deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-04 09:29:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

add download models script and fastapi server to serve tts

Browse Source
This commit is contained in:
iflamed
2024-07-08 18:51:06 +08:00
parent 4e43a9d98b
commit fff6f9f1e0
4 changed files with 61 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
README.md
View File

@@ -37,17 +37,13 @@ We strongly recommend that you download our pretrained `CosyVoice-300M` `CosyVoi
If you are expert in this field, and you are only interested in training your own CosyVoice model from scratch, you can skip this step.
``` python
# SDK模型下载
from modelscope import snapshot_download
snapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M')
snapshot_download('iic/CosyVoice-300M-SFT', local_dir='pretrained_models/CosyVoice-300M-SFT')
snapshot_download('iic/CosyVoice-300M-Instruct', local_dir='pretrained_models/CosyVoice-300M-Instruct')
snapshot_download('iic/CosyVoice-ttsfrd', local_dir='pretrained_models/CosyVoice-ttsfrd')
Download models with python script.
``` shell
python download.py
```
Download models with git, you should install `git lfs` first.
``` sh
# git模型下载，请确保已安装git lfs
mkdir -p pretrained_models
git clone https://www.modelscope.cn/iic/CosyVoice-300M.git pretrained_models/CosyVoice-300M
git clone https://www.modelscope.cn/iic/CosyVoice-300M-SFT.git pretrained_models/CosyVoice-300M-SFT
@@ -120,6 +116,14 @@ python3 webui.py --port 50000 --model_dir pretrained_models/CosyVoice-300M
For advanced user, we have provided train and inference scripts in `examples/libritts/cosyvoice/run.sh`.
You can get familiar with CosyVoice following this recipie.
**Serve with FastAPI**
```sh
# For development
fastapi dev --port 3003
# For production
fastapi run --port 3003
```
**Build for deployment**
Optionally, if you want to use grpc for service deployment,

6
download.py Normal file
View File

@@ -0,0 +1,6 @@
# SDK模型下载
from modelscope import snapshot_download
snapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M')
snapshot_download('iic/CosyVoice-300M-SFT', local_dir='pretrained_models/CosyVoice-300M-SFT')
snapshot_download('iic/CosyVoice-300M-Instruct', local_dir='pretrained_models/CosyVoice-300M-Instruct')
snapshot_download('iic/CosyVoice-ttsfrd', local_dir='pretrained_models/CosyVoice-ttsfrd')

40
main.py Normal file
View File

@@ -0,0 +1,40 @@
import io,time
from fastapi import FastAPI, Response
from fastapi.responses import HTMLResponse
from cosyvoice.cli.cosyvoice import CosyVoice
import torchaudio
cosyvoice = CosyVoice('pretrained_models/CosyVoice-300M-SFT')
# sft usage
print(cosyvoice.list_avaliable_spks())
app = FastAPI()
@app.get("/api/voice/tts")
async def tts(query: str, role: str):
start = time.process_time()
output = cosyvoice.inference_sft(query, role)
end = time.process_time()
print("infer time:", end-start, "seconds")
buffer = io.BytesIO()
torchaudio.save(buffer, output['tts_speech'], 22050, format="wav")
buffer.seek(0)
return Response(content=buffer.read(-1), media_type="audio/wav")
@app.get("/api/voice/roles")
async def roles():
return {"roles": cosyvoice.list_avaliable_spks()}
@app.get("/", response_class=HTMLResponse)
async def root():
return """
<!DOCTYPE html>
<html lang=zh-cn>
<head>
<meta charset=utf-8>
<title>Api information</title>
</head>
<body>
Get the supported tones from the Roles API first, then enter the tones and textual content in the TTS API for synthesis. <a href='./docs'>Documents of API</a>
</body>
</html>
"""

4
requirements.txt
View File

@@ -25,4 +25,6 @@ soundfile==0.12.1
tensorboard==2.14.0
torch==2.0.1
torchaudio==2.0.2
wget==3.2
wget==3.2
fastapi==0.111.0
fastapi-cli==0.0.4