diff --git a/backend/fastrtc/__init__.py b/backend/fastrtc/__init__.py
index 43f61ec..2e08399 100644
--- a/backend/fastrtc/__init__.py
+++ b/backend/fastrtc/__init__.py
@@ -17,7 +17,11 @@ from .reply_on_pause import AlgoOptions, ReplyOnPause
from .reply_on_stopwords import ReplyOnStopWords
from .speech_to_text import MoonshineSTT, get_stt_model
from .stream import Stream, UIArgs
-from .text_to_speech import KokoroTTSOptions, get_tts_model
+from .text_to_speech import (
+ CartesiaTTSOptions,
+ KokoroTTSOptions,
+ get_tts_model,
+)
from .tracks import (
AsyncAudioVideoStreamHandler,
AsyncStreamHandler,
@@ -87,4 +91,5 @@ __all__ = [
"VideoStreamHandler",
"CloseStream",
"get_current_context",
+ "CartesiaTTSOptions",
]
diff --git a/backend/fastrtc/stream.py b/backend/fastrtc/stream.py
index d1b53ac..1a2acdc 100644
--- a/backend/fastrtc/stream.py
+++ b/backend/fastrtc/stream.py
@@ -59,6 +59,8 @@ class UIArgs(TypedDict):
If "submit", the input will be sent when the submit event is triggered by the user.
If "change", the input will be sent whenever the user changes the input value.
"""
+ hide_title: NotRequired[bool]
+ """If True, the title and subtitle will not be displayed."""
class Stream(WebRTCConnectionMixin):
@@ -339,21 +341,22 @@ class Stream(WebRTCConnectionMixin):
same_components.append(component)
if self.modality == "video" and self.mode == "receive":
with gr.Blocks() as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Video Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Video Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
with gr.Column():
if additional_input_components:
@@ -391,21 +394,22 @@ class Stream(WebRTCConnectionMixin):
)
elif self.modality == "video" and self.mode == "send":
with gr.Blocks() as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Video Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Video Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
if additional_input_components:
with gr.Column():
@@ -494,21 +498,22 @@ class Stream(WebRTCConnectionMixin):
)
elif self.modality == "audio" and self.mode == "receive":
with gr.Blocks() as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
with gr.Column():
for component in additional_input_components:
@@ -549,21 +554,22 @@ class Stream(WebRTCConnectionMixin):
)
elif self.modality == "audio" and self.mode == "send":
with gr.Blocks() as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
with gr.Column():
with gr.Group():
@@ -604,21 +610,22 @@ class Stream(WebRTCConnectionMixin):
)
elif self.modality == "audio" and self.mode == "send-receive":
with gr.Blocks() as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Audio Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
with gr.Column():
with gr.Group():
@@ -662,21 +669,22 @@ class Stream(WebRTCConnectionMixin):
css = """.my-group {max-width: 600px !important; max-height: 600 !important;}
.my-column {display: flex !important; justify-content: center !important; align-items: center !important};"""
with gr.Blocks(css=css) as demo:
- gr.HTML(
- f"""
-
- {ui_args.get("title", "Audio Video Streaming (Powered by FastRTC ⚡️)")}
-
- """
- )
- if ui_args.get("subtitle"):
- gr.Markdown(
+ if not ui_args.get("hide_title"):
+ gr.HTML(
f"""
-
- {ui_args.get("subtitle")}
-
- """
+
+ {ui_args.get("title", "Audio Video Streaming (Powered by FastRTC ⚡️)")}
+
+ """
)
+ if ui_args.get("subtitle"):
+ gr.Markdown(
+ f"""
+
+ {ui_args.get("subtitle")}
+
+ """
+ )
with gr.Row():
with gr.Column(elem_classes=["my-column"]):
with gr.Group(elem_classes=["my-group"]):
diff --git a/backend/fastrtc/text_to_speech/__init__.py b/backend/fastrtc/text_to_speech/__init__.py
index 2cc082a..0d55538 100644
--- a/backend/fastrtc/text_to_speech/__init__.py
+++ b/backend/fastrtc/text_to_speech/__init__.py
@@ -1,3 +1,7 @@
-from .tts import KokoroTTSOptions, get_tts_model
+from .tts import (
+ CartesiaTTSOptions,
+ KokoroTTSOptions,
+ get_tts_model,
+)
-__all__ = ["get_tts_model", "KokoroTTSOptions"]
+__all__ = ["get_tts_model", "KokoroTTSOptions", "CartesiaTTSOptions"]
diff --git a/backend/fastrtc/text_to_speech/tts.py b/backend/fastrtc/text_to_speech/tts.py
index 37743be..f800e4b 100644
--- a/backend/fastrtc/text_to_speech/tts.py
+++ b/backend/fastrtc/text_to_speech/tts.py
@@ -2,7 +2,7 @@ import asyncio
import importlib.util
import re
from collections.abc import AsyncGenerator, Generator
-from dataclasses import dataclass
+from dataclasses import dataclass, field
from functools import lru_cache
from typing import Literal, Protocol, TypeVar
@@ -153,10 +153,11 @@ class KokoroTTSModel(TTSModel):
break
+@dataclass
class CartesiaTTSOptions(TTSOptions):
voice: str = "71a7ad14-091c-4e8e-a314-022ece01c121"
language: str = "en"
- emotion: list[str] = []
+ emotion: list[str] = field(default_factory=list)
cartesia_version: str = "2024-06-10"
model: str = "sonic-2"
sample_rate: int = 22_050
diff --git a/demo/talk_to_llama4/AV_Huggy.png b/demo/talk_to_llama4/AV_Huggy.png
new file mode 100644
index 0000000..6a9fb74
Binary files /dev/null and b/demo/talk_to_llama4/AV_Huggy.png differ
diff --git a/demo/talk_to_llama4/app.py b/demo/talk_to_llama4/app.py
index 9929fca..4010fc5 100644
--- a/demo/talk_to_llama4/app.py
+++ b/demo/talk_to_llama4/app.py
@@ -9,11 +9,12 @@ from fastapi import FastAPI
from fastapi.responses import HTMLResponse, StreamingResponse
from fastrtc import (
AdditionalOutputs,
+ CartesiaTTSOptions,
ReplyOnPause,
Stream,
- audio_to_bytes,
get_cloudflare_turn_credentials_async,
get_current_context,
+ get_stt_model,
get_tts_model,
)
from groq import Groq
@@ -22,9 +23,11 @@ from numpy.typing import NDArray
curr_dir = Path(__file__).parent
load_dotenv()
-tts_model = get_tts_model()
+tts_model = get_tts_model(
+ model="cartesia", cartesia_api_key=os.getenv("CARTESIA_API_KEY")
+)
groq = Groq(api_key=os.getenv("GROQ_API_KEY"))
-
+stt_model = get_stt_model()
conversations: dict[str, list[dict[str, str]]] = {}
@@ -43,14 +46,8 @@ def response(user_audio: tuple[int, NDArray[np.int16]]):
]
messages = conversations[context.webrtc_id]
- transcription = groq.audio.transcriptions.create(
- file=("audio.wav", audio_to_bytes(user_audio)),
- model="distil-whisper-large-v3-en",
- response_format="verbose_json",
- )
- print(transcription.text)
-
- messages.append({"role": "user", "content": transcription.text})
+ transcription = stt_model.stt(user_audio)
+ messages.append({"role": "user", "content": transcription})
completion = groq.chat.completions.create( # type: ignore
model="meta-llama/llama-4-scout-17b-16e-instruct",
@@ -68,7 +65,9 @@ def response(user_audio: tuple[int, NDArray[np.int16]]):
long_response = response["long"]
messages.append({"role": "assistant", "content": long_response})
conversations[context.webrtc_id] = messages
- yield from tts_model.stream_tts_sync(short_response)
+ yield from tts_model.stream_tts_sync(
+ short_response, options=CartesiaTTSOptions(sample_rate=24_000)
+ )
yield AdditionalOutputs(messages)
@@ -78,9 +77,22 @@ stream = Stream(
mode="send-receive",
additional_outputs=[gr.Chatbot(type="messages")],
additional_outputs_handler=lambda old, new: new,
- rtc_configuration=get_cloudflare_turn_credentials_async,
+ rtc_configuration=None,
+ ui_args={"hide_title": True},
)
+with gr.Blocks() as demo:
+ gr.HTML(
+ f"""
+
+ .parent / )
FastRTC + Cartesia TTS = Blazing Fast LLM Audio
+
+ """
+ )
+ stream.ui.render()
+
+stream.ui = demo
+
app = FastAPI()
stream.mount(app)
@@ -109,9 +121,13 @@ async def _(webrtc_id: str):
if __name__ == "__main__":
import os
+ from pathlib import Path
if (mode := os.getenv("MODE")) == "UI":
- stream.ui.launch(server_port=7860)
+ stream.ui.launch(
+ server_port=7860,
+ allowed_paths=[str((Path(__file__).parent / "AV_Huggy.png").resolve())],
+ )
elif mode == "PHONE":
raise ValueError("Phone mode not supported")
else: