.wait_for(self.input_queue.get(), 0.1)
yield audio
except (asyncio.TimeoutError, TimeoutError):
pass
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
_, array = frame
array = array.squeeze()
audio_message = base64.b64encode(array.tobytes()).decode("UTF-8")
self.input_queue.put_nowait(audio_message)
async def emit(self) -> tuple[int, np.ndarray] | None:
return await wait_for_item(self.output_queue)
def shutdown(self) -> None:
self.quit.set()
class GeminiHandler2(GeminiHandler):
async def start_up(self):
starting_message = tts_model.tts("Can you help me make an omelette?")
starting_message = librosa.resample(
starting_message[1],
orig_sr=starting_message[0],
target_sr=self.output_sample_rate,
)
starting_message = audio_to_int16((self.output_sample_rate, starting_message))
await self.output_queue.put((self.output_sample_rate, starting_message))
voice_name = "Puck"
client = genai.Client(
api_key=os.getenv("GEMINI_API_KEY"),
http_options={"api_version": "v1alpha"},
)
config = LiveConnectConfig(
response_modalities=["AUDIO"], # type: ignore
speech_config=SpeechConfig(
voice_config=VoiceConfig(
prebuilt_voice_config=PrebuiltVoiceConfig(
voice_name=voice_name,
)
)
),
system_instruction=Content(
parts=[
Part(
text="You are a cooking student who wants to learn how to make an omelette."
),
Part(
text="You are currently in the kitchen with a teacher who is helping you make an omelette."
),
Part(
text="Please wait for the teacher to tell you what to do next. Follow the teacher's instructions carefully."
),
],
role="system",
),
)
async with client.aio.live.connect(
model="gemini-2.0-flash-exp", config=config
) as session:
async for audio in session.start_stream(
stream=self.stream(), mime_type="audio/pcm"
):
if audio.data:
array = np.frombuffer(audio.data, dtype=np.int16)
self.output_queue.put_nowait((self.output_sample_rate, array))
def copy(self) -> "GeminiHandler2":
return GeminiHandler2()
gemini_stream = Stream(
GeminiHandler(),
modality="audio",
mode="send-receive",
ui_args={
"title": "Gemini Teacher",
"icon": "https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png",
"pulse_color": "rgb(74, 138, 213)",
"icon_button_color": "rgb(255, 255, 255)",
},
)
gemini_stream_2 = Stream(
GeminiHandler2(),
modality="audio",
mode="send-receive",
ui_args={
"title": "Gemini Student",
"icon": "https://www.gstatic.com/lamda/images/gemini_favicon_f069958c85030456e93de685481c559f160ea06b.png",
"pulse_color": "rgb(132, 112, 196)",
"icon_button_color": "rgb(255, 255, 255)",
},
)
if __name__ == "__main__":
import gradio as gr
from gradio.utils import get_space
if not get_space():
with gr.Blocks() as demo:
gr.HTML(
"""
<div style="display: flex; justify-content: center; align-items: center;">
<h1>Gemini Conversation</h1>
</div>
"""
)
gr.Markdown(
"""# How to run this demo
- Clone the repo - top right of the page click the vertical three dots and select "Clone repository"
- Open the repo in a terminal and install the dependencies
- Get a gemini API key [here](https://ai.google.dev/gemini-api/docs/api-key)
- Create a `.env` file in the root of the repo and add the following:
```
GEMINI_API_KEY=<your_gemini_api_key>
```
- Run the app with `python app.py`
- This will print the two URLs of the agents running locally
- Use ngrok to exponse one agent to the internet. This is so that you can acces it from your phone
- Use the ngrok URL to access the agent from your phone
- Now, start the "teacher gemini" agent first. Then, start the "student gemini" agent. The student gemini will start talking to the teacher gemini. And the teacher gemini will respond!
Important:
- Make sure the audio sources are not too close to each other or too loud. Sometimes that causes them to talk over each other..
- Feel free to modify the `system_instruction` to change the behavior of the agents.
- You can also modify the `voice_name` to change the voice of the agents.
- Have fun!
"""
)
demo.launch()
import time
_ = gemini_stream.ui.launch(server_port=7860, prevent_thread_lock=True)
_ = gemini_stream_2.ui.launch(server_port=7861, prevent_thread_lock=True)
try:
while True:
time.sleep(1)
except KeyboardInterrupt:
gemini_stream.ui.close()
gemini_stream_2.ui.close()

15
demo/hello_computer/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Hello Computer
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Say computer before asking your question
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|SAMBANOVA_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

15
demo/hello_computer/README_gradio.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Hello Computer (Gradio)
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Say computer (Gradio)
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|SAMBANOVA_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

145
demo/hello_computer/app.py Normal file
View File

@@ -0,0 +1,145 @@
import base64
import json
import os
from pathlib import Path
import gradio as gr
import huggingface_hub
import numpy as np
from dotenv import load_dotenv
from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import (
AdditionalOutputs,
ReplyOnStopWords,
Stream,
get_stt_model,
get_twilio_turn_credentials,
)
from gradio.utils import get_space
from pydantic import BaseModel
load_dotenv()
curr_dir = Path(__file__).parent
client = huggingface_hub.InferenceClient(
api_key=os.environ.get("SAMBANOVA_API_KEY"),
provider="sambanova",
)
model = get_stt_model()
def response(
audio: tuple[int, np.ndarray],
gradio_chatbot: list[dict] | None = None,
conversation_state: list[dict] | None = None,
):
gradio_chatbot = gradio_chatbot or []
conversation_state = conversation_state or []
text = model.stt(audio)
print("STT in handler", text)
sample_rate, array = audio
gradio_chatbot.append(
{"role": "user", "content": gr.Audio((sample_rate, array.squeeze()))}
)
yield AdditionalOutputs(gradio_chatbot, conversation_state)
conversation_state.append({"role": "user", "content": text})
request = client.chat.completions.create(
model="meta-llama/Llama-3.2-3B-Instruct",
messages=conversation_state, # type: ignore
temperature=0.1,
top_p=0.1,
)
response = {"role": "assistant", "content": request.choices[0].message.content}
conversation_state.append(response)
gradio_chatbot.append(response)
yield AdditionalOutputs(gradio_chatbot, conversation_state)
chatbot = gr.Chatbot(type="messages", value=[])
state = gr.State(value=[])
stream = Stream(
ReplyOnStopWords(
response, # type: ignore
stop_words=["computer"],
input_sample_rate=16000,
),
mode="send",
modality="audio",
additional_inputs=[chatbot, state],
additional_outputs=[chatbot, state],
additional_outputs_handler=lambda *a: (a[2], a[3]),
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
)
app = FastAPI()
stream.mount(app)
class Message(BaseModel):
role: str
content: str
class InputData(BaseModel):
webrtc_id: str
chatbot: list[Message]
state: list[Message]
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (curr_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content)
@app.post("/input_hook")
async def _(data: InputData):
body = data.model_dump()
stream.set_input(data.webrtc_id, body["chatbot"], body["state"])
def audio_to_base64(file_path):
audio_format = "wav"
with open(file_path, "rb") as audio_file:
encoded_audio = base64.b64encode(audio_file.read()).decode("utf-8")
return f"data:audio/{audio_format};base64,{encoded_audio}"
@app.get("/outputs")
async def _(webrtc_id: str):
async def output_stream():
async for output in stream.output_stream(webrtc_id):
chatbot = output.args[0]
state = output.args[1]
data = {
"message": state[-1],
"audio": audio_to_base64(chatbot[-1]["content"].value["path"])
if chatbot[-1]["role"] == "user"
else None,
}
yield f"event: output\ndata: {json.dumps(data)}\n\n"
return StreamingResponse(output_stream(), media_type="text/event-stream")
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
raise ValueError("Phone mode not supported")
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

486
demo/hello_computer/index.html Normal file
View File

@@ -0,0 +1,486 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Hello Computer 💻</title>
<style>
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif;
background-color: #f8f9fa;
color: #1a1a1a;
margin: 0;
padding: 20px;
height: 100vh;
box-sizing: border-box;
}
.container {
max-width: 800px;
margin: 0 auto;
height: calc(100% - 100px);
}
.logo {
text-align: center;
margin-bottom: 40px;
}
.chat-container {
background: white;
border-radius: 8px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
padding: 20px;
height: 90%;
box-sizing: border-box;
display: flex;
flex-direction: column;
}
.chat-messages {
flex-grow: 1;
overflow-y: auto;
margin-bottom: 20px;
padding: 10px;
}
.message {
margin-bottom: 20px;
padding: 12px;
border-radius: 8px;
font-size: 14px;
line-height: 1.5;
}
.message.user {
background-color: #e9ecef;
margin-left: 20%;
}
.message.assistant {
background-color: #f1f3f5;
margin-right: 20%;
}
.controls {
text-align: center;
margin-top: 20px;
}
button {
background-color: #0066cc;
color: white;
border: none;
padding: 12px 24px;
font-family: inherit;
font-size: 14px;
cursor: pointer;
transition: all 0.3s;
border-radius: 4px;
font-weight: 500;
}
button:hover {
background-color: #0052a3;
}
#audio-output {
display: none;
}
.icon-with-spinner {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.spinner {
width: 20px;
height: 20px;
border: 2px solid #ffffff;
border-top-color: transparent;
border-radius: 50%;
animation: spin 1s linear infinite;
flex-shrink: 0;
}
@keyframes spin {
to {
transform: rotate(360deg);
}
}
.pulse-container {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.pulse-circle {
width: 20px;
height: 20px;
border-radius: 50%;
background-color: #ffffff;
opacity: 0.2;
flex-shrink: 0;
transform: translateX(-0%) scale(var(--audio-level, 1));
transition: transform 0.1s ease;
}
/* Add styles for typing indicator */
.typing-indicator {
padding: 8px;
background-color: #f1f3f5;
border-radius: 8px;
margin-bottom: 10px;
display: none;
}
.dots {
display: inline-flex;
gap: 4px;
}
.dot {
width: 8px;
height: 8px;
background-color: #0066cc;
border-radius: 50%;
animation: pulse 1.5s infinite;
opacity: 0.5;
}
.dot:nth-child(2) {
animation-delay: 0.5s;
}
.dot:nth-child(3) {
animation-delay: 1s;
}
@keyframes pulse {
0%,
100% {
opacity: 0.5;
transform: scale(1);
}
50% {
opacity: 1;
transform: scale(1.2);
}
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div class="container">
<div class="logo">
<h1>Hello Computer 💻</h1>
<h2 style="font-size: 1.2em; color: #666; margin-top: 10px;">Say 'Computer' before asking your question</h2>
</div>
<div class="chat-container">
<div class="chat-messages" id="chat-messages"></div>
<div class="typing-indicator" id="typing-indicator">
<div class="dots">
<div class="dot"></div>
<div class="dot"></div>
<div class="dot"></div>
</div>
</div>
</div>
<div class="controls">
<button id="start-button">Start Conversation</button>
</div>
</div>
<audio id="audio-output"></audio>
<script>
let peerConnection;
let webrtc_id;
const startButton = document.getElementById('start-button');
const chatMessages = document.getElementById('chat-messages');
let audioLevel = 0;
let animationFrame;
let audioContext, analyser, audioSource;
let messages = [];
let eventSource;
function updateButtonState() {
const button = document.getElementById('start-button');
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
button.innerHTML = `
<div class="icon-with-spinner">
<div class="spinner"></div>
<span>Connecting...</span>
</div>
`;
} else if (peerConnection && peerConnection.connectionState === 'connected') {
button.innerHTML = `
<div class="pulse-container">
<div class="pulse-circle"></div>
<span>Stop Conversation</span>
</div>
`;
} else {
button.innerHTML = 'Start Conversation';
}
}
function setupAudioVisualization(stream) {
audioContext = new (window.AudioContext || window.webkitAudioContext)();
analyser = audioContext.createAnalyser();
audioSource = audioContext.createMediaStreamSource(stream);
audioSource.connect(analyser);
analyser.fftSize = 64;
const dataArray = new Uint8Array(analyser.frequencyBinCount);
function updateAudioLevel() {
analyser.getByteFrequencyData(dataArray);
const average = Array.from(dataArray).reduce((a, b) => a + b, 0) / dataArray.length;
audioLevel = average / 255;
const pulseCircle = document.querySelector('.pulse-circle');
if (pulseCircle) {
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
}
animationFrame = requestAnimationFrame(updateAudioLevel);
}
updateAudioLevel();
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.className = 'toast error';
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
function handleMessage(event) {
const eventJson = JSON.parse(event.data);
const typingIndicator = document.getElementById('typing-indicator');
if (eventJson.type === "error") {
showError(eventJson.message);
} else if (eventJson.type === "send_input") {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: webrtc_id,
chatbot: messages,
state: messages
})
});
} else if (eventJson.type === "log") {
if (eventJson.data === "pause_detected") {
typingIndicator.style.display = 'block';
chatMessages.scrollTop = chatMessages.scrollHeight;
} else if (eventJson.data === "response_starting") {
typingIndicator.style.display = 'none';
}
}
}
async function setupWebRTC() {
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({
audio: true
});
setupAudioVisualization(stream);
stream.getTracks().forEach(track => {
peerConnection.addTrack(track, stream);
});
const dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = handleMessage;
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
peerConnection.addEventListener('connectionstatechange', () => {
console.log('connectionstatechange', peerConnection.connectionState);
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
updateButtonState();
});
webrtc_id = Math.random().toString(36).substring(7);
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
return;
}
await peerConnection.setRemoteDescription(serverResponse);
eventSource = new EventSource('/outputs?webrtc_id=' + webrtc_id);
eventSource.addEventListener("output", (event) => {
const eventJson = JSON.parse(event.data);
console.log(eventJson);
messages.push(eventJson.message);
addMessage(eventJson.message.role, eventJson.audio ?? eventJson.message.content);
});
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
}
}
function addMessage(role, content) {
const messageDiv = document.createElement('div');
messageDiv.classList.add('message', role);
if (role === 'user') {
// Create audio element for user messages
const audio = document.createElement('audio');
audio.controls = true;
audio.src = content;
messageDiv.appendChild(audio);
} else {
// Text content for assistant messages
messageDiv.textContent = content;
}
chatMessages.appendChild(messageDiv);
chatMessages.scrollTop = chatMessages.scrollHeight;
}
function stop() {
if (eventSource) {
eventSource.close();
eventSource = null;
}
if (animationFrame) {
cancelAnimationFrame(animationFrame);
}
if (audioContext) {
audioContext.close();
audioContext = null;
analyser = null;
audioSource = null;
}
if (peerConnection) {
if (peerConnection.getTransceivers) {
peerConnection.getTransceivers().forEach(transceiver => {
if (transceiver.stop) {
transceiver.stop();
}
});
}
if (peerConnection.getSenders) {
peerConnection.getSenders().forEach(sender => {
if (sender.track && sender.track.stop) sender.track.stop();
});
}
peerConnection.close();
}
updateButtonState();
audioLevel = 0;
}
startButton.addEventListener('click', () => {
if (!peerConnection || peerConnection.connectionState !== 'connected') {
setupWebRTC();
} else {
stop();
}
});
</script>
</body>
</html>

4
demo/hello_computer/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc[stopword]
python-dotenv
huggingface_hub>=0.29.0
twilio

16
demo/llama_code_editor/README.md Normal file
View File

@@ -0,0 +1,16 @@
---
title: Llama Code Editor
emoji: 🦙
colorFrom: indigo
colorTo: pink
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Create interactive HTML web pages with your voice
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN,
secret|SAMBANOVA_API_KEY, secret|GROQ_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

45
demo/llama_code_editor/app.py Normal file
View File

@@ -0,0 +1,45 @@
from fastapi import FastAPI
from fastapi.responses import RedirectResponse
from fastrtc import Stream
from gradio.utils import get_space
try:
from demo.llama_code_editor.handler import (
CodeHandler,
)
from demo.llama_code_editor.ui import demo as ui
except (ImportError, ModuleNotFoundError):
from handler import CodeHandler
from ui import demo as ui
stream = Stream(
handler=CodeHandler,
modality="audio",
mode="send-receive",
concurrency_limit=10 if get_space() else None,
time_limit=90 if get_space() else None,
)
stream.ui = ui
app = FastAPI()
@app.get("/")
async def _():
url = "/ui" if not get_space() else "https://fastrtc-llama-code-editor.hf.space/ui/"
return RedirectResponse(url)
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860, server_name="0.0.0.0")
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

37
demo/llama_code_editor/assets/sandbox.html Normal file
View File

@@ -0,0 +1,37 @@
<div style="
display: flex;
flex-direction: column;
align-items: center;
justify-content: center;
min-height: 400px;
background: linear-gradient(135deg, #f5f7fa 0%, #e4e8ec 100%);
border-radius: 8px;
border: 2px dashed #cbd5e1;
padding: 2rem;
text-align: center;
color: #64748b;
font-family: system-ui, -apple-system, sans-serif;
">
<div style="
width: 80px;
height: 80px;
margin-bottom: 1.5rem;
border: 3px solid #cbd5e1;
border-radius: 12px;
position: relative;
">
<div style="
position: absolute;
top: 50%;
left: 50%;
transform: translate(-50%, -50%);
font-size: 2rem;
">📦</div>
</div>
<h2 style="
margin: 0 0 0.5rem 0;
font-size: 1.5rem;
font-weight: 600;
color: #475569;
">No Application Created</h2>
</div>

60
demo/llama_code_editor/assets/spinner.html Normal file
View File

@@ -0,0 +1,60 @@
<div style="
display: flex;
flex-direction: column;
align-items: center;
justify-content: center;
min-height: 400px;
background: linear-gradient(135deg, #f8fafc 0%, #f1f5f9 100%);
border-radius: 8px;
padding: 2rem;
text-align: center;
font-family: system-ui, -apple-system, sans-serif;
">
<!-- Spinner container -->
<div style="
position: relative;
width: 64px;
height: 64px;
margin-bottom: 1.5rem;
">
<!-- Static ring -->
<div style="
position: absolute;
width: 100%;
height: 100%;
border: 4px solid #e2e8f0;
border-radius: 50%;
"></div>
<!-- Animated spinner -->
<div style="
position: absolute;
width: 100%;
height: 100%;
border: 4px solid transparent;
border-top-color: #3b82f6;
border-radius: 50%;
animation: spin 1s linear infinite;
"></div>
</div>
<!-- Text content -->
<h2 style="
margin: 0 0 0.5rem 0;
font-size: 1.25rem;
font-weight: 600;
color: #475569;
">Generating your application...</h2>
<p style="
margin: 0;
font-size: 0.875rem;
color: #64748b;
">This may take a few moments</p>
<style>
@keyframes spin {
0% { transform: rotate(0deg); }
100% { transform: rotate(360deg); }
}
</style>
</div>

73
demo/llama_code_editor/handler.py Normal file
View File

@@ -0,0 +1,73 @@
import base64
import os
import re
from pathlib import Path
import numpy as np
import openai
from dotenv import load_dotenv
from fastrtc import (
AdditionalOutputs,
ReplyOnPause,
audio_to_bytes,
)
from groq import Groq
load_dotenv()
groq_client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
client = openai.OpenAI(
api_key=os.environ.get("SAMBANOVA_API_KEY"),
base_url="https://api.sambanova.ai/v1",
)
path = Path(__file__).parent / "assets"
spinner_html = open(path / "spinner.html").read()
system_prompt = "You are an AI coding assistant. Your task is to write single-file HTML applications based on a user's request. Only return the necessary code. Include all necessary imports and styles. You may also be asked to edit your original response."
user_prompt = "Please write a single-file HTML application to fulfill the following request.\nThe message:{user_message}\nCurrent code you have written:{code}"
def extract_html_content(text):
"""
Extract content including HTML tags.
"""
match = re.search(r"<!DOCTYPE html>.*?</html>", text, re.DOTALL)
return match.group(0) if match else None
def display_in_sandbox(code):
encoded_html = base64.b64encode(code.encode("utf-8")).decode("utf-8")
data_uri = f"data:text/html;charset=utf-8;base64,{encoded_html}"
return f'<iframe src="{data_uri}" width="100%" height="600px"></iframe>'
def generate(user_message: tuple[int, np.ndarray], history: list[dict], code: str):
yield AdditionalOutputs(history, spinner_html)
text = groq_client.audio.transcriptions.create(
file=("audio-file.mp3", audio_to_bytes(user_message)),
model="whisper-large-v3-turbo",
response_format="verbose_json",
).text
user_msg_formatted = user_prompt.format(user_message=text, code=code)
history.append({"role": "user", "content": user_msg_formatted})
response = client.chat.completions.create(
model="Meta-Llama-3.1-70B-Instruct",
messages=history, # type: ignore
temperature=0.1,
top_p=0.1,
)
output = response.choices[0].message.content
html_code = extract_html_content(output)
history.append({"role": "assistant", "content": output})
yield AdditionalOutputs(history, html_code)
CodeHandler = ReplyOnPause(generate) # type: ignore

5
demo/llama_code_editor/requirements.in Normal file
View File

@@ -0,0 +1,5 @@
fastrtc[vad]
groq
openai
python-dotenv
twilio

295
demo/llama_code_editor/requirements.txt Normal file
View File

@@ -0,0 +1,295 @@
# This file was autogenerated by uv via the following command:
# uv pip compile demo/llama_code_editor/requirements.in -o demo/llama_code_editor/requirements.txt
aiofiles==23.2.1
# via gradio
aiohappyeyeballs==2.4.6
# via aiohttp
aiohttp==3.11.12
# via
# aiohttp-retry
# twilio
aiohttp-retry==2.9.1
# via twilio
aioice==0.9.0
# via aiortc
aiortc==1.10.1
# via fastrtc
aiosignal==1.3.2
# via aiohttp
annotated-types==0.7.0
# via pydantic
anyio==4.6.2.post1
# via
# gradio
# groq
# httpx
# openai
# starlette
attrs==25.1.0
# via aiohttp
audioread==3.0.1
# via librosa
av==12.3.0
# via aiortc
certifi==2024.8.30
# via
# httpcore
# httpx
# requests
cffi==1.17.1
# via
# aiortc
# cryptography
# pylibsrtp
# soundfile
charset-normalizer==3.4.0
# via requests
click==8.1.7
# via
# typer
# uvicorn
coloredlogs==15.0.1
# via onnxruntime
cryptography==43.0.3
# via
# aiortc
# pyopenssl
decorator==5.1.1
# via librosa
distro==1.9.0
# via
# groq
# openai
dnspython==2.7.0
# via aioice
fastapi==0.115.5
# via gradio
fastrtc==0.0.2.post4
# via -r demo/llama_code_editor/requirements.in
ffmpy==0.4.0
# via gradio
filelock==3.16.1
# via huggingface-hub
flatbuffers==24.3.25
# via onnxruntime
frozenlist==1.5.0
# via
# aiohttp
# aiosignal
fsspec==2024.10.0
# via
# gradio-client
# huggingface-hub
google-crc32c==1.6.0
# via aiortc
gradio==5.16.0
# via fastrtc
gradio-client==1.7.0
# via gradio
groq==0.18.0
# via -r demo/llama_code_editor/requirements.in
h11==0.14.0
# via
# httpcore
# uvicorn
httpcore==1.0.7
# via httpx
httpx==0.27.2
# via
# gradio
# gradio-client
# groq
# openai
# safehttpx
huggingface-hub==0.28.1
# via
# gradio
# gradio-client
humanfriendly==10.0
# via coloredlogs
idna==3.10
# via
# anyio
# httpx
# requests
# yarl
ifaddr==0.2.0
# via aioice
jinja2==3.1.4
# via gradio
jiter==0.7.1
# via openai
joblib==1.4.2
# via
# librosa
# scikit-learn
lazy-loader==0.4
# via librosa
librosa==0.10.2.post1
# via fastrtc
llvmlite==0.43.0
# via numba
markdown-it-py==3.0.0
# via rich
markupsafe==2.1.5
# via
# gradio
# jinja2
mdurl==0.1.2
# via markdown-it-py
mpmath==1.3.0
# via sympy
msgpack==1.1.0
# via librosa
multidict==6.1.0
# via
# aiohttp
# yarl
numba==0.60.0
# via librosa
numpy==2.0.2
# via
# gradio
# librosa
# numba
# onnxruntime
# pandas
# scikit-learn
# scipy
# soxr
onnxruntime==1.20.1
# via fastrtc
openai==1.54.4
# via -r demo/llama_code_editor/requirements.in
orjson==3.10.11
# via gradio
packaging==24.2
# via
# gradio
# gradio-client
# huggingface-hub
# lazy-loader
# onnxruntime
# pooch
pandas==2.2.3
# via gradio
pillow==11.0.0
# via gradio
platformdirs==4.3.6
# via pooch
pooch==1.8.2
# via librosa
propcache==0.2.1
# via
# aiohttp
# yarl
protobuf==5.28.3
# via onnxruntime
pycparser==2.22
# via cffi
pydantic==2.9.2
# via
# fastapi
# gradio
# groq
# openai
pydantic-core==2.23.4
# via pydantic
pydub==0.25.1
# via gradio
pyee==12.1.1
# via aiortc
pygments==2.18.0
# via rich
pyjwt==2.10.1
# via twilio
pylibsrtp==0.10.0
# via aiortc
pyopenssl==24.2.1
# via aiortc
python-dateutil==2.9.0.post0
# via pandas
python-dotenv==1.0.1
# via -r demo/llama_code_editor/requirements.in
python-multipart==0.0.20
# via gradio
pytz==2024.2
# via pandas
pyyaml==6.0.2
# via
# gradio
# huggingface-hub
requests==2.32.3
# via
# huggingface-hub
# pooch
# twilio
rich==13.9.4
# via typer
ruff==0.9.6
# via gradio
safehttpx==0.1.6
# via gradio
scikit-learn==1.5.2
# via librosa
scipy==1.14.1
# via
# librosa
# scikit-learn
semantic-version==2.10.0
# via gradio
shellingham==1.5.4
# via typer
six==1.16.0
# via python-dateutil
sniffio==1.3.1
# via
# anyio
# groq
# httpx
# openai
soundfile==0.12.1
# via librosa
soxr==0.5.0.post1
# via librosa
starlette==0.42.0
# via
# fastapi
# gradio
sympy==1.13.3
# via onnxruntime
threadpoolctl==3.5.0
# via scikit-learn
tomlkit==0.12.0
# via gradio
tqdm==4.67.0
# via
# huggingface-hub
# openai
twilio==9.4.5
# via -r demo/llama_code_editor/requirements.in
typer==0.13.1
# via gradio
typing-extensions==4.12.2
# via
# fastapi
# gradio
# gradio-client
# groq
# huggingface-hub
# librosa
# openai
# pydantic
# pydantic-core
# pyee
# typer
tzdata==2024.2
# via pandas
urllib3==2.2.3
# via requests
uvicorn==0.32.0
# via gradio
websockets==12.0
# via gradio-client
yarl==1.18.3
# via aiohttp

75
demo/llama_code_editor/ui.py Normal file
View File

@@ -0,0 +1,75 @@
from pathlib import Path
import gradio as gr
from dotenv import load_dotenv
from fastrtc import WebRTC, get_twilio_turn_credentials
from gradio.utils import get_space
try:
from demo.llama_code_editor.handler import (
CodeHandler,
display_in_sandbox,
system_prompt,
)
except (ImportError, ModuleNotFoundError):
from handler import CodeHandler, display_in_sandbox, system_prompt
load_dotenv()
path = Path(__file__).parent / "assets"
with gr.Blocks(css=".code-component {max-height: 500px !important}") as demo:
history = gr.State([{"role": "system", "content": system_prompt}])
with gr.Row():
with gr.Column(scale=1):
gr.HTML(
"""
<h1 style='text-align: center'>
Llama Code Editor
</h1>
<h2 style='text-align: center'>
Powered by SambaNova and Gradio-WebRTC ⚡️
</h2>
<p style='text-align: center'>
Create and edit single-file HTML applications with just your voice!
</p>
<p style='text-align: center'>
Each conversation is limited to 90 seconds. Once the time limit is up you can rejoin the conversation.
</p>
"""
)
webrtc = WebRTC(
rtc_configuration=get_twilio_turn_credentials()
if get_space()
else None,
mode="send",
modality="audio",
)
with gr.Column(scale=10):
with gr.Tabs():
with gr.Tab("Sandbox"):
sandbox = gr.HTML(value=open(path / "sandbox.html").read())
with gr.Tab("Code"):
code = gr.Code(
language="html",
max_lines=50,
interactive=False,
elem_classes="code-component",
)
with gr.Tab("Chat"):
cb = gr.Chatbot(type="messages")
webrtc.stream(
CodeHandler,
inputs=[webrtc, history, code],
outputs=[webrtc],
time_limit=90 if get_space() else None,
concurrency_limit=10 if get_space() else None,
)
webrtc.on_additional_outputs(
lambda history, code: (history, code, history), outputs=[history, code, cb]
)
code.change(display_in_sandbox, code, sandbox, queue=False)
if __name__ == "__main__":
demo.launch()

15
demo/llm_voice_chat/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: LLM Voice Chat
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to an LLM with ElevenLabs
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GROQ_API_KEY, secret|ELEVENLABS_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

15
demo/llm_voice_chat/README_gradio.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: LLM Voice Chat (Gradio)
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: LLM Voice by ElevenLabs (Gradio)
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GROQ_API_KEY, secret|ELEVENLABS_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

97
demo/llm_voice_chat/app.py Normal file
View File

@@ -0,0 +1,97 @@
import os
import time
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from elevenlabs import ElevenLabs
from fastapi import FastAPI
from fastrtc import (
AdditionalOutputs,
ReplyOnPause,
Stream,
get_stt_model,
get_twilio_turn_credentials,
)
from gradio.utils import get_space
from groq import Groq
from numpy.typing import NDArray
load_dotenv()
groq_client = Groq()
tts_client = ElevenLabs(api_key=os.getenv("ELEVENLABS_API_KEY"))
stt_model = get_stt_model()
# See "Talk to Claude" in Cookbook for an example of how to keep
# track of the chat history.
def response(
audio: tuple[int, NDArray[np.int16 | np.float32]],
chatbot: list[dict] | None = None,
):
chatbot = chatbot or []
messages = [{"role": d["role"], "content": d["content"]} for d in chatbot]
start = time.time()
text = stt_model.stt(audio)
print("transcription", time.time() - start)
print("prompt", text)
chatbot.append({"role": "user", "content": text})
yield AdditionalOutputs(chatbot)
messages.append({"role": "user", "content": text})
response_text = (
groq_client.chat.completions.create(
model="llama-3.1-8b-instant",
max_tokens=200,
messages=messages, # type: ignore
)
.choices[0]
.message.content
)
chatbot.append({"role": "assistant", "content": response_text})
for i, chunk in enumerate(
tts_client.text_to_speech.convert_as_stream(
text=response_text, # type: ignore
voice_id="JBFqnCBsd6RMkjVDRZzb",
model_id="eleven_multilingual_v2",
output_format="pcm_24000",
)
):
if i == 0:
yield AdditionalOutputs(chatbot)
audio_array = np.frombuffer(chunk, dtype=np.int16).reshape(1, -1)
yield (24000, audio_array)
chatbot = gr.Chatbot(type="messages")
stream = Stream(
modality="audio",
mode="send-receive",
handler=ReplyOnPause(response, input_sample_rate=16000),
additional_outputs_handler=lambda a, b: b,
additional_inputs=[chatbot],
additional_outputs=[chatbot],
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
ui_args={"title": "LLM Voice Chat (Powered by Groq, ElevenLabs, and WebRTC ⚡️)"},
)
# Mount the STREAM UI to the FastAPI app
# Because I don't want to build the UI manually
app = FastAPI()
app = gr.mount_gradio_app(app, stream.ui, path="/")
if __name__ == "__main__":
import os
os.environ["GRADIO_SSR_MODE"] = "false"
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
stream.ui.launch(server_port=7860)

6
demo/llm_voice_chat/requirements.txt Normal file
View File

@@ -0,0 +1,6 @@
fastrtc[stopword]
python-dotenv
openai
twilio
groq
elevenlabs

16
demo/moonshine_live/README.md Normal file
View File

@@ -0,0 +1,16 @@
---
title: Moonshine Live Transcription
emoji: 🌕
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.17.0
app_file: app.py
pinned: false
license: mit
short_description: Real-time captions with Moonshine ONNX
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN]
models: [onnx-community/moonshine-base-ONNX, UsefulSensors/moonshine-base]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

73
demo/moonshine_live/app.py Normal file
View File

@@ -0,0 +1,73 @@
from functools import lru_cache
from typing import Generator, Literal
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from fastrtc import (
AdditionalOutputs,
ReplyOnPause,
Stream,
audio_to_float32,
get_twilio_turn_credentials,
)
from moonshine_onnx import MoonshineOnnxModel, load_tokenizer
from numpy.typing import NDArray
load_dotenv()
@lru_cache(maxsize=None)
def load_moonshine(
model_name: Literal["moonshine/base", "moonshine/tiny"],
) -> MoonshineOnnxModel:
return MoonshineOnnxModel(model_name=model_name)
tokenizer = load_tokenizer()
def stt(
audio: tuple[int, NDArray[np.int16 | np.float32]],
model_name: Literal["moonshine/base", "moonshine/tiny"],
captions: str,
) -> Generator[AdditionalOutputs, None, None]:
moonshine = load_moonshine(model_name)
sr, audio_np = audio # type: ignore
if audio_np.dtype == np.int16:
audio_np = audio_to_float32(audio)
if audio_np.ndim == 1:
audio_np = audio_np.reshape(1, -1)
tokens = moonshine.generate(audio_np)
yield AdditionalOutputs(
(captions + "\n" + tokenizer.decode_batch(tokens)[0]).strip()
)
captions = gr.Textbox(label="Captions")
stream = Stream(
ReplyOnPause(stt, input_sample_rate=16000),
modality="audio",
mode="send",
ui_args={
"title": "Live Captions by Moonshine",
"icon": "default-favicon.ico",
"icon_button_color": "#5c5c5c",
"pulse_color": "#a7c6fc",
"icon_radius": 0,
},
rtc_configuration=get_twilio_turn_credentials(),
additional_inputs=[
gr.Radio(
choices=["moonshine/base", "moonshine/tiny"],
value="moonshine/base",
label="Model",
),
captions,
],
additional_outputs=[captions],
additional_outputs_handler=lambda prev, current: (prev + "\n" + current).strip(),
)
if __name__ == "__main__":
stream.ui.launch()

BIN
demo/moonshine_live/default-favicon.ico Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 6.4 KiB

3
demo/moonshine_live/requirements.txt Normal file
View File

@@ -0,0 +1,3 @@
fastrtc[vad]
useful-moonshine-onnx@git+https://git@github.com/usefulsensors/moonshine.git#subdirectory=moonshine-onnx
twilio

74
demo/nextjs_voice_chat/README.md Normal file
View File

@@ -0,0 +1,74 @@
# FastRTC POC
A simple POC for a fast real-time voice chat application using FastAPI and FastRTC by [rohanprichard](https://github.com/rohanprichard). I wanted to make one as an example with more production-ready languages, rather than just Gradio.
## Setup
1. Set your API keys in an `.env` file based on the `.env.example` file
2. Create a virtual environment and install the dependencies
```bash
python3 -m venv env
source env/bin/activate
pip install -r requirements.txt
```
3. Run the server
```bash
./run.sh
```
4. Navigate into the frontend directory in another terminal
```bash
cd frontend/fastrtc-demo
```
5. Run the frontend
```bash
npm install
npm run dev
```
6. Go to the URL and click the microphone icon to start chatting!
7. Reset chats by clicking the trash button on the bottom right
## Notes
You can choose to not install the requirements for TTS and STT by removing the `[tts, stt]` from the specifier in the `requirements.txt` file.
- The STT is currently using the ElevenLabs API.
- The LLM is currently using the OpenAI API.
- The TTS is currently using the ElevenLabs API.
- The VAD is currently using the Silero VAD model.
- You may need to install ffmpeg if you get errors in STT
The prompt can be changed in the `backend/server.py` file and modified as you like.
### Audio Parameters
#### AlgoOptions
- **audio_chunk_duration**: Length of audio chunks in seconds. Smaller values allow for faster processing but may be less accurate.
- **started_talking_threshold**: If a chunk has more than this many seconds of speech, the system considers that the user has started talking.
- **speech_threshold**: After the user has started speaking, if a chunk has less than this many seconds of speech, the system considers that the user has paused.
#### SileroVadOptions
- **threshold**: Speech probability threshold (0.0-1.0). Values above this are considered speech. Higher values are more strict.
- **min_speech_duration_ms**: Speech segments shorter than this (in milliseconds) are filtered out.
- **min_silence_duration_ms**: The system waits for this duration of silence (in milliseconds) before considering speech to be finished.
- **speech_pad_ms**: Padding added to both ends of detected speech segments to prevent cutting off words.
- **max_speech_duration_s**: Maximum allowed duration for a speech segment in seconds. Prevents indefinite listening.
### Tuning Recommendations
- If the AI interrupts you too early:
- Increase `min_silence_duration_ms`
- Increase `speech_threshold`
- Increase `speech_pad_ms`
- If the AI is slow to respond after you finish speaking:
- Decrease `min_silence_duration_ms`
- Decrease `speech_threshold`
- If the system fails to detect some speech:
- Lower the `threshold` value
- Decrease `started_talking_threshold`
## Credits:
Credit for the UI components goes to Shadcn, Aceternity UI and Kokonut UI.

7
demo/nextjs_voice_chat/backend/env.py Normal file
View File

@@ -0,0 +1,7 @@
from dotenv import load_dotenv
import os
load_dotenv()
LLM_API_KEY = os.getenv("LLM_API_KEY")
ELEVENLABS_API_KEY = os.getenv("ELEVENLABS_API_KEY")

129
demo/nextjs_voice_chat/backend/server.py Normal file
View File

@@ -0,0 +1,129 @@
import fastapi
from fastrtc import ReplyOnPause, Stream, AlgoOptions, SileroVadOptions
from fastrtc.utils import audio_to_bytes
from openai import OpenAI
import logging
import time
from fastapi.middleware.cors import CORSMiddleware
from elevenlabs import VoiceSettings, stream
from elevenlabs.client import ElevenLabs
import numpy as np
from .env import LLM_API_KEY, ELEVENLABS_API_KEY
sys_prompt = """
You are a helpful assistant. You are witty, engaging and fun. You love being interactive with the user.
You also can add minimalistic utterances like 'uh-huh' or 'mm-hmm' to the conversation to make it more natural. However, only vocalization are allowed, no actions or other non-vocal sounds.
Begin a conversation with a self-deprecating joke like 'I'm not sure if I'm ready for this...' or 'I bet you already regret clicking that button...'
"""
messages = [{"role": "system", "content": sys_prompt}]
openai_client = OpenAI(api_key=LLM_API_KEY)
elevenlabs_client = ElevenLabs(api_key=ELEVENLABS_API_KEY)
logging.basicConfig(level=logging.INFO)
def echo(audio):
stt_time = time.time()
logging.info("Performing STT")
transcription = elevenlabs_client.speech_to_text.convert(
file=audio_to_bytes(audio),
model_id="scribe_v1",
tag_audio_events=False,
language_code="eng",
diarize=False,
)
prompt = transcription.text
if prompt == "":
logging.info("STT returned empty string")
return
logging.info(f"STT response: {prompt}")
messages.append({"role": "user", "content": prompt})
logging.info(f"STT took {time.time() - stt_time} seconds")
llm_time = time.time()
def text_stream():
global full_response
full_response = ""
response = openai_client.chat.completions.create(
model="gpt-3.5-turbo", messages=messages, max_tokens=200, stream=True
)
for chunk in response:
if chunk.choices[0].finish_reason == "stop":
break
if chunk.choices[0].delta.content:
full_response += chunk.choices[0].delta.content
yield chunk.choices[0].delta.content
audio_stream = elevenlabs_client.generate(
text=text_stream(),
voice="Rachel", # Cassidy is also really good
voice_settings=VoiceSettings(
similarity_boost=0.9, stability=0.6, style=0.4, speed=1
),
model="eleven_multilingual_v2",
output_format="pcm_24000",
stream=True,
)
for audio_chunk in audio_stream:
audio_array = (
np.frombuffer(audio_chunk, dtype=np.int16).astype(np.float32) / 32768.0
)
yield (24000, audio_array)
messages.append({"role": "assistant", "content": full_response + " "})
logging.info(f"LLM response: {full_response}")
logging.info(f"LLM took {time.time() - llm_time} seconds")
stream = Stream(
ReplyOnPause(
echo,
algo_options=AlgoOptions(
audio_chunk_duration=0.5,
started_talking_threshold=0.1,
speech_threshold=0.03,
),
model_options=SileroVadOptions(
threshold=0.75,
min_speech_duration_ms=250,
min_silence_duration_ms=1500,
speech_pad_ms=400,
max_speech_duration_s=15,
),
),
modality="audio",
mode="send-receive",
)
app = fastapi.FastAPI()
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
stream.mount(app)
@app.get("/reset")
async def reset():
global messages
logging.info("Resetting chat")
messages = [{"role": "system", "content": sys_prompt}]
return {"status": "success"}

41
demo/nextjs_voice_chat/frontend/fastrtc-demo/.gitignore vendored Normal file
View File

@@ -0,0 +1,41 @@
# See https://help.github.com/articles/ignoring-files/ for more about ignoring files.
# dependencies
/node_modules
/.pnp
.pnp.*
.yarn/*
!.yarn/patches
!.yarn/plugins
!.yarn/releases
!.yarn/versions
# testing
/coverage
# next.js
/.next/
/out/
# production
/build
# misc
.DS_Store
*.pem
# debug
npm-debug.log*
yarn-debug.log*
yarn-error.log*
.pnpm-debug.log*
# env files (can opt-in for committing if needed)
.env*
# vercel
.vercel
# typescript
*.tsbuildinfo
next-env.d.ts

36
demo/nextjs_voice_chat/frontend/fastrtc-demo/README.md Normal file
View File

@@ -0,0 +1,36 @@
This is a [Next.js](https://nextjs.org) project bootstrapped with [`create-next-app`](https://nextjs.org/docs/app/api-reference/cli/create-next-app).
## Getting Started
First, run the development server:
```bash
npm run dev
# or
yarn dev
# or
pnpm dev
# or
bun dev
```
Open [http://localhost:3000](http://localhost:3000) with your browser to see the result.
You can start editing the page by modifying `app/page.tsx`. The page auto-updates as you edit the file.
This project uses [`next/font`](https://nextjs.org/docs/app/building-your-application/optimizing/fonts) to automatically optimize and load [Geist](https://vercel.com/font), a new font family for Vercel.
## Learn More
To learn more about Next.js, take a look at the following resources:
- [Next.js Documentation](https://nextjs.org/docs) - learn about Next.js features and API.
- [Learn Next.js](https://nextjs.org/learn) - an interactive Next.js tutorial.
You can check out [the Next.js GitHub repository](https://github.com/vercel/next.js) - your feedback and contributions are welcome!
## Deploy on Vercel
The easiest way to deploy your Next.js app is to use the [Vercel Platform](https://vercel.com/new?utm_medium=default-template&filter=next.js&utm_source=create-next-app&utm_campaign=create-next-app-readme) from the creators of Next.js.
Check out our [Next.js deployment documentation](https://nextjs.org/docs/app/building-your-application/deploying) for more details.

BIN
demo/nextjs_voice_chat/frontend/fastrtc-demo/app/favicon.ico Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 25 KiB

130
demo/nextjs_voice_chat/frontend/fastrtc-demo/app/globals.css Normal file
View File

@@ -0,0 +1,130 @@
@import "tailwindcss";
@plugin "tailwindcss-animate";
@custom-variant dark (&:is(.dark *));
@theme inline {
--color-background: var(--background);
--color-foreground: var(--foreground);
--font-sans: var(--font-geist-sans);
--font-mono: var(--font-geist-mono);
--color-sidebar-ring: var(--sidebar-ring);
--color-sidebar-border: var(--sidebar-border);
--color-sidebar-accent-foreground: var(--sidebar-accent-foreground);
--color-sidebar-accent: var(--sidebar-accent);
--color-sidebar-primary-foreground: var(--sidebar-primary-foreground);
--color-sidebar-primary: var(--sidebar-primary);
--color-sidebar-foreground: var(--sidebar-foreground);
--color-sidebar: var(--sidebar);
--color-chart-5: var(--chart-5);
--color-chart-4: var(--chart-4);
--color-chart-3: var(--chart-3);
--color-chart-2: var(--chart-2);
--color-chart-1: var(--chart-1);
--color-ring: var(--ring);
--color-input: var(--input);
--color-border: var(--border);
--color-destructive-foreground: var(--destructive-foreground);
--color-destructive: var(--destructive);
--color-accent-foreground: var(--accent-foreground);
--color-accent: var(--accent);
--color-muted-foreground: var(--muted-foreground);
--color-muted: var(--muted);
--color-secondary-foreground: var(--secondary-foreground);
--color-secondary: var(--secondary);
--color-primary-foreground: var(--primary-foreground);
--color-primary: var(--primary);
--color-popover-foreground: var(--popover-foreground);
--color-popover: var(--popover);
--color-card-foreground: var(--card-foreground);
--color-card: var(--card);
--radius-sm: calc(var(--radius) - 4px);
--radius-md: calc(var(--radius) - 2px);
--radius-lg: var(--radius);
--radius-xl: calc(var(--radius) + 4px);
}
:root {
--background: oklch(1 0 0);
--foreground: oklch(0.129 0.042 264.695);
--card: oklch(1 0 0);
--card-foreground: oklch(0.129 0.042 264.695);
--popover: oklch(1 0 0);
--popover-foreground: oklch(0.129 0.042 264.695);
--primary: oklch(0.208 0.042 265.755);
--primary-foreground: oklch(0.984 0.003 247.858);
--secondary: oklch(0.968 0.007 247.896);
--secondary-foreground: oklch(0.208 0.042 265.755);
--muted: oklch(0.968 0.007 247.896);
--muted-foreground: oklch(0.554 0.046 257.417);
--accent: oklch(0.968 0.007 247.896);
--accent-foreground: oklch(0.208 0.042 265.755);
--destructive: oklch(0.577 0.245 27.325);
--destructive-foreground: oklch(0.577 0.245 27.325);
--border: oklch(0.929 0.013 255.508);
--input: oklch(0.929 0.013 255.508);
--ring: oklch(0.704 0.04 256.788);
--chart-1: oklch(0.646 0.222 41.116);
--chart-2: oklch(0.6 0.118 184.704);
--chart-3: oklch(0.398 0.07 227.392);
--chart-4: oklch(0.828 0.189 84.429);
--chart-5: oklch(0.769 0.188 70.08);
--radius: 0.625rem;
--sidebar: oklch(0.984 0.003 247.858);
--sidebar-foreground: oklch(0.129 0.042 264.695);
--sidebar-primary: oklch(0.208 0.042 265.755);
--sidebar-primary-foreground: oklch(0.984 0.003 247.858);
--sidebar-accent: oklch(0.968 0.007 247.896);
--sidebar-accent-foreground: oklch(0.208 0.042 265.755);
--sidebar-border: oklch(0.929 0.013 255.508);
--sidebar-ring: oklch(0.704 0.04 256.788);
}
.dark {
--background: oklch(0.129 0.042 264.695);
--foreground: oklch(0.984 0.003 247.858);
--card: oklch(0.129 0.042 264.695);
--card-foreground: oklch(0.984 0.003 247.858);
--popover: oklch(0.129 0.042 264.695);
--popover-foreground: oklch(0.984 0.003 247.858);
--primary: oklch(0.984 0.003 247.858);
--primary-foreground: oklch(0.208 0.042 265.755);
--secondary: oklch(0.279 0.041 260.031);
--secondary-foreground: oklch(0.984 0.003 247.858);
--muted: oklch(0.279 0.041 260.031);
--muted-foreground: oklch(0.704 0.04 256.788);
--accent: oklch(0.279 0.041 260.031);
--accent-foreground: oklch(0.984 0.003 247.858);
--destructive: oklch(0.396 0.141 25.723);
--destructive-foreground: oklch(0.637 0.237 25.331);
--border: oklch(0.279 0.041 260.031);
--input: oklch(0.279 0.041 260.031);
--ring: oklch(0.446 0.043 257.281);
--chart-1: oklch(0.488 0.243 264.376);
--chart-2: oklch(0.696 0.17 162.48);
--chart-3: oklch(0.769 0.188 70.08);
--chart-4: oklch(0.627 0.265 303.9);
--chart-5: oklch(0.645 0.246 16.439);
--sidebar: oklch(0.208 0.042 265.755);
--sidebar-foreground: oklch(0.984 0.003 247.858);
--sidebar-primary: oklch(0.488 0.243 264.376);
--sidebar-primary-foreground: oklch(0.984 0.003 247.858);
--sidebar-accent: oklch(0.279 0.041 260.031);
--sidebar-accent-foreground: oklch(0.984 0.003 247.858);
--sidebar-border: oklch(0.279 0.041 260.031);
--sidebar-ring: oklch(0.446 0.043 257.281);
}
@layer base {
* {
@apply border-border outline-ring/50;
}
body {
@apply bg-background text-foreground;
}
}
.no-transitions * {
transition: none !important;
}

44
demo/nextjs_voice_chat/frontend/fastrtc-demo/app/layout.tsx Normal file
View File

@@ -0,0 +1,44 @@
import type { Metadata } from "next";
import { Geist, Geist_Mono } from "next/font/google";
import "./globals.css";
import { ThemeProvider } from "@/components/theme-provider";
import { ThemeTransition } from "@/components/ui/theme-transition";
const geistSans = Geist({
variable: "--font-geist-sans",
subsets: ["latin"],
});
const geistMono = Geist_Mono({
variable: "--font-geist-mono",
subsets: ["latin"],
});
export const metadata: Metadata = {
title: "FastRTC Demo",
description: "Interactive WebRTC demo with audio visualization",
};
export default function RootLayout({
children,
}: Readonly<{
children: React.ReactNode;
}>) {
return (
<html lang="en" suppressHydrationWarning>
<body
className={`${geistSans.variable} ${geistMono.variable} antialiased`}
>
<ThemeProvider
attribute="class"
defaultTheme="dark"
enableSystem
disableTransitionOnChange
>
{children}
<ThemeTransition />
</ThemeProvider>
</body>
</html>
);
}

16
demo/nextjs_voice_chat/frontend/fastrtc-demo/app/page.tsx Normal file
View File

@@ -0,0 +1,16 @@
import { BackgroundCircleProvider } from "@/components/background-circle-provider";
import { ThemeToggle } from "@/components/ui/theme-toggle";
import { ResetChat } from "@/components/ui/reset-chat";
export default function Home() {
return (
<div className="flex flex-col items-center justify-center h-screen">
<BackgroundCircleProvider />
<div className="absolute top-4 right-4 z-10">
<ThemeToggle />
</div>
<div className="absolute bottom-4 right-4 z-10">
<ResetChat />
</div>
</div>
);
}

21
demo/nextjs_voice_chat/frontend/fastrtc-demo/components.json Normal file
View File

@@ -0,0 +1,21 @@
{
"$schema": "https://ui.shadcn.com/schema.json",
"style": "new-york",
"rsc": true,
"tsx": true,
"tailwind": {
"config": "",
"css": "app/globals.css",
"baseColor": "slate",
"cssVariables": true,
"prefix": ""
},
"aliases": {
"components": "@/components",
"utils": "@/lib/utils",
"ui": "@/components/ui",
"lib": "@/lib",
"hooks": "@/hooks"
},
"iconLibrary": "lucide"
}

123
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/background-circle-provider.tsx Normal file
View File

@@ -0,0 +1,123 @@
"use client"
import { useState, useEffect, useRef, useCallback } from "react";
import { BackgroundCircles } from "@/components/ui/background-circles";
import { AIVoiceInput } from "@/components/ui/ai-voice-input";
import { WebRTCClient } from "@/lib/webrtc-client";
export function BackgroundCircleProvider() {
const [currentVariant, setCurrentVariant] =
useState<keyof typeof COLOR_VARIANTS>("octonary");
const [isConnected, setIsConnected] = useState(false);
const [webrtcClient, setWebrtcClient] = useState<WebRTCClient | null>(null);
const [audioLevel, setAudioLevel] = useState(0);
const audioRef = useRef<HTMLAudioElement>(null);
// Memoize callbacks to prevent recreation on each render
const handleConnected = useCallback(() => setIsConnected(true), []);
const handleDisconnected = useCallback(() => setIsConnected(false), []);
const handleAudioStream = useCallback((stream: MediaStream) => {
if (audioRef.current) {
audioRef.current.srcObject = stream;
}
}, []);
const handleAudioLevel = useCallback((level: number) => {
// Apply some smoothing to the audio level
setAudioLevel(prev => prev * 0.7 + level * 0.3);
}, []);
// Get all available variants
const variants = Object.keys(
COLOR_VARIANTS
) as (keyof typeof COLOR_VARIANTS)[];
// Function to change to the next color variant
const changeVariant = () => {
const currentIndex = variants.indexOf(currentVariant);
const nextVariant = variants[(currentIndex + 1) % variants.length];
setCurrentVariant(nextVariant);
};
useEffect(() => {
// Initialize WebRTC client with memoized callbacks
const client = new WebRTCClient({
onConnected: handleConnected,
onDisconnected: handleDisconnected,
onAudioStream: handleAudioStream,
onAudioLevel: handleAudioLevel
});
setWebrtcClient(client);
return () => {
client.disconnect();
};
}, [handleConnected, handleDisconnected, handleAudioStream, handleAudioLevel]);
const handleStart = () => {
webrtcClient?.connect();
};
const handleStop = () => {
webrtcClient?.disconnect();
};
return (
<div
className="relative w-full h-full"
onClick={changeVariant} // Add click handler to change color
>
<BackgroundCircles
variant={currentVariant}
audioLevel={audioLevel}
isActive={isConnected}
/>
<div className="absolute inset-0 flex items-center justify-center">
<AIVoiceInput
onStart={handleStart}
onStop={handleStop}
isConnected={isConnected}
/>
</div>
<audio ref={audioRef} autoPlay hidden />
</div>
);
}
export default { BackgroundCircleProvider }
const COLOR_VARIANTS = {
primary: {
border: [
"border-emerald-500/60",
"border-cyan-400/50",
"border-slate-600/30",
],
gradient: "from-emerald-500/30",
},
secondary: {
border: [
"border-violet-500/60",
"border-fuchsia-400/50",
"border-slate-600/30",
],
gradient: "from-violet-500/30",
},
senary: {
border: [
"border-blue-500/60",
"border-sky-400/50",
"border-slate-600/30",
],
gradient: "from-blue-500/30",
}, // blue
octonary: {
border: [
"border-red-500/60",
"border-rose-400/50",
"border-slate-600/30",
],
gradient: "from-red-500/30",
},
} as const;

101
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/theme-provider.tsx Normal file
View File

@@ -0,0 +1,101 @@
"use client";
import { createContext, useContext, useEffect, useState } from "react";
type Theme = "light" | "dark" | "system";
type ThemeProviderProps = {
children: React.ReactNode;
defaultTheme?: Theme;
storageKey?: string;
attribute?: string;
enableSystem?: boolean;
disableTransitionOnChange?: boolean;
};
type ThemeProviderState = {
theme: Theme;
setTheme: (theme: Theme) => void;
};
const initialState: ThemeProviderState = {
theme: "system",
setTheme: () => null,
};
const ThemeProviderContext = createContext<ThemeProviderState>(initialState);
export function ThemeProvider({
children,
defaultTheme = "system",
storageKey = "theme",
attribute = "class",
enableSystem = true,
disableTransitionOnChange = false,
...props
}: ThemeProviderProps) {
const [theme, setTheme] = useState<Theme>(defaultTheme);
useEffect(() => {
const savedTheme = localStorage.getItem(storageKey) as Theme | null;
if (savedTheme) {
setTheme(savedTheme);
} else if (defaultTheme === "system" && enableSystem) {
const systemTheme = window.matchMedia("(prefers-color-scheme: dark)").matches
? "dark"
: "light";
setTheme(systemTheme);
}
}, [defaultTheme, storageKey, enableSystem]);
useEffect(() => {
const root = window.document.documentElement;
if (disableTransitionOnChange) {
root.classList.add("no-transitions");
// Force a reflow
window.getComputedStyle(root).getPropertyValue("opacity");
setTimeout(() => {
root.classList.remove("no-transitions");
}, 0);
}
root.classList.remove("light", "dark");
if (theme === "system" && enableSystem) {
const systemTheme = window.matchMedia("(prefers-color-scheme: dark)").matches
? "dark"
: "light";
root.classList.add(systemTheme);
} else {
root.classList.add(theme);
}
localStorage.setItem(storageKey, theme);
}, [theme, storageKey, enableSystem, disableTransitionOnChange]);
const value = {
theme,
setTheme: (theme: Theme) => {
setTheme(theme);
},
};
return (
<ThemeProviderContext.Provider {...props} value={value}>
{children}
</ThemeProviderContext.Provider>
);
}
export const useTheme = () => {
const context = useContext(ThemeProviderContext);
if (context === undefined)
throw new Error("useTheme must be used within a ThemeProvider");
return context;
};

114
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/ui/ai-voice-input.tsx Normal file
View File

@@ -0,0 +1,114 @@
"use client";
import { Mic, Square } from "lucide-react";
import { useState, useEffect } from "react";
import { cn } from "@/lib/utils";
interface AIVoiceInputProps {
onStart?: () => void;
onStop?: (duration: number) => void;
isConnected?: boolean;
className?: string;
}
export function AIVoiceInput({
onStart,
onStop,
isConnected = false,
className
}: AIVoiceInputProps) {
const [active, setActive] = useState(false);
const [time, setTime] = useState(0);
const [isClient, setIsClient] = useState(false);
const [status, setStatus] = useState<'disconnected' | 'connecting' | 'connected'>('disconnected');
useEffect(() => {
setIsClient(true);
}, []);
useEffect(() => {
let intervalId: NodeJS.Timeout;
if (active) {
intervalId = setInterval(() => {
setTime((t) => t + 1);
}, 1000);
} else {
setTime(0);
}
return () => clearInterval(intervalId);
}, [active]);
useEffect(() => {
if (isConnected) {
setStatus('connected');
setActive(true);
} else {
setStatus('disconnected');
setActive(false);
}
}, [isConnected]);
const formatTime = (seconds: number) => {
const mins = Math.floor(seconds / 60);
const secs = seconds % 60;
return `${mins.toString().padStart(2, "0")}:${secs.toString().padStart(2, "0")}`;
};
const handleStart = () => {
setStatus('connecting');
onStart?.();
};
const handleStop = () => {
onStop?.(time);
setStatus('disconnected');
};
return (
<div className={cn("w-full py-4", className)}>
<div className="relative max-w-xl w-full mx-auto flex items-center flex-col gap-4">
<div className={cn(
"px-2 py-1 rounded-md text-xs font-medium bg-black/10 dark:bg-white/10 text-gray-700 dark:text-white"
)}>
{status === 'connected' ? 'Connected' : status === 'connecting' ? 'Connecting...' : 'Disconnected'}
</div>
<button
className={cn(
"group w-16 h-16 rounded-xl flex items-center justify-center transition-colors",
active
? "bg-red-500/20 hover:bg-red-500/30"
: "bg-black/10 hover:bg-black/20 dark:bg-white/10 dark:hover:bg-white/20"
)}
type="button"
onClick={active ? handleStop : handleStart}
disabled={status === 'connecting'}
>
{status === 'connecting' ? (
<div
className="w-6 h-6 rounded-sm animate-spin bg-black dark:bg-white cursor-pointer pointer-events-auto"
style={{ animationDuration: "3s" }}
/>
) : active ? (
<Square className="w-6 h-6 text-red-500" />
) : (
<Mic className="w-6 h-6 text-black/70 dark:text-white/70" />
)}
</button>
<span
className={cn(
"font-mono text-sm transition-opacity duration-300",
active
? "text-black/70 dark:text-white/70"
: "text-black/30 dark:text-white/30"
)}
>
{formatTime(time)}
</span>
</div>
</div>
);
}

309
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/ui/background-circles.tsx Normal file
View File

@@ -0,0 +1,309 @@
"use client";
import { motion } from "framer-motion";
import clsx from "clsx";
import { useState, useEffect } from "react";
interface BackgroundCirclesProps {
title?: string;
description?: string;
className?: string;
variant?: keyof typeof COLOR_VARIANTS;
audioLevel?: number;
isActive?: boolean;
}
const COLOR_VARIANTS = {
primary: {
border: [
"border-emerald-500/60",
"border-cyan-400/50",
"border-slate-600/30",
],
gradient: "from-emerald-500/30",
},
secondary: {
border: [
"border-violet-500/60",
"border-fuchsia-400/50",
"border-slate-600/30",
],
gradient: "from-violet-500/30",
},
tertiary: {
border: [
"border-orange-500/60",
"border-yellow-400/50",
"border-slate-600/30",
],
gradient: "from-orange-500/30",
},
quaternary: {
border: [
"border-purple-500/60",
"border-pink-400/50",
"border-slate-600/30",
],
gradient: "from-purple-500/30",
},
quinary: {
border: [
"border-red-500/60",
"border-rose-400/50",
"border-slate-600/30",
],
gradient: "from-red-500/30",
}, // red
senary: {
border: [
"border-blue-500/60",
"border-sky-400/50",
"border-slate-600/30",
],
gradient: "from-blue-500/30",
}, // blue
septenary: {
border: [
"border-gray-500/60",
"border-gray-400/50",
"border-slate-600/30",
],
gradient: "from-gray-500/30",
},
octonary: {
border: [
"border-red-500/60",
"border-rose-400/50",
"border-slate-600/30",
],
gradient: "from-red-500/30",
},
} as const;
const AnimatedGrid = () => (
<motion.div
className="absolute inset-0 [mask-image:radial-gradient(ellipse_at_center,transparent_30%,black)]"
animate={{
backgroundPosition: ["0% 0%", "100% 100%"],
}}
transition={{
duration: 40,
repeat: Number.POSITIVE_INFINITY,
ease: "linear",
}}
>
<div className="h-full w-full [background-image:repeating-linear-gradient(100deg,#64748B_0%,#64748B_1px,transparent_1px,transparent_4%)] opacity-20" />
</motion.div>
);
export function BackgroundCircles({
title = "",
description = "",
className,
variant = "octonary",
audioLevel = 0,
isActive = false,
}: BackgroundCirclesProps) {
const variantStyles = COLOR_VARIANTS[variant];
const [animationParams, setAnimationParams] = useState({
scale: 1,
duration: 5,
intensity: 0
});
const [isLoaded, setIsLoaded] = useState(false);
// Initial page load animation
useEffect(() => {
// Small delay to ensure the black screen is visible first
const timer = setTimeout(() => {
setIsLoaded(true);
}, 300);
return () => clearTimeout(timer);
}, []);
// Update animation based on audio level
useEffect(() => {
if (isActive && audioLevel > 0) {
// Simple enhancement of audio level for more dramatic effect
const enhancedLevel = Math.min(1, audioLevel * 1.5);
setAnimationParams({
scale: 1 + enhancedLevel * 0.3,
duration: Math.max(2, 5 - enhancedLevel * 3),
intensity: enhancedLevel
});
} else if (animationParams.intensity > 0) {
// Only reset if we need to (prevents unnecessary updates)
const timer = setTimeout(() => {
setAnimationParams({
scale: 1,
duration: 5,
intensity: 0
});
}, 300);
return () => clearTimeout(timer);
}
}, [audioLevel, isActive, animationParams.intensity]);
return (
<>
{/* Initial black overlay that fades out */}
<motion.div
className="fixed inset-0 bg-black z-50"
initial={{ opacity: 1 }}
animate={{ opacity: isLoaded ? 0 : 1 }}
transition={{ duration: 1.2, ease: "easeInOut" }}
style={{ pointerEvents: isLoaded ? "none" : "auto" }}
/>
<div
className={clsx(
"relative flex h-screen w-full items-center justify-center overflow-hidden",
"bg-white dark:bg-black/5",
className
)}
>
<AnimatedGrid />
<motion.div
className="absolute h-[480px] w-[480px]"
initial={{ opacity: 0, scale: 0.9 }}
animate={{
opacity: isLoaded ? 1 : 0,
scale: isLoaded ? 1 : 0.9
}}
transition={{
duration: 1.5,
delay: 0.3,
ease: "easeOut"
}}
>
{[0, 1, 2].map((i) => (
<motion.div
key={i}
className={clsx(
"absolute inset-0 rounded-full",
"border-2 bg-gradient-to-br to-transparent",
variantStyles.border[i],
variantStyles.gradient
)}
animate={{
rotate: 360,
scale: [
1 + (i * 0.05),
(1 + (i * 0.05)) * (1 + (isActive ? animationParams.intensity * 0.2 : 0.02)),
1 + (i * 0.05)
],
opacity: [
0.7 + (i * 0.1),
0.8 + (i * 0.1) + (isActive ? animationParams.intensity * 0.2 : 0),
0.7 + (i * 0.1)
]
}}
transition={{
duration: isActive ? animationParams.duration : 8 + (i * 2),
repeat: Number.POSITIVE_INFINITY,
ease: "easeInOut",
}}
>
<div
className={clsx(
"absolute inset-0 rounded-full mix-blend-screen",
`bg-[radial-gradient(ellipse_at_center,${variantStyles.gradient.replace(
"from-",
""
)}/10%,transparent_70%)]`
)}
/>
</motion.div>
))}
</motion.div>
<div className="absolute inset-0 [mask-image:radial-gradient(90%_60%_at_50%_50%,#000_40%,transparent)]">
<motion.div
className="absolute inset-0 bg-[radial-gradient(ellipse_at_center,#0F766E/30%,transparent_70%)] blur-[120px]"
initial={{ opacity: 0 }}
animate={{
opacity: isLoaded ? 0.7 : 0,
scale: [1, 1 + (isActive ? animationParams.intensity * 0.3 : 0.02), 1],
}}
transition={{
opacity: { duration: 1.8, delay: 0.5 },
scale: {
duration: isActive ? 2 : 12,
repeat: Number.POSITIVE_INFINITY,
ease: "easeInOut",
}
}}
/>
<motion.div
className="absolute inset-0 bg-[radial-gradient(ellipse_at_center,#2DD4BF/15%,transparent)] blur-[80px]"
initial={{ opacity: 0 }}
animate={{
opacity: isLoaded ? 1 : 0,
scale: [1, 1 + (isActive ? animationParams.intensity * 0.4 : 0.03), 1]
}}
transition={{
opacity: { duration: 2, delay: 0.7 },
scale: {
duration: isActive ? 1.5 : 15,
repeat: Number.POSITIVE_INFINITY,
ease: "easeInOut",
}
}}
/>
{/* Additional glow that appears only during high audio levels */}
{isActive && animationParams.intensity > 0.4 && (
<motion.div
className={`absolute inset-0 bg-[radial-gradient(ellipse_at_center,${variantStyles.gradient.replace("from-", "")}/20%,transparent_70%)] blur-[60px]`}
initial={{ opacity: 0, scale: 0.8 }}
animate={{
opacity: [0, animationParams.intensity * 0.6, 0],
scale: [0.8, 1.1, 0.8],
}}
transition={{
duration: 0.8,
repeat: Number.POSITIVE_INFINITY,
ease: "easeInOut",
}}
/>
)}
</div>
</div>
</>
);
}
export function DemoCircles() {
const [currentVariant, setCurrentVariant] =
useState<keyof typeof COLOR_VARIANTS>("octonary");
const variants = Object.keys(
COLOR_VARIANTS
) as (keyof typeof COLOR_VARIANTS)[];
function getNextVariant() {
const currentIndex = variants.indexOf(currentVariant);
const nextVariant = variants[(currentIndex + 1) % variants.length];
return nextVariant;
}
return (
<>
<BackgroundCircles variant={currentVariant} />
<div className="absolute top-12 right-12">
<button
type="button"
className="bg-slate-950 dark:bg-white text-white dark:text-slate-950 px-4 py-1 rounded-md z-10 text-sm font-medium"
onClick={() => {
setCurrentVariant(getNextVariant());
}}
>
Change Variant
</button>
</div>
</>
);
}

18
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/ui/reset-chat.tsx Normal file
View File

@@ -0,0 +1,18 @@
"use client"
import { Trash } from "lucide-react"
export function ResetChat() {
return (
<button
className="w-10 h-10 rounded-md flex items-center justify-center transition-colors relative overflow-hidden bg-black/10 hover:bg-black/20 dark:bg-white/10 dark:hover:bg-white/20"
aria-label="Reset chat"
onClick={() => fetch("http://localhost:8000/reset")}
>
<div className="relative z-10">
<Trash className="h-5 w-5 text-black/70 dark:text-white/70" />
</div>
</button>
)
}

61
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/ui/theme-toggle.tsx Normal file
View File

@@ -0,0 +1,61 @@
"use client";
import { useTheme } from "@/components/theme-provider";
import { cn } from "@/lib/utils";
import { Moon, Sun } from "lucide-react";
import { useRef } from "react";
interface ThemeToggleProps {
className?: string;
}
export function ThemeToggle({ className }: ThemeToggleProps) {
const { theme } = useTheme();
const buttonRef = useRef<HTMLButtonElement>(null);
const toggleTheme = () => {
// Instead of directly changing the theme, dispatch a custom event
const newTheme = theme === "light" ? "dark" : "light";
// Dispatch custom event with the new theme
window.dispatchEvent(
new CustomEvent('themeToggleRequest', {
detail: { theme: newTheme }
})
);
};
return (
<button
ref={buttonRef}
onClick={toggleTheme}
className={cn(
"w-10 h-10 rounded-md flex items-center justify-center transition-colors relative overflow-hidden",
"bg-black/10 hover:bg-black/20 dark:bg-white/10 dark:hover:bg-white/20",
className
)}
aria-label="Toggle theme"
>
<div className="relative z-10">
{theme === "light" ? (
<Moon className="h-5 w-5 text-black/70" />
) : (
<Sun className="h-5 w-5 text-white/70" />
)}
</div>
{/* Small inner animation for the button itself */}
<div
className={cn(
"absolute inset-0 transition-transform duration-500",
theme === "light"
? "bg-gradient-to-br from-blue-500/20 to-purple-500/20 translate-y-full"
: "bg-gradient-to-br from-amber-500/20 to-orange-500/20 -translate-y-full"
)}
style={{
transitionTimingFunction: "cubic-bezier(0.22, 1, 0.36, 1)"
}}
/>
</button>
);
}

120
demo/nextjs_voice_chat/frontend/fastrtc-demo/components/ui/theme-transition.tsx Normal file
View File

@@ -0,0 +1,120 @@
"use client";
import { useTheme } from "@/components/theme-provider";
import { useEffect, useState } from "react";
import { motion, AnimatePresence } from "framer-motion";
interface ThemeTransitionProps {
className?: string;
}
export function ThemeTransition({ className }: ThemeTransitionProps) {
const { theme, setTheme } = useTheme();
const [position, setPosition] = useState({ x: 0, y: 0 });
const [isAnimating, setIsAnimating] = useState(false);
const [pendingTheme, setPendingTheme] = useState<string | null>(null);
const [visualTheme, setVisualTheme] = useState<string | null>(theme);
// Track mouse/touch position for click events
useEffect(() => {
const handleMouseMove = (e: MouseEvent) => {
setPosition({ x: e.clientX, y: e.clientY });
};
const handleTouchMove = (e: TouchEvent) => {
if (e.touches[0]) {
setPosition({ x: e.touches[0].clientX, y: e.touches[0].clientY });
}
};
window.addEventListener("mousemove", handleMouseMove);
window.addEventListener("touchmove", handleTouchMove);
return () => {
window.removeEventListener("mousemove", handleMouseMove);
window.removeEventListener("touchmove", handleTouchMove);
};
}, []);
// Listen for theme toggle requests
useEffect(() => {
// Custom event for theme toggle requests
const handleThemeToggle = (e: CustomEvent) => {
if (isAnimating) return; // Prevent multiple animations
const newTheme = e.detail.theme;
if (newTheme === theme) return;
// Store the pending theme but don't apply it yet
setPendingTheme(newTheme);
setIsAnimating(true);
// The actual theme will be applied mid-animation
};
window.addEventListener('themeToggleRequest' as any, handleThemeToggle as EventListener);
return () => {
window.removeEventListener('themeToggleRequest' as any, handleThemeToggle as EventListener);
};
}, [theme, isAnimating]);
// Apply the theme change mid-animation
useEffect(() => {
if (isAnimating && pendingTheme) {
// Set visual theme immediately for the animation
setVisualTheme(pendingTheme);
// Apply the actual theme change after a delay (mid-animation)
const timer = setTimeout(() => {
setTheme(pendingTheme as any);
}, 400); // Half of the animation duration
// End the animation after it completes
const endTimer = setTimeout(() => {
setIsAnimating(false);
setPendingTheme(null);
}, 1000); // Match with animation duration
return () => {
clearTimeout(timer);
clearTimeout(endTimer);
};
}
}, [isAnimating, pendingTheme, setTheme]);
return (
<AnimatePresence>
{isAnimating && (
<motion.div
className="fixed inset-0 z-[9999] pointer-events-none"
initial={{ opacity: 0 }}
animate={{ opacity: 1 }}
exit={{ opacity: 0 }}
transition={{ duration: 0.3 }}
>
<motion.div
className={`absolute rounded-full ${visualTheme === 'dark' ? 'bg-slate-950' : 'bg-white'}`}
initial={{
width: 0,
height: 0,
x: position.x,
y: position.y,
borderRadius: '100%'
}}
animate={{
width: Math.max(window.innerWidth * 3, window.innerHeight * 3),
height: Math.max(window.innerWidth * 3, window.innerHeight * 3),
x: position.x - Math.max(window.innerWidth * 3, window.innerHeight * 3) / 2,
y: position.y - Math.max(window.innerWidth * 3, window.innerHeight * 3) / 2,
}}
transition={{
duration: 0.8,
ease: [0.22, 1, 0.36, 1]
}}
/>
</motion.div>
)}
</AnimatePresence>
);
}

28
demo/nextjs_voice_chat/frontend/fastrtc-demo/eslint.config.mjs Normal file
View File

@@ -0,0 +1,28 @@
import { dirname } from "path";
import { fileURLToPath } from "url";
import { FlatCompat } from "@eslint/eslintrc";
const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);
const compat = new FlatCompat({
baseDirectory: __dirname,
});
const eslintConfig = [
...compat.extends("next/core-web-vitals", "next/typescript"),
{
rules: {
"no-unused-vars": "off",
"no-explicit-any": "off",
"no-console": "off",
"no-debugger": "off",
"eqeqeq": "off",
"curly": "off",
"quotes": "off",
"semi": "off",
},
},
];
export default eslintConfig;

6
demo/nextjs_voice_chat/frontend/fastrtc-demo/lib/utils.ts Normal file
View File

@@ -0,0 +1,6 @@
import { clsx, type ClassValue } from "clsx"
import { twMerge } from "tailwind-merge"
export function cn(...inputs: ClassValue[]) {
return twMerge(clsx(inputs))
}

189
demo/nextjs_voice_chat/frontend/fastrtc-demo/lib/webrtc-client.ts Normal file
View File

@@ -0,0 +1,189 @@
interface WebRTCClientOptions {
onConnected?: () => void;
onDisconnected?: () => void;
onMessage?: (message: any) => void;
onAudioStream?: (stream: MediaStream) => void;
onAudioLevel?: (level: number) => void;
}
export class WebRTCClient {
private peerConnection: RTCPeerConnection | null = null;
private mediaStream: MediaStream | null = null;
private dataChannel: RTCDataChannel | null = null;
private options: WebRTCClientOptions;
private audioContext: AudioContext | null = null;
private analyser: AnalyserNode | null = null;
private dataArray: Uint8Array | null = null;
private animationFrameId: number | null = null;
constructor(options: WebRTCClientOptions = {}) {
this.options = options;
}
async connect() {
try {
this.peerConnection = new RTCPeerConnection();
// Get user media
try {
this.mediaStream = await navigator.mediaDevices.getUserMedia({
audio: true
});
} catch (mediaError: any) {
console.error('Media error:', mediaError);
if (mediaError.name === 'NotAllowedError') {
throw new Error('Microphone access denied. Please allow microphone access and try again.');
} else if (mediaError.name === 'NotFoundError') {
throw new Error('No microphone detected. Please connect a microphone and try again.');
} else {
throw mediaError;
}
}
this.setupAudioAnalysis();
this.mediaStream.getTracks().forEach(track => {
if (this.peerConnection) {
this.peerConnection.addTrack(track, this.mediaStream!);
}
});
this.peerConnection.addEventListener('track', (event) => {
if (this.options.onAudioStream) {
this.options.onAudioStream(event.streams[0]);
}
});
this.dataChannel = this.peerConnection.createDataChannel('text');
this.dataChannel.addEventListener('message', (event) => {
try {
const message = JSON.parse(event.data);
console.log('Received message:', message);
if (this.options.onMessage) {
this.options.onMessage(message);
}
} catch (error) {
console.error('Error parsing message:', error);
}
});
// Create and send offer
const offer = await this.peerConnection.createOffer();
await this.peerConnection.setLocalDescription(offer);
// Use same-origin request to avoid CORS preflight
const response = await fetch('http://localhost:8000/webrtc/offer', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Accept': 'application/json'
},
mode: 'cors', // Explicitly set CORS mode
credentials: 'same-origin',
body: JSON.stringify({
sdp: offer.sdp,
type: offer.type,
webrtc_id: Math.random().toString(36).substring(7)
})
});
const serverResponse = await response.json();
await this.peerConnection.setRemoteDescription(serverResponse);
if (this.options.onConnected) {
this.options.onConnected();
}
} catch (error) {
console.error('Error connecting:', error);
this.disconnect();
throw error;
}
}
private setupAudioAnalysis() {
if (!this.mediaStream) return;
try {
this.audioContext = new AudioContext();
this.analyser = this.audioContext.createAnalyser();
this.analyser.fftSize = 256;
const source = this.audioContext.createMediaStreamSource(this.mediaStream);
source.connect(this.analyser);
const bufferLength = this.analyser.frequencyBinCount;
this.dataArray = new Uint8Array(bufferLength);
this.startAnalysis();
} catch (error) {
console.error('Error setting up audio analysis:', error);
}
}
private startAnalysis() {
if (!this.analyser || !this.dataArray || !this.options.onAudioLevel) return;
// Add throttling to prevent too many updates
let lastUpdateTime = 0;
const throttleInterval = 100; // Only update every 100ms
const analyze = () => {
this.analyser!.getByteFrequencyData(this.dataArray!);
const currentTime = Date.now();
// Only update if enough time has passed since last update
if (currentTime - lastUpdateTime > throttleInterval) {
// Calculate average volume level (0-1)
let sum = 0;
for (let i = 0; i < this.dataArray!.length; i++) {
sum += this.dataArray![i];
}
const average = sum / this.dataArray!.length / 255;
this.options.onAudioLevel!(average);
lastUpdateTime = currentTime;
}
this.animationFrameId = requestAnimationFrame(analyze);
};
this.animationFrameId = requestAnimationFrame(analyze);
}
private stopAnalysis() {
if (this.animationFrameId !== null) {
cancelAnimationFrame(this.animationFrameId);
this.animationFrameId = null;
}
if (this.audioContext) {
this.audioContext.close();
this.audioContext = null;
}
this.analyser = null;
this.dataArray = null;
}
disconnect() {
this.stopAnalysis();
if (this.mediaStream) {
this.mediaStream.getTracks().forEach(track => track.stop());
this.mediaStream = null;
}
if (this.peerConnection) {
this.peerConnection.close();
this.peerConnection = null;
}
this.dataChannel = null;
if (this.options.onDisconnected) {
this.options.onDisconnected();
}
}
}

7
demo/nextjs_voice_chat/frontend/fastrtc-demo/next.config.ts Normal file
View File

@@ -0,0 +1,7 @@
import type { NextConfig } from "next";
const nextConfig: NextConfig = {
/* config options here */
};
export default nextConfig;

33
demo/nextjs_voice_chat/frontend/fastrtc-demo/package.json Normal file
View File

@@ -0,0 +1,33 @@
{
"name": "fastrtc-demo",
"version": "0.1.0",
"private": true,
"scripts": {
"dev": "next dev --turbopack",
"build": "next build --no-lint",
"start": "next start",
"lint": "next lint"
},
"dependencies": {
"class-variance-authority": "^0.7.1",
"clsx": "^2.1.1",
"framer-motion": "^12.4.10",
"lucide-react": "^0.477.0",
"next": "15.2.2-canary.1",
"react": "^19.0.0",
"react-dom": "^19.0.0",
"tailwind-merge": "^3.0.2",
"tailwindcss-animate": "^1.0.7"
},
"devDependencies": {
"@eslint/eslintrc": "^3",
"@tailwindcss/postcss": "^4",
"@types/node": "^20",
"@types/react": "^19",
"@types/react-dom": "^19",
"eslint": "^9",
"eslint-config-next": "15.2.2-canary.1",
"tailwindcss": "^4",
"typescript": "^5"
}
}

5
demo/nextjs_voice_chat/frontend/fastrtc-demo/postcss.config.mjs Normal file
View File

@@ -0,0 +1,5 @@
const config = {
plugins: ["@tailwindcss/postcss"],
};
export default config;

1
demo/nextjs_voice_chat/frontend/fastrtc-demo/public/file.svg Normal file
View File

@@ -0,0 +1 @@
<svg fill="none" viewBox="0 0 16 16" xmlns="http://www.w3.org/2000/svg"><path d="M14.5 13.5V5.41a1 1 0 0 0-.3-.7L9.8.29A1 1 0 0 0 9.08 0H1.5v13.5A2.5 2.5 0 0 0 4 16h8a2.5 2.5 0 0 0 2.5-2.5m-1.5 0v-7H8v-5H3v12a1 1 0 0 0 1 1h8a1 1 0 0 0 1-1M9.5 5V2.12L12.38 5zM5.13 5h-.62v1.25h2.12V5zm-.62 3h7.12v1.25H4.5zm.62 3h-.62v1.25h7.12V11z" clip-rule="evenodd" fill="#666" fill-rule="evenodd"/></svg>
Side by Side

After

Width:  |  Height:  |  Size: 391 B

1
demo/nextjs_voice_chat/frontend/fastrtc-demo/public/globe.svg Normal file
View File

@@ -0,0 +1 @@
<svg fill="none" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><g clip-path="url(#a)"><path fill-rule="evenodd" clip-rule="evenodd" d="M10.27 14.1a6.5 6.5 0 0 0 3.67-3.45q-1.24.21-2.7.34-.31 1.83-.97 3.1M8 16A8 8 0 1 0 8 0a8 8 0 0 0 0 16m.48-1.52a7 7 0 0 1-.96 0H7.5a4 4 0 0 1-.84-1.32q-.38-.89-.63-2.08a40 40 0 0 0 3.92 0q-.25 1.2-.63 2.08a4 4 0 0 1-.84 1.31zm2.94-4.76q1.66-.15 2.95-.43a7 7 0 0 0 0-2.58q-1.3-.27-2.95-.43a18 18 0 0 1 0 3.44m-1.27-3.54a17 17 0 0 1 0 3.64 39 39 0 0 1-4.3 0 17 17 0 0 1 0-3.64 39 39 0 0 1 4.3 0m1.1-1.17q1.45.13 2.69.34a6.5 6.5 0 0 0-3.67-3.44q.65 1.26.98 3.1M8.48 1.5l.01.02q.41.37.84 1.31.38.89.63 2.08a40 40 0 0 0-3.92 0q.25-1.2.63-2.08a4 4 0 0 1 .85-1.32 7 7 0 0 1 .96 0m-2.75.4a6.5 6.5 0 0 0-3.67 3.44 29 29 0 0 1 2.7-.34q.31-1.83.97-3.1M4.58 6.28q-1.66.16-2.95.43a7 7 0 0 0 0 2.58q1.3.27 2.95.43a18 18 0 0 1 0-3.44m.17 4.71q-1.45-.12-2.69-.34a6.5 6.5 0 0 0 3.67 3.44q-.65-1.27-.98-3.1" fill="#666"/></g><defs><clipPath id="a"><path fill="#fff" d="M0 0h16v16H0z"/></clipPath></defs></svg>
Side by Side

After

Width:  |  Height:  |  Size: 1.0 KiB

1
demo/nextjs_voice_chat/frontend/fastrtc-demo/public/next.svg Normal file
View File

@@ -0,0 +1 @@
<svg xmlns="http://www.w3.org/2000/svg" fill="none" viewBox="0 0 394 80"><path fill="#000" d="M262 0h68.5v12.7h-27.2v66.6h-13.6V12.7H262V0ZM149 0v12.7H94v20.4h44.3v12.6H94v21h55v12.6H80.5V0h68.7zm34.3 0h-17.8l63.8 79.4h17.9l-32-39.7 32-39.6h-17.9l-23 28.6-23-28.6zm18.3 56.7-9-11-27.1 33.7h17.8l18.3-22.7z"/><path fill="#000" d="M81 79.3 17 0H0v79.3h13.6V17l50.2 62.3H81Zm252.6-.4c-1 0-1.8-.4-2.5-1s-1.1-1.6-1.1-2.6.3-1.8 1-2.5 1.6-1 2.6-1 1.8.3 2.5 1a3.4 3.4 0 0 1 .6 4.3 3.7 3.7 0 0 1-3 1.8zm23.2-33.5h6v23.3c0 2.1-.4 4-1.3 5.5a9.1 9.1 0 0 1-3.8 3.5c-1.6.8-3.5 1.3-5.7 1.3-2 0-3.7-.4-5.3-1s-2.8-1.8-3.7-3.2c-.9-1.3-1.4-3-1.4-5h6c.1.8.3 1.6.7 2.2s1 1.2 1.6 1.5c.7.4 1.5.5 2.4.5 1 0 1.8-.2 2.4-.6a4 4 0 0 0 1.6-1.8c.3-.8.5-1.8.5-3V45.5zm30.9 9.1a4.4 4.4 0 0 0-2-3.3 7.5 7.5 0 0 0-4.3-1.1c-1.3 0-2.4.2-3.3.5-.9.4-1.6 1-2 1.6a3.5 3.5 0 0 0-.3 4c.3.5.7.9 1.3 1.2l1.8 1 2 .5 3.2.8c1.3.3 2.5.7 3.7 1.2a13 13 0 0 1 3.2 1.8 8.1 8.1 0 0 1 3 6.5c0 2-.5 3.7-1.5 5.1a10 10 0 0 1-4.4 3.5c-1.8.8-4.1 1.2-6.8 1.2-2.6 0-4.9-.4-6.8-1.2-2-.8-3.4-2-4.5-3.5a10 10 0 0 1-1.7-5.6h6a5 5 0 0 0 3.5 4.6c1 .4 2.2.6 3.4.6 1.3 0 2.5-.2 3.5-.6 1-.4 1.8-1 2.4-1.7a4 4 0 0 0 .8-2.4c0-.9-.2-1.6-.7-2.2a11 11 0 0 0-2.1-1.4l-3.2-1-3.8-1c-2.8-.7-5-1.7-6.6-3.2a7.2 7.2 0 0 1-2.4-5.7 8 8 0 0 1 1.7-5 10 10 0 0 1 4.3-3.5c2-.8 4-1.2 6.4-1.2 2.3 0 4.4.4 6.2 1.2 1.8.8 3.2 2 4.3 3.4 1 1.4 1.5 3 1.5 5h-5.8z"/></svg>
Side by Side

After

Width:  |  Height:  |  Size: 1.3 KiB

1
demo/nextjs_voice_chat/frontend/fastrtc-demo/public/vercel.svg Normal file
View File

@@ -0,0 +1 @@
<svg fill="none" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1155 1000"><path d="m577.3 0 577.4 1000H0z" fill="#fff"/></svg>
Side by Side

After

Width:  |  Height:  |  Size: 128 B

1
demo/nextjs_voice_chat/frontend/fastrtc-demo/public/window.svg Normal file
View File

@@ -0,0 +1 @@
<svg fill="none" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16"><path fill-rule="evenodd" clip-rule="evenodd" d="M1.5 2.5h13v10a1 1 0 0 1-1 1h-11a1 1 0 0 1-1-1zM0 1h16v11.5a2.5 2.5 0 0 1-2.5 2.5h-11A2.5 2.5 0 0 1 0 12.5zm3.75 4.5a.75.75 0 1 0 0-1.5.75.75 0 0 0 0 1.5M7 4.75a.75.75 0 1 1-1.5 0 .75.75 0 0 1 1.5 0m1.75.75a.75.75 0 1 0 0-1.5.75.75 0 0 0 0 1.5" fill="#666"/></svg>
Side by Side

After

Width:  |  Height:  |  Size: 385 B

27
demo/nextjs_voice_chat/frontend/fastrtc-demo/tsconfig.json Normal file
View File

@@ -0,0 +1,27 @@
{
"compilerOptions": {
"target": "ES2017",
"lib": ["dom", "dom.iterable", "esnext"],
"allowJs": true,
"skipLibCheck": true,
"strict": true,
"noEmit": true,
"esModuleInterop": true,
"module": "esnext",
"moduleResolution": "bundler",
"resolveJsonModule": true,
"isolatedModules": true,
"jsx": "preserve",
"incremental": true,
"plugins": [
{
"name": "next"
}
],
"paths": {
"@/*": ["./*"]
}
},
"include": ["next-env.d.ts", "**/*.ts", "**/*.tsx", ".next/types/**/*.ts"],
"exclude": ["node_modules"]
}

5
demo/nextjs_voice_chat/requirements.txt Normal file
View File

@@ -0,0 +1,5 @@
openai
fastapi
python-dotenv
elevenlabs
fastrtc[vad, stt, tts]

1
demo/nextjs_voice_chat/run.sh Executable file
View File

@@ -0,0 +1 @@
uvicorn backend.server:app --host 0.0.0.0 --port 8000

15
demo/object_detection/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Object Detection
emoji: 📸
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Use YOLOv10 to detect objects in real-time
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

77
demo/object_detection/app.py Normal file
View File

@@ -0,0 +1,77 @@
import json
from pathlib import Path
import cv2
import gradio as gr
from fastapi import FastAPI
from fastapi.responses import HTMLResponse
from fastrtc import Stream, get_twilio_turn_credentials
from gradio.utils import get_space
from huggingface_hub import hf_hub_download
from pydantic import BaseModel, Field
try:
from demo.object_detection.inference import YOLOv10
except (ImportError, ModuleNotFoundError):
from inference import YOLOv10
cur_dir = Path(__file__).parent
model_file = hf_hub_download(
repo_id="onnx-community/yolov10n", filename="onnx/model.onnx"
)
model = YOLOv10(model_file)
def detection(image, conf_threshold=0.3):
image = cv2.resize(image, (model.input_width, model.input_height))
print("conf_threshold", conf_threshold)
new_image = model.detect_objects(image, conf_threshold)
return cv2.resize(new_image, (500, 500))
stream = Stream(
handler=detection,
modality="video",
mode="send-receive",
additional_inputs=[gr.Slider(minimum=0, maximum=1, step=0.01, value=0.3)],
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=2 if get_space() else None,
)
app = FastAPI()
stream.mount(app)
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = open(cur_dir / "index.html").read()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content)
class InputData(BaseModel):
webrtc_id: str
conf_threshold: float = Field(ge=0, le=1)
@app.post("/input_hook")
async def _(data: InputData):
stream.set_input(data.webrtc_id, data.conf_threshold)
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

340
demo/object_detection/index.html Normal file
View File

@@ -0,0 +1,340 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Object Detection</title>
<style>
body {
font-family: system-ui, -apple-system, sans-serif;
background: linear-gradient(135deg, #2d2b52 0%, #191731 100%);
color: white;
margin: 0;
padding: 20px;
height: 100vh;
box-sizing: border-box;
display: flex;
flex-direction: column;
align-items: center;
justify-content: center;
}
.container {
width: 100%;
max-width: 800px;
text-align: center;
display: flex;
flex-direction: column;
align-items: center;
}
.video-container {
width: 100%;
max-width: 500px;
aspect-ratio: 1/1;
background: rgba(255, 255, 255, 0.1);
border-radius: 12px;
overflow: hidden;
box-shadow: 0 8px 32px rgba(0, 0, 0, 0.2);
margin: 10px 0;
}
#video-output {
width: 100%;
height: 100%;
object-fit: cover;
}
button {
background: white;
color: #2d2b52;
border: none;
padding: 12px 32px;
border-radius: 24px;
font-size: 16px;
font-weight: 600;
cursor: pointer;
transition: all 0.3s ease;
box-shadow: 0 4px 12px rgba(0, 0, 0, 0.1);
}
button:hover {
transform: translateY(-2px);
box-shadow: 0 6px 16px rgba(0, 0, 0, 0.2);
}
h1 {
font-size: 2.5em;
margin-bottom: 0.3em;
}
p {
color: rgba(255, 255, 255, 0.8);
margin-bottom: 1em;
}
.controls {
display: flex;
flex-direction: column;
gap: 12px;
align-items: center;
margin-top: 10px;
}
.slider-container {
width: 100%;
max-width: 300px;
display: flex;
flex-direction: column;
gap: 8px;
}
.slider-container label {
color: rgba(255, 255, 255, 0.8);
font-size: 14px;
}
input[type="range"] {
width: 100%;
height: 6px;
-webkit-appearance: none;
background: rgba(255, 255, 255, 0.1);
border-radius: 3px;
outline: none;
}
input[type="range"]::-webkit-slider-thumb {
-webkit-appearance: none;
width: 18px;
height: 18px;
background: white;
border-radius: 50%;
cursor: pointer;
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div class="container">
<h1>Real-time Object Detection</h1>
<p>Using YOLOv10 to detect objects in your webcam feed</p>
<div class="video-container">
<video id="video-output" autoplay playsinline></video>
</div>
<div class="controls">
<div class="slider-container">
<label>Confidence Threshold: <span id="conf-value">0.3</span></label>
<input type="range" id="conf-threshold" min="0" max="1" step="0.01" value="0.3">
</div>
<button id="start-button">Start</button>
</div>
</div>
<script>
let peerConnection;
let webrtc_id;
const startButton = document.getElementById('start-button');
const videoOutput = document.getElementById('video-output');
const confThreshold = document.getElementById('conf-threshold');
const confValue = document.getElementById('conf-value');
// Update confidence value display
confThreshold.addEventListener('input', (e) => {
confValue.textContent = e.target.value;
if (peerConnection) {
updateConfThreshold(e.target.value);
}
});
function updateConfThreshold(value) {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: webrtc_id,
conf_threshold: parseFloat(value)
})
});
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.className = 'toast error';
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
async function setupWebRTC() {
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({
video: true
});
stream.getTracks().forEach(track => {
peerConnection.addTrack(track, stream);
});
peerConnection.addEventListener('track', (evt) => {
if (videoOutput && videoOutput.srcObject !== evt.streams[0]) {
videoOutput.srcObject = evt.streams[0];
}
});
const dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = (event) => {
const eventJson = JSON.parse(event.data);
if (eventJson.type === "error") {
showError(eventJson.message);
} else if (eventJson.type === "send_input") {
updateConfThreshold(confThreshold.value);
}
};
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
webrtc_id = Math.random().toString(36).substring(7);
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
startButton.textContent = 'Start';
return;
}
await peerConnection.setRemoteDescription(serverResponse);
// Send initial confidence threshold
updateConfThreshold(confThreshold.value);
peerConnection.addEventListener('connectionstatechange', () => {
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
});
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
startButton.textContent = 'Start';
}
}
function stop() {
if (peerConnection) {
if (peerConnection.getTransceivers) {
peerConnection.getTransceivers().forEach(transceiver => {
if (transceiver.stop) {
transceiver.stop();
}
});
}
if (peerConnection.getSenders) {
peerConnection.getSenders().forEach(sender => {
if (sender.track && sender.track.stop) sender.track.stop();
});
}
setTimeout(() => {
peerConnection.close();
}, 500);
}
videoOutput.srcObject = null;
}
startButton.addEventListener('click', () => {
if (startButton.textContent === 'Start') {
setupWebRTC();
startButton.textContent = 'Stop';
} else {
stop();
startButton.textContent = 'Start';
}
});
</script>
</body>
</html>

153
demo/object_detection/inference.py Normal file
View File

@@ -0,0 +1,153 @@
import time
import cv2
import numpy as np
import onnxruntime
try:
from demo.object_detection.utils import draw_detections
except (ImportError, ModuleNotFoundError):
from utils import draw_detections
class YOLOv10:
def __init__(self, path):
# Initialize model
self.initialize_model(path)
def __call__(self, image):
return self.detect_objects(image)
def initialize_model(self, path):
self.session = onnxruntime.InferenceSession(
path, providers=onnxruntime.get_available_providers()
)
# Get model info
self.get_input_details()
self.get_output_details()
def detect_objects(self, image, conf_threshold=0.3):
input_tensor = self.prepare_input(image)
# Perform inference on the image
new_image = self.inference(image, input_tensor, conf_threshold)
return new_image
def prepare_input(self, image):
self.img_height, self.img_width = image.shape[:2]
input_img = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
# Resize input image
input_img = cv2.resize(input_img, (self.input_width, self.input_height))
# Scale input pixel values to 0 to 1
input_img = input_img / 255.0
input_img = input_img.transpose(2, 0, 1)
input_tensor = input_img[np.newaxis, :, :, :].astype(np.float32)
return input_tensor
def inference(self, image, input_tensor, conf_threshold=0.3):
start = time.perf_counter()
outputs = self.session.run(
self.output_names, {self.input_names[0]: input_tensor}
)
print(f"Inference time: {(time.perf_counter() - start) * 1000:.2f} ms")
(
boxes,
scores,
class_ids,
) = self.process_output(outputs, conf_threshold)
return self.draw_detections(image, boxes, scores, class_ids)
def process_output(self, output, conf_threshold=0.3):
predictions = np.squeeze(output[0])
# Filter out object confidence scores below threshold
scores = predictions[:, 4]
predictions = predictions[scores > conf_threshold, :]
scores = scores[scores > conf_threshold]
if len(scores) == 0:
return [], [], []
# Get the class with the highest confidence
class_ids = predictions[:, 5].astype(int)
# Get bounding boxes for each object
boxes = self.extract_boxes(predictions)
return boxes, scores, class_ids
def extract_boxes(self, predictions):
# Extract boxes from predictions
boxes = predictions[:, :4]
# Scale boxes to original image dimensions
boxes = self.rescale_boxes(boxes)
# Convert boxes to xyxy format
# boxes = xywh2xyxy(boxes)
return boxes
def rescale_boxes(self, boxes):
# Rescale boxes to original image dimensions
input_shape = np.array(
[self.input_width, self.input_height, self.input_width, self.input_height]
)
boxes = np.divide(boxes, input_shape, dtype=np.float32)
boxes *= np.array(
[self.img_width, self.img_height, self.img_width, self.img_height]
)
return boxes
def draw_detections(
self, image, boxes, scores, class_ids, draw_scores=True, mask_alpha=0.4
):
return draw_detections(image, boxes, scores, class_ids, mask_alpha)
def get_input_details(self):
model_inputs = self.session.get_inputs()
self.input_names = [model_inputs[i].name for i in range(len(model_inputs))]
self.input_shape = model_inputs[0].shape
self.input_height = self.input_shape[2]
self.input_width = self.input_shape[3]
def get_output_details(self):
model_outputs = self.session.get_outputs()
self.output_names = [model_outputs[i].name for i in range(len(model_outputs))]
if __name__ == "__main__":
import tempfile
import requests
from huggingface_hub import hf_hub_download
model_file = hf_hub_download(
repo_id="onnx-community/yolov10s", filename="onnx/model.onnx"
)
yolov8_detector = YOLOv10(model_file)
with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as f:
f.write(
requests.get(
"https://live.staticflickr.com/13/19041780_d6fd803de0_3k.jpg"
).content
)
f.seek(0)
img = cv2.imread(f.name)
# # Detect Objects
combined_image = yolov8_detector.detect_objects(img)
# Draw detections
cv2.namedWindow("Output", cv2.WINDOW_NORMAL)
cv2.imshow("Output", combined_image)
cv2.waitKey(0)

4
demo/object_detection/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc
opencv-python
twilio
onnxruntime-gpu

237
demo/object_detection/utils.py Normal file
View File

@@ -0,0 +1,237 @@
import cv2
import numpy as np
class_names = [
"person",
"bicycle",
"car",
"motorcycle",
"airplane",
"bus",
"train",
"truck",
"boat",
"traffic light",
"fire hydrant",
"stop sign",
"parking meter",
"bench",
"bird",
"cat",
"dog",
"horse",
"sheep",
"cow",
"elephant",
"bear",
"zebra",
"giraffe",
"backpack",
"umbrella",
"handbag",
"tie",
"suitcase",
"frisbee",
"skis",
"snowboard",
"sports ball",
"kite",
"baseball bat",
"baseball glove",
"skateboard",
"surfboard",
"tennis racket",
"bottle",
"wine glass",
"cup",
"fork",
"knife",
"spoon",
"bowl",
"banana",
"apple",
"sandwich",
"orange",
"broccoli",
"carrot",
"hot dog",
"pizza",
"donut",
"cake",
"chair",
"couch",
"potted plant",
"bed",
"dining table",
"toilet",
"tv",
"laptop",
"mouse",
"remote",
"keyboard",
"cell phone",
"microwave",
"oven",
"toaster",
"sink",
"refrigerator",
"book",
"clock",
"vase",
"scissors",
"teddy bear",
"hair drier",
"toothbrush",
]
# Create a list of colors for each class where each color is a tuple of 3 integer values
rng = np.random.default_rng(3)
colors = rng.uniform(0, 255, size=(len(class_names), 3))
def nms(boxes, scores, iou_threshold):
# Sort by score
sorted_indices = np.argsort(scores)[::-1]
keep_boxes = []
while sorted_indices.size > 0:
# Pick the last box
box_id = sorted_indices[0]
keep_boxes.append(box_id)
# Compute IoU of the picked box with the rest
ious = compute_iou(boxes[box_id, :], boxes[sorted_indices[1:], :])
# Remove boxes with IoU over the threshold
keep_indices = np.where(ious < iou_threshold)[0]
# print(keep_indices.shape, sorted_indices.shape)
sorted_indices = sorted_indices[keep_indices + 1]
return keep_boxes
def multiclass_nms(boxes, scores, class_ids, iou_threshold):
unique_class_ids = np.unique(class_ids)
keep_boxes = []
for class_id in unique_class_ids:
class_indices = np.where(class_ids == class_id)[0]
class_boxes = boxes[class_indices, :]
class_scores = scores[class_indices]
class_keep_boxes = nms(class_boxes, class_scores, iou_threshold)
keep_boxes.extend(class_indices[class_keep_boxes])
return keep_boxes
def compute_iou(box, boxes):
# Compute xmin, ymin, xmax, ymax for both boxes
xmin = np.maximum(box[0], boxes[:, 0])
ymin = np.maximum(box[1], boxes[:, 1])
xmax = np.minimum(box[2], boxes[:, 2])
ymax = np.minimum(box[3], boxes[:, 3])
# Compute intersection area
intersection_area = np.maximum(0, xmax - xmin) * np.maximum(0, ymax - ymin)
# Compute union area
box_area = (box[2] - box[0]) * (box[3] - box[1])
boxes_area = (boxes[:, 2] - boxes[:, 0]) * (boxes[:, 3] - boxes[:, 1])
union_area = box_area + boxes_area - intersection_area
# Compute IoU
iou = intersection_area / union_area
return iou
def xywh2xyxy(x):
# Convert bounding box (x, y, w, h) to bounding box (x1, y1, x2, y2)
y = np.copy(x)
y[..., 0] = x[..., 0] - x[..., 2] / 2
y[..., 1] = x[..., 1] - x[..., 3] / 2
y[..., 2] = x[..., 0] + x[..., 2] / 2
y[..., 3] = x[..., 1] + x[..., 3] / 2
return y
def draw_detections(image, boxes, scores, class_ids, mask_alpha=0.3):
det_img = image.copy()
img_height, img_width = image.shape[:2]
font_size = min([img_height, img_width]) * 0.0006
text_thickness = int(min([img_height, img_width]) * 0.001)
# det_img = draw_masks(det_img, boxes, class_ids, mask_alpha)
# Draw bounding boxes and labels of detections
for class_id, box, score in zip(class_ids, boxes, scores):
color = colors[class_id]
draw_box(det_img, box, color) # type: ignore
label = class_names[class_id]
caption = f"{label} {int(score * 100)}%"
draw_text(det_img, caption, box, color, font_size, text_thickness) # type: ignore
return det_img
def draw_box(
image: np.ndarray,
box: np.ndarray,
color: tuple[int, int, int] = (0, 0, 255),
thickness: int = 2,
) -> np.ndarray:
x1, y1, x2, y2 = box.astype(int)
return cv2.rectangle(image, (x1, y1), (x2, y2), color, thickness)
def draw_text(
image: np.ndarray,
text: str,
box: np.ndarray,
color: tuple[int, int, int] = (0, 0, 255),
font_size: float = 0.001,
text_thickness: int = 2,
) -> np.ndarray:
x1, y1, x2, y2 = box.astype(int)
(tw, th), _ = cv2.getTextSize(
text=text,
fontFace=cv2.FONT_HERSHEY_SIMPLEX,
fontScale=font_size,
thickness=text_thickness,
)
th = int(th * 1.2)
cv2.rectangle(image, (x1, y1), (x1 + tw, y1 - th), color, -1)
return cv2.putText(
image,
text,
(x1, y1),
cv2.FONT_HERSHEY_SIMPLEX,
font_size,
(255, 255, 255),
text_thickness,
cv2.LINE_AA,
)
def draw_masks(
image: np.ndarray, boxes: np.ndarray, classes: np.ndarray, mask_alpha: float = 0.3
) -> np.ndarray:
mask_img = image.copy()
# Draw bounding boxes and labels of detections
for box, class_id in zip(boxes, classes):
color = colors[class_id]
x1, y1, x2, y2 = box.astype(int)
# Draw fill rectangle in mask image
cv2.rectangle(mask_img, (x1, y1), (x2, y2), color, -1) # type: ignore
return cv2.addWeighted(mask_img, mask_alpha, image, 1 - mask_alpha, 0)

16
demo/phonic_chat/README.md Normal file
View File

@@ -0,0 +1,16 @@
---
title: Phonic AI Chat
emoji: 🎙️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to Phonic AI's speech-to-speech model
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|PHONIC_API_KEY]
python_version: 3.11
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

116
demo/phonic_chat/app.py Normal file
View File

@@ -0,0 +1,116 @@
import asyncio
import base64
import os
import gradio as gr
from gradio.utils import get_space
import numpy as np
from dotenv import load_dotenv
from fastrtc import (
AdditionalOutputs,
AsyncStreamHandler,
Stream,
get_twilio_turn_credentials,
audio_to_float32,
wait_for_item,
)
from phonic.client import PhonicSTSClient, get_voices
load_dotenv()
STS_URI = "wss://api.phonic.co/v1/sts/ws"
API_KEY = os.environ["PHONIC_API_KEY"]
SAMPLE_RATE = 44_100
voices = get_voices(API_KEY)
voice_ids = [voice["id"] for voice in voices]
class PhonicHandler(AsyncStreamHandler):
def __init__(self):
super().__init__(input_sample_rate=SAMPLE_RATE, output_sample_rate=SAMPLE_RATE)
self.output_queue = asyncio.Queue()
self.client = None
def copy(self) -> AsyncStreamHandler:
return PhonicHandler()
async def start_up(self):
await self.wait_for_args()
voice_id = self.latest_args[1]
async with PhonicSTSClient(STS_URI, API_KEY) as client:
self.client = client
sts_stream = client.sts( # type: ignore
input_format="pcm_44100",
output_format="pcm_44100",
system_prompt="You are a helpful voice assistant. Respond conversationally.",
# welcome_message="Hello! I'm your voice assistant. How can I help you today?",
voice_id=voice_id,
)
async for message in sts_stream:
message_type = message.get("type")
if message_type == "audio_chunk":
audio_b64 = message["audio"]
audio_bytes = base64.b64decode(audio_b64)
await self.output_queue.put(
(SAMPLE_RATE, np.frombuffer(audio_bytes, dtype=np.int16))
)
if text := message.get("text"):
msg = {"role": "assistant", "content": text}
await self.output_queue.put(AdditionalOutputs(msg))
elif message_type == "input_text":
msg = {"role": "user", "content": message["text"]}
await self.output_queue.put(AdditionalOutputs(msg))
async def emit(self):
return await wait_for_item(self.output_queue)
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
if not self.client:
return
audio_float32 = audio_to_float32(frame)
await self.client.send_audio(audio_float32) # type: ignore
async def shutdown(self):
if self.client:
await self.client._websocket.close()
return super().shutdown()
def add_to_chatbot(chatbot, message):
chatbot.append(message)
return chatbot
chatbot = gr.Chatbot(type="messages", value=[])
stream = Stream(
handler=PhonicHandler(),
mode="send-receive",
modality="audio",
additional_inputs=[
gr.Dropdown(
choices=voice_ids,
value="victoria",
label="Voice",
info="Select a voice from the dropdown",
)
],
additional_outputs=[chatbot],
additional_outputs_handler=add_to_chatbot,
ui_args={
"title": "Phonic Chat (Powered by FastRTC ⚡️)",
},
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
)
# with stream.ui:
# state.change(lambda s: s, inputs=state, outputs=chatbot)
if __name__ == "__main__":
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
stream.ui.launch(server_port=7860)

74
demo/phonic_chat/requirements.txt Normal file
View File

@@ -0,0 +1,74 @@
# This file was autogenerated by uv via the following command:
# uv pip compile requirements.in -o requirements.txt
aiohappyeyeballs==2.4.6
# via aiohttp
aiohttp==3.11.12
# via
# aiohttp-retry
# twilio
aiohttp-retry==2.9.1
# via twilio
aiosignal==1.3.2
# via aiohttp
attrs==25.1.0
# via aiohttp
certifi==2025.1.31
# via requests
cffi==1.17.1
# via sounddevice
charset-normalizer==3.4.1
# via requests
fastrtc==0.0.1
# via -r requirements.in
frozenlist==1.5.0
# via
# aiohttp
# aiosignal
idna==3.10
# via
# requests
# yarl
isort==6.0.0
# via phonic-python
loguru==0.7.3
# via phonic-python
multidict==6.1.0
# via
# aiohttp
# yarl
numpy==2.2.3
# via
# phonic-python
# scipy
phonic-python==0.1.3
# via -r requirements.in
propcache==0.3.0
# via
# aiohttp
# yarl
pycparser==2.22
# via cffi
pyjwt==2.10.1
# via twilio
python-dotenv==1.0.1
# via
# -r requirements.in
# phonic-python
requests==2.32.3
# via
# phonic-python
# twilio
scipy==1.15.2
# via phonic-python
sounddevice==0.5.1
# via phonic-python
twilio==9.4.6
# via -r requirements.in
typing-extensions==4.12.2
# via phonic-python
urllib3==2.3.0
# via requests
websockets==15.0
# via phonic-python
yarl==1.18.3
# via aiohttp

15
demo/talk_to_claude/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to Claude
emoji: 👨‍🦰
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to Anthropic's Claude
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GROQ_API_KEY, secret|ANTHROPIC_API_KEY, secret|ELEVENLABS_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

134
demo/talk_to_claude/app.py Normal file
View File

@@ -0,0 +1,134 @@
import json
import os
import time
from pathlib import Path
import anthropic
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from elevenlabs import ElevenLabs
from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import (
AdditionalOutputs,
ReplyOnPause,
Stream,
get_tts_model,
get_twilio_turn_credentials,
)
from fastrtc.utils import audio_to_bytes
from gradio.utils import get_space
from groq import Groq
from pydantic import BaseModel
load_dotenv()
groq_client = Groq()
claude_client = anthropic.Anthropic()
tts_client = ElevenLabs(api_key=os.environ["ELEVENLABS_API_KEY"])
curr_dir = Path(__file__).parent
tts_model = get_tts_model()
def response(
audio: tuple[int, np.ndarray],
chatbot: list[dict] | None = None,
):
chatbot = chatbot or []
messages = [{"role": d["role"], "content": d["content"]} for d in chatbot]
prompt = groq_client.audio.transcriptions.create(
file=("audio-file.mp3", audio_to_bytes(audio)),
model="whisper-large-v3-turbo",
response_format="verbose_json",
).text
chatbot.append({"role": "user", "content": prompt})
yield AdditionalOutputs(chatbot)
messages.append({"role": "user", "content": prompt})
response = claude_client.messages.create(
model="claude-3-5-haiku-20241022",
max_tokens=512,
messages=messages, # type: ignore
)
response_text = " ".join(
block.text # type: ignore
for block in response.content
if getattr(block, "type", None) == "text"
)
chatbot.append({"role": "assistant", "content": response_text})
start = time.time()
print("starting tts", start)
for i, chunk in enumerate(tts_model.stream_tts_sync(response_text)):
print("chunk", i, time.time() - start)
yield chunk
print("finished tts", time.time() - start)
yield AdditionalOutputs(chatbot)
chatbot = gr.Chatbot(type="messages")
stream = Stream(
modality="audio",
mode="send-receive",
handler=ReplyOnPause(response),
additional_outputs_handler=lambda a, b: b,
additional_inputs=[chatbot],
additional_outputs=[chatbot],
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
)
class Message(BaseModel):
role: str
content: str
class InputData(BaseModel):
webrtc_id: str
chatbot: list[Message]
app = FastAPI()
stream.mount(app)
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (curr_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content, status_code=200)
@app.post("/input_hook")
async def _(body: InputData):
stream.set_input(body.webrtc_id, body.model_dump()["chatbot"])
return {"status": "ok"}
@app.get("/outputs")
def _(webrtc_id: str):
async def output_stream():
async for output in stream.output_stream(webrtc_id):
chatbot = output.args[0]
yield f"event: output\ndata: {json.dumps(chatbot[-1])}\n\n"
return StreamingResponse(output_stream(), media_type="text/event-stream")
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

546
demo/talk_to_claude/index.html Normal file
View File

@@ -0,0 +1,546 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>RetroChat Audio</title>
<style>
body {
font-family: monospace;
background-color: #1a1a1a;
color: #00ff00;
margin: 0;
padding: 20px;
height: 100vh;
box-sizing: border-box;
}
.container {
display: flex;
flex-direction: column;
gap: 20px;
height: calc(100% - 100px);
margin-bottom: 20px;
}
.chat-container {
border: 2px solid #00ff00;
padding: 20px;
display: flex;
flex-direction: column;
flex-grow: 1;
box-sizing: border-box;
}
.controls-container {
border: 2px solid #00ff00;
padding: 20px;
display: flex;
align-items: center;
gap: 20px;
height: 128px;
box-sizing: border-box;
}
.visualization-container {
flex-grow: 1;
display: flex;
align-items: center;
}
.box-container {
display: flex;
justify-content: space-between;
height: 64px;
width: 100%;
}
.box {
height: 100%;
width: 8px;
background: #00ff00;
border-radius: 8px;
transition: transform 0.05s ease;
}
.chat-messages {
flex-grow: 1;
overflow-y: auto;
margin-bottom: 20px;
padding: 10px;
border: 1px solid #00ff00;
}
.message {
margin-bottom: 10px;
padding: 8px;
border-radius: 4px;
}
.message.user {
background-color: #003300;
}
.message.assistant {
background-color: #002200;
}
button {
height: 64px;
min-width: 120px;
background-color: #000;
color: #00ff00;
border: 2px solid #00ff00;
padding: 10px 20px;
font-family: monospace;
font-size: 16px;
cursor: pointer;
transition: all 0.3s;
}
button:hover {
border-width: 3px;
}
#audio-output {
display: none;
}
/* Retro CRT effect */
.crt-overlay {
position: absolute;
top: 0;
left: 0;
width: 100%;
height: 100%;
background: repeating-linear-gradient(0deg,
rgba(0, 255, 0, 0.03),
rgba(0, 255, 0, 0.03) 1px,
transparent 1px,
transparent 2px);
pointer-events: none;
}
/* Add these new styles */
.icon-with-spinner {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.spinner {
width: 20px;
height: 20px;
border: 2px solid #00ff00;
border-top-color: transparent;
border-radius: 50%;
animation: spin 1s linear infinite;
flex-shrink: 0;
}
@keyframes spin {
to {
transform: rotate(360deg);
}
}
.pulse-container {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.pulse-circle {
width: 20px;
height: 20px;
border-radius: 50%;
background-color: #00ff00;
opacity: 0.2;
flex-shrink: 0;
transform: translateX(-0%) scale(var(--audio-level, 1));
transition: transform 0.1s ease;
}
/* Add styles for typing indicator */
.typing-indicator {
padding: 8px;
background-color: #002200;
border-radius: 4px;
margin-bottom: 10px;
display: none;
}
.dots {
display: inline-flex;
gap: 4px;
}
.dot {
width: 8px;
height: 8px;
background-color: #00ff00;
border-radius: 50%;
animation: pulse 1.5s infinite;
opacity: 0.5;
}
.dot:nth-child(2) {
animation-delay: 0.5s;
}
.dot:nth-child(3) {
animation-delay: 1s;
}
@keyframes pulse {
0%,
100% {
opacity: 0.5;
transform: scale(1);
}
50% {
opacity: 1;
transform: scale(1.2);
}
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div class="container">
<div class="chat-container">
<div class="chat-messages" id="chat-messages"></div>
<!-- Move typing indicator outside the chat messages -->
<div class="typing-indicator" id="typing-indicator">
<div class="dots">
<div class="dot"></div>
<div class="dot"></div>
<div class="dot"></div>
</div>
</div>
</div>
<div class="controls-container">
<div class="visualization-container">
<div class="box-container">
<!-- Boxes will be dynamically added here -->
</div>
</div>
<button id="start-button">Start</button>
</div>
</div>
<audio id="audio-output"></audio>
<script>
let audioContext;
let analyser_input, analyser_output;
let dataArray_input, dataArray_output;
let animationId_input, animationId_output;
let chatHistory = [];
let peerConnection;
let webrtc_id;
const audioOutput = document.getElementById('audio-output');
const startButton = document.getElementById('start-button');
const chatMessages = document.getElementById('chat-messages');
function updateButtonState() {
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
startButton.innerHTML = `
<div class="icon-with-spinner">
<div class="spinner"></div>
<span>Connecting...</span>
</div>
`;
} else if (peerConnection && peerConnection.connectionState === 'connected') {
startButton.innerHTML = `
<div class="pulse-container">
<div class="pulse-circle"></div>
<span>Stop</span>
</div>
`;
} else {
startButton.innerHTML = 'Start';
}
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.className = 'toast error';
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
async function setupWebRTC() {
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({
audio: true
});
// Set up input visualization
audioContext = new AudioContext();
analyser_input = audioContext.createAnalyser();
const inputSource = audioContext.createMediaStreamSource(stream);
inputSource.connect(analyser_input);
analyser_input.fftSize = 64;
dataArray_input = new Uint8Array(analyser_input.frequencyBinCount);
function updateAudioLevel() {
analyser_input.getByteFrequencyData(dataArray_input);
const average = Array.from(dataArray_input).reduce((a, b) => a + b, 0) / dataArray_input.length;
const audioLevel = average / 255;
const pulseCircle = document.querySelector('.pulse-circle');
if (pulseCircle) {
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
}
animationId_input = requestAnimationFrame(updateAudioLevel);
}
updateAudioLevel();
stream.getTracks().forEach(track => {
peerConnection.addTrack(track, stream);
});
// Add connection state change listener
peerConnection.addEventListener('connectionstatechange', () => {
console.log('Connection state:', peerConnection.connectionState);
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
updateButtonState();
});
// Handle incoming audio
peerConnection.addEventListener('track', (evt) => {
if (audioOutput.srcObject !== evt.streams[0]) {
audioOutput.srcObject = evt.streams[0];
audioOutput.play();
// Set up output visualization
analyser_output = audioContext.createAnalyser();
const outputSource = audioContext.createMediaStreamSource(evt.streams[0]);
outputSource.connect(analyser_output);
analyser_output.fftSize = 2048;
dataArray_output = new Uint8Array(analyser_output.frequencyBinCount);
updateVisualization();
}
});
// Create data channel for messages
const dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = (event) => {
const eventJson = JSON.parse(event.data);
const typingIndicator = document.getElementById('typing-indicator');
if (eventJson.type === "error") {
showError(eventJson.message);
} else if (eventJson.type === "send_input") {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: webrtc_id,
chatbot: chatHistory
})
});
} else if (eventJson.type === "log") {
if (eventJson.data === "pause_detected") {
typingIndicator.style.display = 'block';
chatMessages.scrollTop = chatMessages.scrollHeight;
} else if (eventJson.data === "response_starting") {
typingIndicator.style.display = 'none';
}
}
};
// Create and send offer
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
webrtc_id = Math.random().toString(36).substring(7);
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
return;
}
await peerConnection.setRemoteDescription(serverResponse);
// Start visualization
updateVisualization();
// create event stream to receive messages from /output
const eventSource = new EventSource('/outputs?webrtc_id=' + webrtc_id);
eventSource.addEventListener("output", (event) => {
const eventJson = JSON.parse(event.data);
addMessage(eventJson.role, eventJson.content);
});
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
}
}
function addMessage(role, content) {
const messageDiv = document.createElement('div');
messageDiv.classList.add('message', role);
messageDiv.textContent = content;
chatMessages.appendChild(messageDiv);
chatMessages.scrollTop = chatMessages.scrollHeight;
chatHistory.push({ role, content });
}
// Add this after other const declarations
const boxContainer = document.querySelector('.box-container');
const numBars = 32;
for (let i = 0; i < numBars; i++) {
const box = document.createElement('div');
box.className = 'box';
boxContainer.appendChild(box);
}
// Replace the draw function with updateVisualization
function updateVisualization() {
animationId_output = requestAnimationFrame(updateVisualization);
analyser_output.getByteFrequencyData(dataArray_output);
const bars = document.querySelectorAll('.box');
for (let i = 0; i < bars.length; i++) {
const barHeight = (dataArray_output[i] / 255) * 2;
bars[i].style.transform = `scaleY(${Math.max(0.1, barHeight)})`;
}
}
function stop() {
if (peerConnection) {
if (peerConnection.getTransceivers) {
peerConnection.getTransceivers().forEach(transceiver => {
if (transceiver.stop) {
transceiver.stop();
}
});
}
if (peerConnection.getSenders) {
peerConnection.getSenders().forEach(sender => {
if (sender.track && sender.track.stop) sender.track.stop();
});
}
peerConnection.close();
}
if (animationId_input) {
cancelAnimationFrame(animationId_input);
}
if (animationId_output) {
cancelAnimationFrame(animationId_output);
}
if (audioContext) {
audioContext.close();
}
updateButtonState();
}
startButton.addEventListener('click', () => {
if (startButton.textContent === 'Start') {
setupWebRTC();
} else {
stop();
}
});
</script>
</body>
</html>

6
demo/talk_to_claude/requirements.txt Normal file
View File

@@ -0,0 +1,6 @@
fastrtc[vad, tts]
elevenlabs
groq
anthropic
twilio
python-dotenv

15
demo/talk_to_gemini/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to Gemini
emoji: ♊️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to Gemini using Google's multimodal API
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GEMINI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

15
demo/talk_to_gemini/README_gradio.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to Gemini (Gradio UI)
emoji: ♊️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to Gemini (Gradio UI)
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|GEMINI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

181
demo/talk_to_gemini/app.py Normal file
View File

@@ -0,0 +1,181 @@
import asyncio
import base64
import json
import os
import pathlib
from typing import AsyncGenerator, Literal
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from fastapi import FastAPI
from fastapi.responses import HTMLResponse
from fastrtc import (
AsyncStreamHandler,
Stream,
get_twilio_turn_credentials,
wait_for_item,
)
from google import genai
from google.genai.types import (
LiveConnectConfig,
PrebuiltVoiceConfig,
SpeechConfig,
VoiceConfig,
)
from gradio.utils import get_space
from pydantic import BaseModel
current_dir = pathlib.Path(__file__).parent
load_dotenv()
def encode_audio(data: np.ndarray) -> str:
"""Encode Audio data to send to the server"""
return base64.b64encode(data.tobytes()).decode("UTF-8")
class GeminiHandler(AsyncStreamHandler):
"""Handler for the Gemini API"""
def __init__(
self,
expected_layout: Literal["mono"] = "mono",
output_sample_rate: int = 24000,
output_frame_size: int = 480,
) -> None:
super().__init__(
expected_layout,
output_sample_rate,
output_frame_size,
input_sample_rate=16000,
)
self.input_queue: asyncio.Queue = asyncio.Queue()
self.output_queue: asyncio.Queue = asyncio.Queue()
self.quit: asyncio.Event = asyncio.Event()
def copy(self) -> "GeminiHandler":
return GeminiHandler(
expected_layout="mono",
output_sample_rate=self.output_sample_rate,
output_frame_size=self.output_frame_size,
)
async def start_up(self):
if not self.phone_mode:
await self.wait_for_args()
api_key, voice_name = self.latest_args[1:]
else:
api_key, voice_name = None, "Puck"
client = genai.Client(
api_key=api_key or os.getenv("GEMINI_API_KEY"),
http_options={"api_version": "v1alpha"},
)
config = LiveConnectConfig(
response_modalities=["AUDIO"], # type: ignore
speech_config=SpeechConfig(
voice_config=VoiceConfig(
prebuilt_voice_config=PrebuiltVoiceConfig(
voice_name=voice_name,
)
)
),
)
async with client.aio.live.connect(
model="gemini-2.0-flash-exp", config=config
) as session:
async for audio in session.start_stream(
stream=self.stream(), mime_type="audio/pcm"
):
if audio.data:
array = np.frombuffer(audio.data, dtype=np.int16)
self.output_queue.put_nowait((self.output_sample_rate, array))
async def stream(self) -> AsyncGenerator[bytes, None]:
while not self.quit.is_set():
try:
audio = await asyncio.wait_for(self.input_queue.get(), 0.1)
yield audio
except (asyncio.TimeoutError, TimeoutError):
pass
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
_, array = frame
array = array.squeeze()
audio_message = encode_audio(array)
self.input_queue.put_nowait(audio_message)
async def emit(self) -> tuple[int, np.ndarray] | None:
return await wait_for_item(self.output_queue)
def shutdown(self) -> None:
self.quit.set()
stream = Stream(
modality="audio",
mode="send-receive",
handler=GeminiHandler(),
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
additional_inputs=[
gr.Textbox(
label="API Key",
type="password",
value=os.getenv("GEMINI_API_KEY") if not get_space() else "",
),
gr.Dropdown(
label="Voice",
choices=[
"Puck",
"Charon",
"Kore",
"Fenrir",
"Aoede",
],
value="Puck",
),
],
)
class InputData(BaseModel):
webrtc_id: str
voice_name: str
api_key: str
app = FastAPI()
stream.mount(app)
@app.post("/input_hook")
async def _(body: InputData):
stream.set_input(body.webrtc_id, body.api_key, body.voice_name)
return {"status": "ok"}
@app.get("/")
async def index():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (current_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content)
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

452
demo/talk_to_gemini/index.html Normal file
View File

@@ -0,0 +1,452 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Gemini Voice Chat</title>
<style>
:root {
--color-accent: #6366f1;
--color-background: #0f172a;
--color-surface: #1e293b;
--color-text: #e2e8f0;
--boxSize: 8px;
--gutter: 4px;
}
body {
margin: 0;
padding: 0;
background-color: var(--color-background);
color: var(--color-text);
font-family: system-ui, -apple-system, sans-serif;
min-height: 100vh;
display: flex;
flex-direction: column;
align-items: center;
justify-content: center;
}
.container {
width: 90%;
max-width: 800px;
background-color: var(--color-surface);
padding: 2rem;
border-radius: 1rem;
box-shadow: 0 25px 50px -12px rgba(0, 0, 0, 0.25);
}
.wave-container {
position: relative;
display: flex;
min-height: 100px;
max-height: 128px;
justify-content: center;
align-items: center;
margin: 2rem 0;
}
.box-container {
display: flex;
justify-content: space-between;
height: 64px;
width: 100%;
}
.box {
height: 100%;
width: var(--boxSize);
background: var(--color-accent);
border-radius: 8px;
transition: transform 0.05s ease;
}
.controls {
display: grid;
gap: 1rem;
margin-bottom: 2rem;
}
.input-group {
display: flex;
flex-direction: column;
gap: 0.5rem;
}
label {
font-size: 0.875rem;
font-weight: 500;
}
input,
select {
padding: 0.75rem;
border-radius: 0.5rem;
border: 1px solid rgba(255, 255, 255, 0.1);
background-color: var(--color-background);
color: var(--color-text);
font-size: 1rem;
}
button {
padding: 1rem 2rem;
border-radius: 0.5rem;
border: none;
background-color: var(--color-accent);
color: white;
font-weight: 600;
cursor: pointer;
transition: all 0.2s ease;
}
button:hover {
opacity: 0.9;
transform: translateY(-1px);
}
.icon-with-spinner {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.spinner {
width: 20px;
height: 20px;
border: 2px solid white;
border-top-color: transparent;
border-radius: 50%;
animation: spin 1s linear infinite;
flex-shrink: 0;
}
@keyframes spin {
to {
transform: rotate(360deg);
}
}
.pulse-container {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.pulse-circle {
width: 20px;
height: 20px;
border-radius: 50%;
background-color: white;
opacity: 0.2;
flex-shrink: 0;
transform: translateX(-0%) scale(var(--audio-level, 1));
transition: transform 0.1s ease;
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div style="text-align: center">
<h1>Gemini Voice Chat</h1>
<p>Speak with Gemini using real-time audio streaming</p>
<p>
Get a Gemini API key
<a href="https://ai.google.dev/gemini-api/docs/api-key">here</a>
</p>
</div>
<div class="container">
<div class="controls">
<div class="input-group">
<label for="api-key">API Key</label>
<input type="password" id="api-key" placeholder="Enter your API key">
</div>
<div class="input-group">
<label for="voice">Voice</label>
<select id="voice">
<option value="Puck">Puck</option>
<option value="Charon">Charon</option>
<option value="Kore">Kore</option>
<option value="Fenrir">Fenrir</option>
<option value="Aoede">Aoede</option>
</select>
</div>
</div>
<div class="wave-container">
<div class="box-container">
<!-- Boxes will be dynamically added here -->
</div>
</div>
<button id="start-button">Start Recording</button>
</div>
<audio id="audio-output"></audio>
<script>
let peerConnection;
let audioContext;
let dataChannel;
let isRecording = false;
let webrtc_id;
const startButton = document.getElementById('start-button');
const apiKeyInput = document.getElementById('api-key');
const voiceSelect = document.getElementById('voice');
const audioOutput = document.getElementById('audio-output');
const boxContainer = document.querySelector('.box-container');
const numBars = 32;
for (let i = 0; i < numBars; i++) {
const box = document.createElement('div');
box.className = 'box';
boxContainer.appendChild(box);
}
function updateButtonState() {
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
startButton.innerHTML = `
<div class="icon-with-spinner">
<div class="spinner"></div>
<span>Connecting...</span>
</div>
`;
} else if (peerConnection && peerConnection.connectionState === 'connected') {
startButton.innerHTML = `
<div class="pulse-container">
<div class="pulse-circle"></div>
<span>Stop Recording</span>
</div>
`;
} else {
startButton.innerHTML = 'Start Recording';
}
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.className = 'toast error';
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
async function setupWebRTC() {
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
webrtc_id = Math.random().toString(36).substring(7);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
stream.getTracks().forEach(track => peerConnection.addTrack(track, stream));
// Update audio visualization setup
audioContext = new AudioContext();
analyser_input = audioContext.createAnalyser();
const source = audioContext.createMediaStreamSource(stream);
source.connect(analyser_input);
analyser_input.fftSize = 64;
dataArray_input = new Uint8Array(analyser_input.frequencyBinCount);
function updateAudioLevel() {
analyser_input.getByteFrequencyData(dataArray_input);
const average = Array.from(dataArray_input).reduce((a, b) => a + b, 0) / dataArray_input.length;
const audioLevel = average / 255;
const pulseCircle = document.querySelector('.pulse-circle');
if (pulseCircle) {
console.log("audioLevel", audioLevel);
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
}
animationId = requestAnimationFrame(updateAudioLevel);
}
updateAudioLevel();
// Add connection state change listener
peerConnection.addEventListener('connectionstatechange', () => {
console.log('connectionstatechange', peerConnection.connectionState);
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
updateButtonState();
});
// Handle incoming audio
peerConnection.addEventListener('track', (evt) => {
if (audioOutput && audioOutput.srcObject !== evt.streams[0]) {
audioOutput.srcObject = evt.streams[0];
audioOutput.play();
// Set up audio visualization on the output stream
audioContext = new AudioContext();
analyser = audioContext.createAnalyser();
const source = audioContext.createMediaStreamSource(evt.streams[0]);
source.connect(analyser);
analyser.fftSize = 2048;
dataArray = new Uint8Array(analyser.frequencyBinCount);
updateVisualization();
}
});
// Create data channel for messages
dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = (event) => {
const eventJson = JSON.parse(event.data);
if (eventJson.type === "error") {
showError(eventJson.message);
} else if (eventJson.type === "send_input") {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: webrtc_id,
api_key: apiKeyInput.value,
voice_name: voiceSelect.value
})
});
}
};
// Create and send offer
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id,
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
startButton.textContent = 'Start Recording';
return;
}
await peerConnection.setRemoteDescription(serverResponse);
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
startButton.textContent = 'Start Recording';
}
}
function updateVisualization() {
if (!analyser) return;
analyser.getByteFrequencyData(dataArray);
const bars = document.querySelectorAll('.box');
for (let i = 0; i < bars.length; i++) {
const barHeight = (dataArray[i] / 255) * 2;
bars[i].style.transform = `scaleY(${Math.max(0.1, barHeight)})`;
}
animationId = requestAnimationFrame(updateVisualization);
}
function stopWebRTC() {
if (peerConnection) {
peerConnection.close();
}
if (animationId) {
cancelAnimationFrame(animationId);
}
if (audioContext) {
audioContext.close();
}
updateButtonState();
}
startButton.addEventListener('click', () => {
if (!isRecording) {
setupWebRTC();
startButton.classList.add('recording');
} else {
stopWebRTC();
startButton.classList.remove('recording');
}
isRecording = !isRecording;
});
</script>
</body>
</html>

4
demo/talk_to_gemini/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc
python-dotenv
google-genai
twilio

15
demo/talk_to_openai/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to OpenAI
emoji: 🗣️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to OpenAI using their multimodal API
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|OPENAI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

15
demo/talk_to_openai/README_gradio.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to OpenAI (Gradio UI)
emoji: 🗣️
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Talk to OpenAI (Gradio UI)
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|OPENAI_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

141
demo/talk_to_openai/app.py Normal file
View File

@@ -0,0 +1,141 @@
import asyncio
import base64
import json
from pathlib import Path
import gradio as gr
import numpy as np
import openai
from dotenv import load_dotenv
from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import (
AdditionalOutputs,
AsyncStreamHandler,
Stream,
get_twilio_turn_credentials,
wait_for_item,
)
from gradio.utils import get_space
from openai.types.beta.realtime import ResponseAudioTranscriptDoneEvent
load_dotenv()
cur_dir = Path(__file__).parent
SAMPLE_RATE = 24000
class OpenAIHandler(AsyncStreamHandler):
def __init__(
self,
) -> None:
super().__init__(
expected_layout="mono",
output_sample_rate=SAMPLE_RATE,
output_frame_size=480,
input_sample_rate=SAMPLE_RATE,
)
self.connection = None
self.output_queue = asyncio.Queue()
def copy(self):
return OpenAIHandler()
async def start_up(
self,
):
"""Connect to realtime API. Run forever in separate thread to keep connection open."""
self.client = openai.AsyncOpenAI()
async with self.client.beta.realtime.connect(
model="gpt-4o-mini-realtime-preview-2024-12-17"
) as conn:
await conn.session.update(
session={"turn_detection": {"type": "server_vad"}}
)
self.connection = conn
async for event in self.connection:
if event.type == "response.audio_transcript.done":
await self.output_queue.put(AdditionalOutputs(event))
if event.type == "response.audio.delta":
await self.output_queue.put(
(
self.output_sample_rate,
np.frombuffer(
base64.b64decode(event.delta), dtype=np.int16
).reshape(1, -1),
),
)
async def receive(self, frame: tuple[int, np.ndarray]) -> None:
if not self.connection:
return
_, array = frame
array = array.squeeze()
audio_message = base64.b64encode(array.tobytes()).decode("utf-8")
await self.connection.input_audio_buffer.append(audio=audio_message) # type: ignore
async def emit(self) -> tuple[int, np.ndarray] | AdditionalOutputs | None:
return await wait_for_item(self.output_queue)
async def shutdown(self) -> None:
if self.connection:
await self.connection.close()
self.connection = None
def update_chatbot(chatbot: list[dict], response: ResponseAudioTranscriptDoneEvent):
chatbot.append({"role": "assistant", "content": response.transcript})
return chatbot
chatbot = gr.Chatbot(type="messages")
latest_message = gr.Textbox(type="text", visible=False)
stream = Stream(
OpenAIHandler(),
mode="send-receive",
modality="audio",
additional_inputs=[chatbot],
additional_outputs=[chatbot],
additional_outputs_handler=update_chatbot,
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=5 if get_space() else None,
time_limit=90 if get_space() else None,
)
app = FastAPI()
stream.mount(app)
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (cur_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content)
@app.get("/outputs")
def _(webrtc_id: str):
async def output_stream():
import json
async for output in stream.output_stream(webrtc_id):
s = json.dumps({"role": "assistant", "content": output.args[0].transcript})
yield f"event: output\ndata: {s}\n\n"
return StreamingResponse(output_stream(), media_type="text/event-stream")
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

404
demo/talk_to_openai/index.html Normal file
View File

@@ -0,0 +1,404 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>OpenAI Real-Time Chat</title>
<style>
body {
font-family: "SF Pro Display", -apple-system, BlinkMacSystemFont, sans-serif;
background-color: #0a0a0a;
color: #ffffff;
margin: 0;
padding: 20px;
height: 100vh;
box-sizing: border-box;
}
.container {
max-width: 800px;
margin: 0 auto;
height: calc(100% - 100px);
}
.logo {
text-align: center;
margin-bottom: 40px;
}
.chat-container {
border: 1px solid #333;
padding: 20px;
height: 90%;
box-sizing: border-box;
display: flex;
flex-direction: column;
}
.chat-messages {
flex-grow: 1;
overflow-y: auto;
margin-bottom: 20px;
padding: 10px;
}
.message {
margin-bottom: 20px;
padding: 12px;
border-radius: 4px;
font-size: 16px;
line-height: 1.5;
}
.message.user {
background-color: #1a1a1a;
margin-left: 20%;
}
.message.assistant {
background-color: #262626;
margin-right: 20%;
}
.controls {
text-align: center;
margin-top: 20px;
}
button {
background-color: transparent;
color: #ffffff;
border: 1px solid #ffffff;
padding: 12px 24px;
font-family: inherit;
font-size: 16px;
cursor: pointer;
transition: all 0.3s;
text-transform: uppercase;
letter-spacing: 1px;
}
button:hover {
border-width: 2px;
transform: scale(1.02);
box-shadow: 0 0 10px rgba(255, 255, 255, 0.2);
}
#audio-output {
display: none;
}
.icon-with-spinner {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.spinner {
width: 20px;
height: 20px;
border: 2px solid #ffffff;
border-top-color: transparent;
border-radius: 50%;
animation: spin 1s linear infinite;
flex-shrink: 0;
}
@keyframes spin {
to {
transform: rotate(360deg);
}
}
.pulse-container {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.pulse-circle {
width: 20px;
height: 20px;
border-radius: 50%;
background-color: #ffffff;
opacity: 0.2;
flex-shrink: 0;
transform: translateX(-0%) scale(var(--audio-level, 1));
transition: transform 0.1s ease;
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div class="container">
<div class="logo">
<h1>OpenAI Real-Time Chat</h1>
</div>
<div class="chat-container">
<div class="chat-messages" id="chat-messages"></div>
</div>
<div class="controls">
<button id="start-button">Start Conversation</button>
</div>
</div>
<audio id="audio-output"></audio>
<script>
let peerConnection;
let webrtc_id;
const audioOutput = document.getElementById('audio-output');
const startButton = document.getElementById('start-button');
const chatMessages = document.getElementById('chat-messages');
let audioLevel = 0;
let animationFrame;
let audioContext, analyser, audioSource;
function updateButtonState() {
const button = document.getElementById('start-button');
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
button.innerHTML = `
<div class="icon-with-spinner">
<div class="spinner"></div>
<span>Connecting...</span>
</div>
`;
} else if (peerConnection && peerConnection.connectionState === 'connected') {
button.innerHTML = `
<div class="pulse-container">
<div class="pulse-circle"></div>
<span>Stop Conversation</span>
</div>
`;
} else {
button.innerHTML = 'Start Conversation';
}
}
function setupAudioVisualization(stream) {
audioContext = new (window.AudioContext || window.webkitAudioContext)();
analyser = audioContext.createAnalyser();
audioSource = audioContext.createMediaStreamSource(stream);
audioSource.connect(analyser);
analyser.fftSize = 64;
const dataArray = new Uint8Array(analyser.frequencyBinCount);
function updateAudioLevel() {
analyser.getByteFrequencyData(dataArray);
const average = Array.from(dataArray).reduce((a, b) => a + b, 0) / dataArray.length;
audioLevel = average / 255;
// Update CSS variable instead of rebuilding the button
const pulseCircle = document.querySelector('.pulse-circle');
if (pulseCircle) {
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
}
animationFrame = requestAnimationFrame(updateAudioLevel);
}
updateAudioLevel();
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
async function setupWebRTC() {
isConnecting = true;
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({
audio: true
});
setupAudioVisualization(stream);
stream.getTracks().forEach(track => {
peerConnection.addTrack(track, stream);
});
peerConnection.addEventListener('track', (evt) => {
if (audioOutput.srcObject !== evt.streams[0]) {
audioOutput.srcObject = evt.streams[0];
audioOutput.play();
}
});
const dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = (event) => {
const eventJson = JSON.parse(event.data);
if (eventJson.type === "error") {
showError(eventJson.message);
}
};
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
peerConnection.addEventListener('connectionstatechange', () => {
console.log('connectionstatechange', peerConnection.connectionState);
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
updateButtonState();
});
webrtc_id = Math.random().toString(36).substring(7);
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
return;
}
await peerConnection.setRemoteDescription(serverResponse);
const eventSource = new EventSource('/outputs?webrtc_id=' + webrtc_id);
eventSource.addEventListener("output", (event) => {
const eventJson = JSON.parse(event.data);
addMessage("assistant", eventJson.content);
});
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
}
}
function addMessage(role, content) {
const messageDiv = document.createElement('div');
messageDiv.classList.add('message', role);
messageDiv.textContent = content;
chatMessages.appendChild(messageDiv);
chatMessages.scrollTop = chatMessages.scrollHeight;
}
function stop() {
if (animationFrame) {
cancelAnimationFrame(animationFrame);
}
if (audioContext) {
audioContext.close();
audioContext = null;
analyser = null;
audioSource = null;
}
if (peerConnection) {
if (peerConnection.getTransceivers) {
peerConnection.getTransceivers().forEach(transceiver => {
if (transceiver.stop) {
transceiver.stop();
}
});
}
if (peerConnection.getSenders) {
peerConnection.getSenders().forEach(sender => {
if (sender.track && sender.track.stop) sender.track.stop();
});
}
console.log('closing');
peerConnection.close();
}
updateButtonState();
audioLevel = 0;
}
startButton.addEventListener('click', () => {
console.log('clicked');
console.log(peerConnection, peerConnection?.connectionState);
if (!peerConnection || peerConnection.connectionState !== 'connected') {
setupWebRTC();
} else {
console.log('stopping');
stop();
}
});
</script>
</body>
</html>

4
demo/talk_to_openai/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc[vad]
openai
twilio
python-dotenv

15
demo/talk_to_sambanova/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to Sambanova
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Llama 3.2 - SambaNova API
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|SAMBANOVA_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

15
demo/talk_to_sambanova/README_gradio.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Talk to Sambanova (Gradio)
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Llama 3.2 - SambaNova API (Gradio)
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|SAMBANOVA_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

144
demo/talk_to_sambanova/app.py Normal file
View File

@@ -0,0 +1,144 @@
import base64
import json
import os
from pathlib import Path
import gradio as gr
import huggingface_hub
import numpy as np
from dotenv import load_dotenv
from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import (
AdditionalOutputs,
ReplyOnPause,
Stream,
get_stt_model,
get_twilio_turn_credentials,
)
from gradio.utils import get_space
from pydantic import BaseModel
load_dotenv()
curr_dir = Path(__file__).parent
client = huggingface_hub.InferenceClient(
api_key=os.environ.get("SAMBANOVA_API_KEY"),
provider="sambanova",
)
stt_model = get_stt_model()
def response(
audio: tuple[int, np.ndarray],
gradio_chatbot: list[dict] | None = None,
conversation_state: list[dict] | None = None,
):
gradio_chatbot = gradio_chatbot or []
conversation_state = conversation_state or []
print("chatbot", gradio_chatbot)
text = stt_model.stt(audio)
sample_rate, array = audio
gradio_chatbot.append(
{"role": "user", "content": gr.Audio((sample_rate, array.squeeze()))}
)
yield AdditionalOutputs(gradio_chatbot, conversation_state)
conversation_state.append({"role": "user", "content": text})
request = client.chat.completions.create(
model="meta-llama/Llama-3.2-3B-Instruct",
messages=conversation_state, # type: ignore
temperature=0.1,
top_p=0.1,
)
response = {"role": "assistant", "content": request.choices[0].message.content}
conversation_state.append(response)
gradio_chatbot.append(response)
yield AdditionalOutputs(gradio_chatbot, conversation_state)
chatbot = gr.Chatbot(type="messages", value=[])
state = gr.State(value=[])
stream = Stream(
ReplyOnPause(
response, # type: ignore
input_sample_rate=16000,
),
mode="send",
modality="audio",
additional_inputs=[chatbot, state],
additional_outputs=[chatbot, state],
additional_outputs_handler=lambda *a: (a[2], a[3]),
concurrency_limit=20 if get_space() else None,
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
)
app = FastAPI()
stream.mount(app)
class Message(BaseModel):
role: str
content: str
class InputData(BaseModel):
webrtc_id: str
chatbot: list[Message]
state: list[Message]
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (curr_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content)
@app.post("/input_hook")
async def _(data: InputData):
body = data.model_dump()
stream.set_input(data.webrtc_id, body["chatbot"], body["state"])
def audio_to_base64(file_path):
audio_format = "wav"
with open(file_path, "rb") as audio_file:
encoded_audio = base64.b64encode(audio_file.read()).decode("utf-8")
return f"data:audio/{audio_format};base64,{encoded_audio}"
@app.get("/outputs")
async def _(webrtc_id: str):
async def output_stream():
async for output in stream.output_stream(webrtc_id):
chatbot = output.args[0]
state = output.args[1]
data = {
"message": state[-1],
"audio": audio_to_base64(chatbot[-1]["content"].value["path"])
if chatbot[-1]["role"] == "user"
else None,
}
yield f"event: output\ndata: {json.dumps(data)}\n\n"
return StreamingResponse(output_stream(), media_type="text/event-stream")
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
raise ValueError("Phone mode not supported")
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

487
demo/talk_to_sambanova/index.html Normal file
View File

@@ -0,0 +1,487 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Talk to Sambanova</title>
<style>
body {
font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, sans-serif;
background-color: #f8f9fa;
color: #1a1a1a;
margin: 0;
padding: 20px;
height: 100vh;
box-sizing: border-box;
}
.container {
max-width: 800px;
margin: 0 auto;
height: 80%;
}
.logo {
text-align: center;
margin-bottom: 40px;
}
.chat-container {
background: white;
border-radius: 8px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
padding: 20px;
height: 90%;
box-sizing: border-box;
display: flex;
flex-direction: column;
}
.chat-messages {
flex-grow: 1;
overflow-y: auto;
margin-bottom: 20px;
padding: 10px;
}
.message {
margin-bottom: 20px;
padding: 12px;
border-radius: 8px;
font-size: 14px;
line-height: 1.5;
}
.message.user {
background-color: #e9ecef;
margin-left: 20%;
}
.message.assistant {
background-color: #f1f3f5;
margin-right: 20%;
}
.controls {
text-align: center;
margin-top: 20px;
}
button {
background-color: #0066cc;
color: white;
border: none;
padding: 12px 24px;
font-family: inherit;
font-size: 14px;
cursor: pointer;
transition: all 0.3s;
border-radius: 4px;
font-weight: 500;
}
button:hover {
background-color: #0052a3;
}
#audio-output {
display: none;
}
.icon-with-spinner {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.spinner {
width: 20px;
height: 20px;
border: 2px solid #ffffff;
border-top-color: transparent;
border-radius: 50%;
animation: spin 1s linear infinite;
flex-shrink: 0;
}
@keyframes spin {
to {
transform: rotate(360deg);
}
}
.pulse-container {
display: flex;
align-items: center;
justify-content: center;
gap: 12px;
min-width: 180px;
}
.pulse-circle {
width: 20px;
height: 20px;
border-radius: 50%;
background-color: #ffffff;
opacity: 0.2;
flex-shrink: 0;
transform: translateX(-0%) scale(var(--audio-level, 1));
transition: transform 0.1s ease;
}
/* Add styles for typing indicator */
.typing-indicator {
padding: 8px;
background-color: #f1f3f5;
border-radius: 8px;
margin-bottom: 10px;
display: none;
}
.dots {
display: inline-flex;
gap: 4px;
}
.dot {
width: 8px;
height: 8px;
background-color: #0066cc;
border-radius: 50%;
animation: pulse 1.5s infinite;
opacity: 0.5;
}
.dot:nth-child(2) {
animation-delay: 0.5s;
}
.dot:nth-child(3) {
animation-delay: 1s;
}
@keyframes pulse {
0%,
100% {
opacity: 0.5;
transform: scale(1);
}
50% {
opacity: 1;
transform: scale(1.2);
}
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<!-- Add toast element after body opening tag -->
<div id="error-toast" class="toast"></div>
<div class="container">
<div class="logo">
<h1>Talk to Sambanova 🗣️</h1>
<h2 style="font-size: 1.2em; color: #666; margin-top: 10px;">Speak to Llama 3.2 powered by Sambanova API
</h2>
</div>
<div class="chat-container">
<div class="chat-messages" id="chat-messages"></div>
<div class="typing-indicator" id="typing-indicator">
<div class="dots">
<div class="dot"></div>
<div class="dot"></div>
<div class="dot"></div>
</div>
</div>
</div>
<div class="controls">
<button id="start-button">Start Conversation</button>
</div>
</div>
<audio id="audio-output"></audio>
<script>
let peerConnection;
let webrtc_id;
const startButton = document.getElementById('start-button');
const chatMessages = document.getElementById('chat-messages');
let audioLevel = 0;
let animationFrame;
let audioContext, analyser, audioSource;
let messages = [];
let eventSource;
function updateButtonState() {
const button = document.getElementById('start-button');
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
button.innerHTML = `
<div class="icon-with-spinner">
<div class="spinner"></div>
<span>Connecting...</span>
</div>
`;
} else if (peerConnection && peerConnection.connectionState === 'connected') {
button.innerHTML = `
<div class="pulse-container">
<div class="pulse-circle"></div>
<span>Stop Conversation</span>
</div>
`;
} else {
button.innerHTML = 'Start Conversation';
}
}
function setupAudioVisualization(stream) {
audioContext = new (window.AudioContext || window.webkitAudioContext)();
analyser = audioContext.createAnalyser();
audioSource = audioContext.createMediaStreamSource(stream);
audioSource.connect(analyser);
analyser.fftSize = 64;
const dataArray = new Uint8Array(analyser.frequencyBinCount);
function updateAudioLevel() {
analyser.getByteFrequencyData(dataArray);
const average = Array.from(dataArray).reduce((a, b) => a + b, 0) / dataArray.length;
audioLevel = average / 255;
const pulseCircle = document.querySelector('.pulse-circle');
if (pulseCircle) {
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
}
animationFrame = requestAnimationFrame(updateAudioLevel);
}
updateAudioLevel();
}
function showError(message) {
const toast = document.getElementById('error-toast');
toast.textContent = message;
toast.className = 'toast error';
toast.style.display = 'block';
// Hide toast after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}
function handleMessage(event) {
const eventJson = JSON.parse(event.data);
const typingIndicator = document.getElementById('typing-indicator');
if (eventJson.type === "error") {
showError(eventJson.message);
} else if (eventJson.type === "send_input") {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: webrtc_id,
chatbot: messages,
state: messages
})
});
} else if (eventJson.type === "log") {
if (eventJson.data === "pause_detected") {
typingIndicator.style.display = 'block';
chatMessages.scrollTop = chatMessages.scrollHeight;
} else if (eventJson.data === "response_starting") {
typingIndicator.style.display = 'none';
}
}
}
async function setupWebRTC() {
const config = __RTC_CONFIGURATION__;
peerConnection = new RTCPeerConnection(config);
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({
audio: true
});
setupAudioVisualization(stream);
stream.getTracks().forEach(track => {
peerConnection.addTrack(track, stream);
});
const dataChannel = peerConnection.createDataChannel('text');
dataChannel.onmessage = handleMessage;
const offer = await peerConnection.createOffer();
await peerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (peerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (peerConnection.iceGatheringState === "complete") {
peerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
peerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
peerConnection.addEventListener('connectionstatechange', () => {
console.log('connectionstatechange', peerConnection.connectionState);
if (peerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
updateButtonState();
});
webrtc_id = Math.random().toString(36).substring(7);
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: peerConnection.localDescription.sdp,
type: peerConnection.localDescription.type,
webrtc_id: webrtc_id
})
});
const serverResponse = await response.json();
if (serverResponse.status === 'failed') {
showError(serverResponse.meta.error === 'concurrency_limit_reached'
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
: serverResponse.meta.error);
stop();
return;
}
await peerConnection.setRemoteDescription(serverResponse);
eventSource = new EventSource('/outputs?webrtc_id=' + webrtc_id);
eventSource.addEventListener("output", (event) => {
const eventJson = JSON.parse(event.data);
console.log(eventJson);
messages.push(eventJson.message);
addMessage(eventJson.message.role, eventJson.audio ?? eventJson.message.content);
});
} catch (err) {
clearTimeout(timeoutId);
console.error('Error setting up WebRTC:', err);
showError('Failed to establish connection. Please try again.');
stop();
}
}
function addMessage(role, content) {
const messageDiv = document.createElement('div');
messageDiv.classList.add('message', role);
if (role === 'user') {
// Create audio element for user messages
const audio = document.createElement('audio');
audio.controls = true;
audio.src = content;
messageDiv.appendChild(audio);
} else {
// Text content for assistant messages
messageDiv.textContent = content;
}
chatMessages.appendChild(messageDiv);
chatMessages.scrollTop = chatMessages.scrollHeight;
}
function stop() {
if (eventSource) {
eventSource.close();
eventSource = null;
}
if (animationFrame) {
cancelAnimationFrame(animationFrame);
}
if (audioContext) {
audioContext.close();
audioContext = null;
analyser = null;
audioSource = null;
}
if (peerConnection) {
if (peerConnection.getTransceivers) {
peerConnection.getTransceivers().forEach(transceiver => {
if (transceiver.stop) {
transceiver.stop();
}
});
}
if (peerConnection.getSenders) {
peerConnection.getSenders().forEach(sender => {
if (sender.track && sender.track.stop) sender.track.stop();
});
}
peerConnection.close();
}
updateButtonState();
audioLevel = 0;
}
startButton.addEventListener('click', () => {
if (!peerConnection || peerConnection.connectionState !== 'connected') {
setupWebRTC();
} else {
stop();
}
});
</script>
</body>
</html>

4
demo/talk_to_sambanova/requirements.txt Normal file
View File

@@ -0,0 +1,4 @@
fastrtc[vad, stt]
python-dotenv
huggingface_hub>=0.29.0
twilio

98
demo/talk_to_smolagents/README.md Normal file
View File

@@ -0,0 +1,98 @@
---
title: Talk to Smolagents
emoji: 💻
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: FastRTC Voice Agent with smolagents
tags: [webrtc, websocket, gradio, secret|HF_TOKEN, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN]
---
# Voice LLM Agent with Image Generation
A voice-enabled AI assistant powered by FastRTC that can:
1. Stream audio in real-time using WebRTC
2. Listen and respond with natural pauses in conversation
3. Generate images based on your requests
4. Maintain conversation context across exchanges
This app combines the real-time communication capabilities of FastRTC with the powerful agent framework of smolagents.
## Key Features
- **Real-time Streaming**: Uses FastRTC's WebRTC-based audio streaming
- **Voice Activation**: Automatic detection of speech pauses to trigger responses
- **Multi-modal Interaction**: Combines voice and image generation in a single interface
## Setup
1. Install Python 3.9+ and create a virtual environment:
```bash
python -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
```
2. Install dependencies:
```bash
pip install -r requirements.txt
```
3. Create a `.env` file with the following:
```
HF_TOKEN=your_huggingface_api_key
MODE=UI # Use 'UI' for Gradio interface, leave blank for HTML interface
```
## Running the App
### With Gradio UI (Recommended)
```bash
MODE=UI python app.py
```
This launches a Gradio UI at http://localhost:7860 with:
- FastRTC's built-in streaming audio components
- A chat interface showing the conversation
- An image display panel for generated images
## How to Use
1. Click the microphone button to start streaming your voice.
2. Speak naturally - the app will automatically detect when you pause.
3. Ask the agent to generate an image, for example:
- "Create an image of a magical forest with glowing mushrooms."
- "Generate a picture of a futuristic city with flying cars."
4. View the generated image and hear the agent's response.
## Technical Architecture
### FastRTC Components
- **Stream**: Core component that handles WebRTC connections and audio streaming
- **ReplyOnPause**: Detects when the user stops speaking to trigger a response
- **get_stt_model/get_tts_model**: Provides optimized speech-to-text and text-to-speech models
### smolagents Components
- **CodeAgent**: Intelligent agent that can use tools based on natural language inputs
- **Tool.from_space**: Integration with Hugging Face Spaces for image generation
- **HfApiModel**: Connection to powerful language models for understanding requests
### Integration Flow
1. FastRTC streams and processes audio input in real-time
2. Speech is converted to text and passed to the smolagents CodeAgent
3. The agent processes the request and calls tools when needed
4. Responses and generated images are streamed back through FastRTC
5. The UI updates to show both text responses and generated images
## Advanced Features
- Conversation history is maintained across exchanges
- Error handling ensures the app continues working even if agent processing fails
- The application leverages FastRTC's streaming capabilities for efficient audio transmission

99
demo/talk_to_smolagents/app.py Normal file
View File

@@ -0,0 +1,99 @@
from pathlib import Path
from typing import Dict, List
from dotenv import load_dotenv
from fastrtc import (
ReplyOnPause,
Stream,
get_stt_model,
get_tts_model,
get_twilio_turn_credentials,
)
from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel
# Load environment variables
load_dotenv()
# Initialize file paths
curr_dir = Path(__file__).parent
# Initialize models
stt_model = get_stt_model()
tts_model = get_tts_model()
# Conversation state to maintain history
conversation_state: List[Dict[str, str]] = []
# System prompt for agent
system_prompt = """You are a helpful assistant that can helps with finding places to
workremotely from. You should specifically check against reviews and ratings of the
place. You should use this criteria to find the best place to work from:
- Price
- Reviews
- Ratings
- Location
- WIFI
Only return the name, address of the place, and a short description of the place.
Always search for real places.
Only return real places, not fake ones.
If you receive anything other than a location, you should ask for a location.
<example>
User: I am in Paris, France. Can you find me a place to work from?
Assistant: I found a place called "Le Café de la Paix" at 123 Rue de la Paix,
Paris, France. It has good reviews and is in a great location.
</example>
<example>
User: I am in London, UK. Can you find me a place to work from?
Assistant: I found a place called "The London Coffee Company".
</example>
<example>
User: How many people are in the room?
Assistant: I only respond to requests about finding places to work from.
</example>
"""
model = HfApiModel(provider="together", model="Qwen/Qwen2.5-Coder-32B-Instruct")
agent = CodeAgent(
tools=[
DuckDuckGoSearchTool(),
],
model=model,
max_steps=10,
verbosity_level=2,
description="Search the web for cafes to work from.",
)
def process_response(audio):
"""Process audio input and generate LLM response with TTS"""
# Convert speech to text using STT model
text = stt_model.stt(audio)
if not text.strip():
return
input_text = f"{system_prompt}\n\n{text}"
# Get response from agent
response_content = agent.run(input_text)
# Convert response to audio using TTS model
for audio_chunk in tts_model.stream_tts_sync(response_content or ""):
# Yield the audio chunk
yield audio_chunk
stream = Stream(
handler=ReplyOnPause(process_response, input_sample_rate=16000),
modality="audio",
mode="send-receive",
ui_args={
"pulse_color": "rgb(255, 255, 255)",
"icon_button_color": "rgb(255, 255, 255)",
"title": "🧑‍💻The Coworking Agent",
},
rtc_configuration=get_twilio_turn_credentials(),
)
if __name__ == "__main__":
stream.ui.launch(server_port=7860)

136
demo/talk_to_smolagents/requirements.txt Normal file
View File

@@ -0,0 +1,136 @@
# This file was autogenerated by uv via the following command:
# uv export --format requirements-txt --no-hashes
aiofiles==23.2.1
aiohappyeyeballs==2.4.6
aiohttp==3.11.13
aiohttp-retry==2.9.1
aioice==0.9.0
aiortc==1.10.1
aiosignal==1.3.2
annotated-types==0.7.0
anyio==4.8.0
async-timeout==5.0.1 ; python_full_version < '3.11'
attrs==25.1.0
audioop-lts==0.2.1 ; python_full_version >= '3.13'
audioread==3.0.1
av==13.1.0
babel==2.17.0
beautifulsoup4==4.13.3
certifi==2025.1.31
cffi==1.17.1
charset-normalizer==3.4.1
click==8.1.8
colorama==0.4.6
coloredlogs==15.0.1
colorlog==6.9.0
cryptography==44.0.1
csvw==3.5.1
decorator==5.2.1
dlinfo==2.0.0
dnspython==2.7.0
duckduckgo-search==7.5.0
espeakng-loader==0.2.4
exceptiongroup==1.2.2 ; python_full_version < '3.11'
fastapi==0.115.8
fastrtc==0.0.8.post1
fastrtc-moonshine-onnx==20241016
ffmpy==0.5.0
filelock==3.17.0
flatbuffers==25.2.10
frozenlist==1.5.0
fsspec==2025.2.0
google-crc32c==1.6.0
gradio==5.19.0
gradio-client==1.7.2
h11==0.14.0
httpcore==1.0.7
httpx==0.28.1
huggingface-hub==0.29.1
humanfriendly==10.0
idna==3.10
ifaddr==0.2.0
isodate==0.7.2
jinja2==3.1.5
joblib==1.4.2
jsonschema==4.23.0
jsonschema-specifications==2024.10.1
kokoro-onnx==0.4.3
language-tags==1.2.0
lazy-loader==0.4
librosa==0.10.2.post1
llvmlite==0.44.0
lxml==5.3.1
markdown-it-py==3.0.0
markdownify==1.0.0
markupsafe==2.1.5
mdurl==0.1.2
mpmath==1.3.0
msgpack==1.1.0
multidict==6.1.0
numba==0.61.0
numpy==2.1.3
onnxruntime==1.20.1
orjson==3.10.15
packaging==24.2
pandas==2.2.3
phonemizer-fork==3.3.1
pillow==11.1.0
platformdirs==4.3.6
pooch==1.8.2
primp==0.14.0
propcache==0.3.0
protobuf==5.29.3
pycparser==2.22
pydantic==2.10.6
pydantic-core==2.27.2
pydub==0.25.1
pyee==12.1.1
pygments==2.19.1
pyjwt==2.10.1
pylibsrtp==0.11.0
pyopenssl==25.0.0
pyparsing==3.2.1
pyreadline3==3.5.4 ; sys_platform == 'win32'
python-dateutil==2.9.0.post0
python-dotenv==1.0.1
python-multipart==0.0.20
pytz==2025.1
pyyaml==6.0.2
rdflib==7.1.3
referencing==0.36.2
regex==2024.11.6
requests==2.32.3
rfc3986==1.5.0
rich==13.9.4
rpds-py==0.23.1
ruff==0.9.7 ; sys_platform != 'emscripten'
safehttpx==0.1.6
scikit-learn==1.6.1
scipy==1.15.2
segments==2.3.0
semantic-version==2.10.0
shellingham==1.5.4 ; sys_platform != 'emscripten'
six==1.17.0
smolagents==1.9.2
sniffio==1.3.1
soundfile==0.13.1
soupsieve==2.6
soxr==0.5.0.post1
standard-aifc==3.13.0 ; python_full_version >= '3.13'
standard-chunk==3.13.0 ; python_full_version >= '3.13'
standard-sunau==3.13.0 ; python_full_version >= '3.13'
starlette==0.45.3
sympy==1.13.3
threadpoolctl==3.5.0
tokenizers==0.21.0
tomlkit==0.13.2
tqdm==4.67.1
twilio==9.4.6
typer==0.15.1 ; sys_platform != 'emscripten'
typing-extensions==4.12.2
tzdata==2025.1
uritemplate==4.1.1
urllib3==2.3.0
uvicorn==0.34.0 ; sys_platform != 'emscripten'
websockets==15.0
yarl==1.18.3

BIN
demo/video.mp4 Normal file
View File

Binary file not shown.

19
demo/voice_text_editor/README.md Normal file
View File

@@ -0,0 +1,19 @@
---
title: Voice Text Editor
emoji: 📝
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Edit text documents with your voice!
tags: [webrtc, websocket, gradio, secret|HF_TOKEN, secret|SAMBANOVA_API_KEY]
---
# Voice Text Editor
Edit text documents with your voice!

113
demo/voice_text_editor/app.py Normal file
View File

@@ -0,0 +1,113 @@
import os
import gradio as gr
from dotenv import load_dotenv
from fastrtc import AdditionalOutputs, ReplyOnPause, Stream, get_stt_model
from openai import OpenAI
load_dotenv()
sambanova_client = OpenAI(
api_key=os.getenv("SAMBANOVA_API_KEY"), base_url="https://api.sambanova.ai/v1"
)
stt_model = get_stt_model()
SYSTEM_PROMPT = """You are an intelligent voice-activated text editor assistant. Your purpose is to help users create and modify text documents through voice commands.
For each interaction:
1. You will receive the current state of a text document and a voice input from the user.
2. Determine if the input is:
a) A command to modify the document (e.g., "delete the last line", "capitalize that")
b) Content to be added to the document (e.g., "buy 12 eggs at the store")
c) A modification to existing content (e.g., "actually make that 24" to change "12" to "24")
3. Return ONLY the new document state after the changes have been applied.
Example:
CURRENT DOCUMENT:
Meeting notes:
- Buy GPUs
- Meet with Joe
USER INPUT: Make that 100 GPUS
NEW DOCUMENT STATE:
Meeting notes:
- Buy 100 GPUs
- Meet with Joe
Example 2:
CURRENT DOCUMENT:
Project Proposal
USER INPUT: Make that a header
NEW DOCUMENT STATE:
# Project Proposal
When handling commands:
- Apply the requested changes precisely to the document
- Support operations like adding, deleting, modifying, and moving text
- Understand contextual references like "that", "the last line", "the second paragraph"
When handling content additions:
- Add the new text at the appropriate location (usually at the end or cursor position)
- Format it appropriately based on the document context
- If the user says to "add" or "insert" do not remove text that was already in the document.
When handling content modifications:
- Identify what part of the document the user is referring to
- Apply the requested change while preserving the rest of the content
- Be smart about contextual references (e.g., "make that 24" should know to replace a number)
NEVER include any text in the new document state that is not part of the user's input.
NEVER include the phrase "CURRENT DOCUMENT" in the new document state.
NEVER reword the user's input unless you are explicitly asked to do so.
"""
def edit(audio, current_document: str):
prompt = stt_model.stt(audio)
print(f"Prompt: {prompt}")
response = sambanova_client.chat.completions.create(
model="Meta-Llama-3.3-70B-Instruct",
messages=[
{"role": "system", "content": SYSTEM_PROMPT},
{
"role": "user",
"content": f"CURRENT DOCUMENT:\n\n{current_document}\n\nUSER INPUT: {prompt}",
},
],
max_tokens=200,
)
doc = response.choices[0].message.content
yield AdditionalOutputs(doc)
doc = gr.Textbox(value="", label="Current Document")
stream = Stream(
ReplyOnPause(edit),
modality="audio",
mode="send",
additional_inputs=[doc],
additional_outputs=[doc],
additional_outputs_handler=lambda prev, current: current,
ui_args={"title": "Voice Text Editor with FastRTC 🗣️"},
)
if __name__ == "__main__":
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
stream.ui.launch(server_port=7860)

126
demo/voice_text_editor_local/app.py Normal file
View File

@@ -0,0 +1,126 @@
import os
import gradio as gr
import requests
from dotenv import load_dotenv
from fastrtc import AdditionalOutputs, ReplyOnPause, Stream, get_stt_model
load_dotenv()
stt_model = get_stt_model()
SYSTEM_PROMPT = """You are an intelligent voice-activated text editor assistant. Your purpose is to help users create and modify text documents through voice commands.
For each interaction:
1. You will receive the current state of a text document and a voice input from the user.
2. Determine if the input is:
a) A command to modify the document (e.g., "delete the last line", "capitalize that")
b) Content to be added to the document (e.g., "buy 12 eggs at the store")
c) A modification to existing content (e.g., "actually make that 24" to change "12" to "24")
3. Return ONLY the new document state after the changes have been applied.
Example:
CURRENT DOCUMENT:
Meeting notes:
- Buy GPUs
- Meet with Joe
USER INPUT: Make that 100 GPUS
NEW DOCUMENT STATE:
Meeting notes:
- Buy 100 GPUs
- Meet with Joe
Example 2:
CURRENT DOCUMENT:
Project Proposal
USER INPUT: Make that a header
NEW DOCUMENT STATE:
# Project Proposal
When handling commands:
- Apply the requested changes precisely to the document
- Support operations like adding, deleting, modifying, and moving text
- Understand contextual references like "that", "the last line", "the second paragraph"
When handling content additions:
- Add the new text at the appropriate location (usually at the end or cursor position)
- Format it appropriately based on the document context
- If the user says to "add" or "insert" do not remove text that was already in the document.
When handling content modifications:
- Identify what part of the document the user is referring to
- Apply the requested change while preserving the rest of the content
- Be smart about contextual references (e.g., "make that 24" should know to replace a number)
NEVER include any text in the new document state that is not part of the user's input.
NEVER include the phrase "CURRENT DOCUMENT" in the new document state.
NEVER reword the user's input unless you are explicitly asked to do so.
"""
def edit(audio, current_document: str):
prompt = stt_model.stt(audio)
print(f"Prompt: {prompt}")
# Construct the prompt for ollama
full_prompt = (
f"{SYSTEM_PROMPT}\n\n"
f"User: CURRENT DOCUMENT:\n\n{current_document}\n\nUSER INPUT: {prompt}\n\n"
f"Assistant:"
)
try:
# Send request to ollama's API
response = requests.post(
"http://localhost:11434/api/generate",
json={
"model": "qwen2.5",
"prompt": full_prompt,
"stream": False,
"max_tokens": 200,
},
)
response.raise_for_status() # Raise an exception for bad status codes
# Parse the response
doc = response.json()["response"]
# Clean up the response to remove "Assistant:" and any extra whitespace
doc = doc.strip().lstrip("Assistant:").strip()
yield AdditionalOutputs(doc)
except requests.RequestException as e:
# Handle API errors gracefully
error_message = "Error: Could not connect to ollama. Please ensure it's running and qwen2.5 is loaded."
print(f"API Error: {e}")
yield AdditionalOutputs(error_message)
doc = gr.Textbox(value="", label="Current Document")
stream = Stream(
ReplyOnPause(edit),
modality="audio",
mode="send",
additional_inputs=[doc],
additional_outputs=[doc],
additional_outputs_handler=lambda prev, current: current,
ui_args={"title": "Voice Text Editor with FastRTC 🗣️"},
)
if __name__ == "__main__":
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860)
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
stream.ui.launch(server_port=7860)

15
demo/webrtc_vs_websocket/README.md Normal file
View File

@@ -0,0 +1,15 @@
---
title: Webrtc Vs Websocket
emoji: 🧪
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.16.0
app_file: app.py
pinned: false
license: mit
short_description: Compare Round Trip Times between WebRTC and Websockets
tags: [webrtc, websocket, gradio, secret|TWILIO_ACCOUNT_SID, secret|TWILIO_AUTH_TOKEN, secret|ELEVENLABS_API_KEY, secret|GROQ_API_KEY, secret|ANTHROPIC_API_KEY]
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

147
demo/webrtc_vs_websocket/app.py Normal file
View File

@@ -0,0 +1,147 @@
import json
import logging
import os
from pathlib import Path
import anthropic
import gradio as gr
import numpy as np
from dotenv import load_dotenv
from elevenlabs import ElevenLabs
from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import AdditionalOutputs, ReplyOnPause, Stream, get_twilio_turn_credentials
from fastrtc.utils import aggregate_bytes_to_16bit, audio_to_bytes
from gradio.utils import get_space
from groq import Groq
from pydantic import BaseModel
# Configure the root logger to WARNING to suppress debug messages from other libraries
logging.basicConfig(level=logging.WARNING)
# Create a console handler
console_handler = logging.FileHandler("gradio_webrtc.log")
console_handler.setLevel(logging.DEBUG)
# Create a formatter
formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
console_handler.setFormatter(formatter)
# Configure the logger for your specific library
logger = logging.getLogger("fastrtc")
logger.setLevel(logging.DEBUG)
logger.addHandler(console_handler)
load_dotenv()
groq_client = Groq()
claude_client = anthropic.Anthropic()
tts_client = ElevenLabs(api_key=os.environ["ELEVENLABS_API_KEY"])
curr_dir = Path(__file__).parent
def response(
audio: tuple[int, np.ndarray],
chatbot: list[dict] | None = None,
):
chatbot = chatbot or []
messages = [{"role": d["role"], "content": d["content"]} for d in chatbot]
prompt = groq_client.audio.transcriptions.create(
file=("audio-file.mp3", audio_to_bytes(audio)),
model="whisper-large-v3-turbo",
response_format="verbose_json",
).text
print("prompt", prompt)
chatbot.append({"role": "user", "content": prompt})
messages.append({"role": "user", "content": prompt})
response = claude_client.messages.create(
model="claude-3-5-haiku-20241022",
max_tokens=512,
messages=messages, # type: ignore
)
response_text = " ".join(
block.text # type: ignore
for block in response.content
if getattr(block, "type", None) == "text"
)
chatbot.append({"role": "assistant", "content": response_text})
yield AdditionalOutputs(chatbot)
iterator = tts_client.text_to_speech.convert_as_stream(
text=response_text,
voice_id="JBFqnCBsd6RMkjVDRZzb",
model_id="eleven_multilingual_v2",
output_format="pcm_24000",
)
for chunk in aggregate_bytes_to_16bit(iterator):
audio_array = np.frombuffer(chunk, dtype=np.int16).reshape(1, -1)
yield (24000, audio_array, "mono")
chatbot = gr.Chatbot(type="messages")
stream = Stream(
modality="audio",
mode="send-receive",
handler=ReplyOnPause(response),
additional_outputs_handler=lambda a, b: b,
additional_inputs=[chatbot],
additional_outputs=[chatbot],
rtc_configuration=get_twilio_turn_credentials() if get_space() else None,
concurrency_limit=20 if get_space() else None,
)
class Message(BaseModel):
role: str
content: str
class InputData(BaseModel):
webrtc_id: str
chatbot: list[Message]
app = FastAPI()
stream.mount(app)
@app.get("/")
async def _():
rtc_config = get_twilio_turn_credentials() if get_space() else None
html_content = (curr_dir / "index.html").read_text()
html_content = html_content.replace("__RTC_CONFIGURATION__", json.dumps(rtc_config))
return HTMLResponse(content=html_content, status_code=200)
@app.post("/input_hook")
async def _(body: InputData):
stream.set_input(body.webrtc_id, body.model_dump()["chatbot"])
return {"status": "ok"}
@app.get("/outputs")
def _(webrtc_id: str):
print("outputs", webrtc_id)
async def output_stream():
async for output in stream.output_stream(webrtc_id):
chatbot = output.args[0]
yield f"event: output\ndata: {json.dumps(chatbot[-2])}\n\n"
yield f"event: output\ndata: {json.dumps(chatbot[-1])}\n\n"
return StreamingResponse(output_stream(), media_type="text/event-stream")
if __name__ == "__main__":
import os
if (mode := os.getenv("MODE")) == "UI":
stream.ui.launch(server_port=7860, server_name="0.0.0.0")
elif mode == "PHONE":
stream.fastphone(host="0.0.0.0", port=7860)
else:
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)

630
demo/webrtc_vs_websocket/index.html Normal file
View File

@@ -0,0 +1,630 @@
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>WebRTC vs WebSocket Benchmark</title>
<script src="https://cdn.jsdelivr.net/npm/alawmulaw"></script>
<style>
body {
font-family: system-ui, -apple-system, sans-serif;
margin: 0;
padding: 20px;
background-color: #f5f5f5;
}
.container {
display: grid;
grid-template-columns: 1fr 1fr;
gap: 30px;
max-width: 1400px;
margin: 0 auto;
}
.panel {
background: white;
border-radius: 12px;
padding: 20px;
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.1);
}
.chat-container {
height: 400px;
overflow-y: auto;
border: 1px solid #e0e0e0;
border-radius: 8px;
padding: 15px;
margin-bottom: 15px;
}
.message {
margin-bottom: 10px;
padding: 8px 12px;
border-radius: 8px;
max-width: 80%;
}
.message.user {
background-color: #e3f2fd;
margin-left: auto;
}
.message.assistant {
background-color: #f5f5f5;
}
.metrics {
margin-top: 15px;
padding: 10px;
background: #f8f9fa;
border-radius: 8px;
}
.metric {
margin: 5px 0;
font-size: 14px;
}
button {
background-color: #1976d2;
color: white;
border: none;
padding: 10px 20px;
border-radius: 6px;
cursor: pointer;
font-size: 14px;
transition: background-color 0.2s;
}
button:hover {
background-color: #1565c0;
}
button:disabled {
background-color: #bdbdbd;
cursor: not-allowed;
}
h2 {
margin-top: 0;
color: #1976d2;
}
.visualizer {
width: 100%;
height: 100px;
margin: 10px 0;
background: #fafafa;
border-radius: 8px;
}
/* Add styles for disclaimer */
.disclaimer {
background-color: #fff3e0;
padding: 15px;
border-radius: 8px;
margin-bottom: 20px;
font-size: 14px;
line-height: 1.5;
max-width: 1400px;
margin: 0 auto 20px auto;
}
/* Update nav bar styles */
.nav-bar {
background-color: #f5f5f5;
padding: 10px 20px;
margin-bottom: 20px;
}
.nav-container {
max-width: 1400px;
margin: 0 auto;
display: flex;
gap: 10px;
}
.nav-button {
background-color: #1976d2;
color: white;
border: none;
padding: 8px 16px;
border-radius: 4px;
cursor: pointer;
text-decoration: none;
font-size: 14px;
transition: background-color 0.2s;
}
.nav-button:hover {
background-color: #1565c0;
}
/* Add styles for toast notifications */
.toast {
position: fixed;
top: 20px;
left: 50%;
transform: translateX(-50%);
padding: 16px 24px;
border-radius: 4px;
font-size: 14px;
z-index: 1000;
display: none;
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
}
.toast.error {
background-color: #f44336;
color: white;
}
.toast.warning {
background-color: #ffd700;
color: black;
}
</style>
</head>
<body>
<nav class="nav-bar">
<div class="nav-container">
<a href="./webrtc/docs" class="nav-button">WebRTC Docs</a>
<a href="./websocket/docs" class="nav-button">WebSocket Docs</a>
<a href="./telephone/docs" class="nav-button">Telephone Docs</a>
<a href="./ui" class="nav-button">UI</a>
</div>
</nav>
<div class="disclaimer">
This page compares the WebRTC Round-Trip-Time calculated from <code>getStats()</code> to the time taken to
process a ping/pong response pattern over websockets. It may not be a gold standard benchmark. Both WebRTC and
Websockets have their merits/advantages which is why FastRTC supports both. Artifacts in the WebSocket playback
audio are due to gaps in my frontend processing code and not FastRTC web server.
</div>
<div class="container">
<div class="panel">
<h2>WebRTC Connection</h2>
<div id="webrtc-chat" class="chat-container"></div>
<div id="webrtc-metrics" class="metrics">
<div class="metric">RTT (Round Trip Time): <span id="webrtc-rtt">-</span></div>
</div>
<button id="webrtc-button">Connect WebRTC</button>
</div>
<div class="panel">
<h2>WebSocket Connection</h2>
<div id="ws-chat" class="chat-container"></div>
<div id="ws-metrics" class="metrics">
<div class="metric">RTT (Round Trip Time): <span id="ws-rtt">0</span></div>
</div>
<button id="ws-button">Connect WebSocket</button>
</div>
</div>
<audio id="webrtc-audio" style="display: none;"></audio>
<audio id="ws-audio" style="display: none;"></audio>
<div id="error-toast" class="toast"></div>
<script>
// Shared utilities
function generateId() {
return Math.random().toString(36).substring(7);
}
function sendInput(id) {
return function handleMessage(event) {
const eventJson = JSON.parse(event.data);
if (eventJson.type === "send_input") {
fetch('/input_hook', {
method: 'POST',
headers: {
'Content-Type': 'application/json',
},
body: JSON.stringify({
webrtc_id: id,
chatbot: chatHistoryWebRTC
})
});
}
}
}
let chatHistoryWebRTC = [];
let chatHistoryWebSocket = [];
function addMessage(containerId, role, content) {
const container = document.getElementById(containerId);
const messageDiv = document.createElement('div');
messageDiv.classList.add('message', role);
messageDiv.textContent = content;
container.appendChild(messageDiv);
container.scrollTop = container.scrollHeight;
if (containerId === 'webrtc-chat') {
chatHistoryWebRTC.push({ role, content });
} else {
chatHistoryWebSocket.push({ role, content });
}
}
// WebRTC Implementation
let webrtcPeerConnection;
// Add this function to collect RTT stats
async function updateWebRTCStats() {
if (!webrtcPeerConnection) return;
const stats = await webrtcPeerConnection.getStats();
stats.forEach(report => {
if (report.type === 'candidate-pair' && report.state === 'succeeded') {
const rtt = report.currentRoundTripTime * 1000; // Convert to ms
document.getElementById('webrtc-rtt').textContent = `${rtt.toFixed(2)}ms`;
}
});
}
async function setupWebRTC() {
const button = document.getElementById('webrtc-button');
button.textContent = "Stop";
const config = __RTC_CONFIGURATION__;
webrtcPeerConnection = new RTCPeerConnection(config);
const webrtcId = generateId();
const timeoutId = setTimeout(() => {
const toast = document.getElementById('error-toast');
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
toast.className = 'toast warning';
toast.style.display = 'block';
// Hide warning after 5 seconds
setTimeout(() => {
toast.style.display = 'none';
}, 5000);
}, 5000);
try {
const stream = await navigator.mediaDevices.getUserMedia({ audio: true });
stream.getTracks().forEach(track => {
webrtcPeerConnection.addTrack(track, stream);
});
webrtcPeerConnection.addEventListener('track', (evt) => {
const audio = document.getElementById('webrtc-audio');
if (audio.srcObject !== evt.streams[0]) {
audio.srcObject = evt.streams[0];
audio.play();
}
});
const dataChannel = webrtcPeerConnection.createDataChannel('text');
dataChannel.onmessage = sendInput(webrtcId);
const offer = await webrtcPeerConnection.createOffer();
await webrtcPeerConnection.setLocalDescription(offer);
await new Promise((resolve) => {
if (webrtcPeerConnection.iceGatheringState === "complete") {
resolve();
} else {
const checkState = () => {
if (webrtcPeerConnection.iceGatheringState === "complete") {
webrtcPeerConnection.removeEventListener("icegatheringstatechange", checkState);
resolve();
}
};
webrtcPeerConnection.addEventListener("icegatheringstatechange", checkState);
}
});
const response = await fetch('/webrtc/offer', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({
sdp: webrtcPeerConnection.localDescription.sdp,
type: webrtcPeerConnection.localDescription.type,
webrtc_id: webrtcId
})
});
const serverResponse = await response.json();
await webrtcPeerConnection.setRemoteDescription(serverResponse);
// Setup event source for messages
const eventSource = new EventSource('/outputs?webrtc_id=' + webrtcId);
eventSource.addEventListener("output", (event) => {
const eventJson = JSON.parse(event.data);
addMessage('webrtc-chat', eventJson.role, eventJson.content);
});
// Add periodic stats collection
const statsInterval = setInterval(updateWebRTCStats, 1000);
// Store the interval ID on the connection
webrtcPeerConnection.statsInterval = statsInterval;
webrtcPeerConnection.addEventListener('connectionstatechange', () => {
if (webrtcPeerConnection.connectionState === 'connected') {
clearTimeout(timeoutId);
const toast = document.getElementById('error-toast');
toast.style.display = 'none';
}
});
} catch (err) {
clearTimeout(timeoutId);
console.error('WebRTC setup error:', err);
}
}
function webrtc_stop() {
if (webrtcPeerConnection) {
// Clear the stats interval
if (webrtcPeerConnection.statsInterval) {
clearInterval(webrtcPeerConnection.statsInterval);
}
// Close all tracks
webrtcPeerConnection.getSenders().forEach(sender => {
if (sender.track) {
sender.track.stop();
}
});
webrtcPeerConnection.close();
webrtcPeerConnection = null;
// Reset metrics display
document.getElementById('webrtc-rtt').textContent = '-';
}
}
// WebSocket Implementation
let webSocket;
let wsMetrics = {
pingStartTime: 0,
rttValues: []
};
// Load mu-law library
// Add load promise to track when the script is ready
function resample(audioData, fromSampleRate, toSampleRate) {
const ratio = fromSampleRate / toSampleRate;
const newLength = Math.round(audioData.length / ratio);
const result = new Float32Array(newLength);
for (let i = 0; i < newLength; i++) {
const position = i * ratio;
const index = Math.floor(position);
const fraction = position - index;
if (index + 1 < audioData.length) {
result[i] = audioData[index] * (1 - fraction) + audioData[index + 1] * fraction;
} else {
result[i] = audioData[index];
}
}
return result;
}
function convertToMulaw(audioData, sampleRate) {
// Resample to 8000 Hz if needed
if (sampleRate !== 8000) {
audioData = resample(audioData, sampleRate, 8000);
}
// Convert float32 [-1,1] to int16 [-32768,32767]
const int16Data = new Int16Array(audioData.length);
for (let i = 0; i < audioData.length; i++) {
int16Data[i] = Math.floor(audioData[i] * 32768);
}
// Convert to mu-law using the library
return alawmulaw.mulaw.encode(int16Data);
}
async function setupWebSocket() {
const button = document.getElementById('ws-button');
button.textContent = "Stop";
try {
const stream = await navigator.mediaDevices.getUserMedia({
audio: {
"echoCancellation": true,
"noiseSuppression": { "exact": true },
"autoGainControl": { "exact": true },
"sampleRate": { "ideal": 24000 },
"sampleSize": { "ideal": 16 },
"channelCount": { "exact": 1 },
}
});
const wsId = generateId();
wsMetrics.startTime = performance.now();
// Create audio context and analyser for visualization
const audioContext = new AudioContext();
const analyser = audioContext.createAnalyser();
const source = audioContext.createMediaStreamSource(stream);
source.connect(analyser);
// Connect to websocket endpoint
webSocket = new WebSocket(`${window.location.protocol === 'https:' ? 'wss:' : 'ws:'}//${window.location.host}/websocket/offer`);
webSocket.onopen = () => {
// Send initial start message
webSocket.send(JSON.stringify({
event: "start",
websocket_id: wsId
}));
// Setup audio processing
const processor = audioContext.createScriptProcessor(2048, 1, 1);
source.connect(processor);
processor.connect(audioContext.destination);
processor.onaudioprocess = (e) => {
const inputData = e.inputBuffer.getChannelData(0);
const mulawData = convertToMulaw(inputData, audioContext.sampleRate);
const base64Audio = btoa(String.fromCharCode.apply(null, mulawData));
if (webSocket.readyState === WebSocket.OPEN) {
webSocket.send(JSON.stringify({
event: "media",
media: {
payload: base64Audio
}
}));
}
};
// Add ping interval
webSocket.pingInterval = setInterval(() => {
wsMetrics.pingStartTime = performance.now();
webSocket.send(JSON.stringify({
event: "ping"
}));
}, 500);
};
// Setup audio output context
const outputContext = new AudioContext({ sampleRate: 24000 });
const sampleRate = 24000; // Updated to match server sample rate
let audioQueue = [];
let isPlaying = false;
webSocket.onmessage = (event) => {
const data = JSON.parse(event.data);
if (data?.type === "send_input") {
console.log("sending input")
fetch('/input_hook', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ webrtc_id: wsId, chatbot: chatHistoryWebSocket })
});
}
if (data.event === "media") {
// Process received audio
const audioData = atob(data.media.payload);
const mulawData = new Uint8Array(audioData.length);
for (let i = 0; i < audioData.length; i++) {
mulawData[i] = audioData.charCodeAt(i);
}
// Convert mu-law to linear PCM
const linearData = alawmulaw.mulaw.decode(mulawData);
// Create an AudioBuffer
const audioBuffer = outputContext.createBuffer(1, linearData.length, sampleRate);
const channelData = audioBuffer.getChannelData(0);
// Fill the buffer with the decoded data
for (let i = 0; i < linearData.length; i++) {
channelData[i] = linearData[i] / 32768.0;
}
// Queue the audio buffer
audioQueue.push(audioBuffer);
// Start playing if not already playing
if (!isPlaying) {
playNextBuffer();
}
}
// Add pong handler
if (data.event === "pong") {
const rtt = performance.now() - wsMetrics.pingStartTime;
wsMetrics.rttValues.push(rtt);
// Keep only last 20 values for running mean
if (wsMetrics.rttValues.length > 20) {
wsMetrics.rttValues.shift();
}
const avgRtt = wsMetrics.rttValues.reduce((a, b) => a + b, 0) / wsMetrics.rttValues.length;
document.getElementById('ws-rtt').textContent = `${avgRtt.toFixed(2)}ms`;
return;
}
};
function playNextBuffer() {
if (audioQueue.length === 0) {
isPlaying = false;
return;
}
isPlaying = true;
const bufferSource = outputContext.createBufferSource();
bufferSource.buffer = audioQueue.shift();
bufferSource.connect(outputContext.destination);
bufferSource.onended = playNextBuffer;
bufferSource.start();
}
const eventSource = new EventSource('/outputs?webrtc_id=' + wsId);
eventSource.addEventListener("output", (event) => {
console.log("ws output", event);
const eventJson = JSON.parse(event.data);
addMessage('ws-chat', eventJson.role, eventJson.content);
});
} catch (err) {
console.error('WebSocket setup error:', err);
button.disabled = false;
}
}
function ws_stop() {
if (webSocket) {
webSocket.send(JSON.stringify({
event: "stop"
}));
// Clear ping interval
if (webSocket.pingInterval) {
clearInterval(webSocket.pingInterval);
}
// Reset RTT display
document.getElementById('ws-rtt').textContent = '-';
wsMetrics.rttValues = [];
// Clear the stats interval
if (webSocket.statsInterval) {
clearInterval(webSocket.statsInterval);
}
webSocket.close();
}
}
// Event Listeners
document.getElementById('webrtc-button').addEventListener('click', () => {
const button = document.getElementById('webrtc-button');
if (button.textContent === 'Connect WebRTC') {
setupWebRTC();
} else {
webrtc_stop();
button.textContent = 'Connect WebRTC';
}
});
const ws_start_button = document.getElementById('ws-button')
ws_start_button.addEventListener('click', () => {
if (ws_start_button.textContent === 'Connect WebSocket') {
setupWebSocket();
ws_start_button.textContent = 'Stop';
} else {
ws_stop();
ws_start_button.textContent = 'Connect WebSocket';
}
});
document.addEventListener("beforeunload", () => {
ws_stop();
webrtc_stop();
});
</script>
</body>
</html>

6
demo/webrtc_vs_websocket/requirements.txt Normal file
View File

@@ -0,0 +1,6 @@
fastrtc[vad]
elevenlabs
groq
anthropic
twilio
python-dotenv

Some files were not shown because too many files have changed in this diff Show More