# CosyVoice | Multilingual Large Voice Generation Model

**URL:** <https://onehack.st/t/cosyvoice-multilingual-large-voice-generation-model/291594>\
**Category:** Tools & Scripts\
**Tags:** freebies, audio, ai\
**Created:** [September 17, 2024, 11:34pm UTC](https://onehack.st/t/cosyvoice-multilingual-large-voice-generation-model/291594 "2024-09-17T23:34:08Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![TheStrength](https://onehack.st/user_avatar/onehack.st/thestrength/32/46753_2.png) [@TheStrength](https://onehack.st/u/TheStrength)\
**Post date:** [September 17, 2024, 11:34pm UTC](https://onehack.st/t/cosyvoice-multilingual-large-voice-generation-model/291594/1 "2024-09-17T23:34:08Z")

</div>

# osyVoice

## 👉🏻 [CosyVoice Demos](https://fun-audio-llm.github.io/) 👈🏻

[[CosyVoice Paper](https://fun-audio-llm.github.io/pdf/CosyVoice_v1.pdf)][[CosyVoice Studio](https://www.modelscope.cn/studios/iic/CosyVoice-300M)][[CosyVoice Code](https://github.com/FunAudioLLM/CosyVoice)]

For `SenseVoice`, visit [SenseVoice repo](https://github.com/FunAudioLLM/SenseVoice) and [SenseVoice space](https://www.modelscope.cn/studios/iic/SenseVoice).

## Roadmap

- 2024/07
  - Flow matching training support
  - WeTextProcessing support when ttsfrd is not avaliable
  - Fastapi server and client

- 2024/08
  - Repetition Aware Sampling(RAS) inference for llm stability
  - Streaming inference mode support, including kv cache and sdpa for rtf optimization

- 2024/09
  - 25hz cosyvoice base model
  - 25hz cosyvoice voice conversion model

- 2024/10
  - 50hz llama based llm model which supports lora finetune

- TBD
  - Support more instruction mode
  - Voice conversion
  - Music generation
  - Training script sample based on Mandarin
  - CosyVoice-500M trained with more multi-lingual data
  - More…

## Install

**Clone and install**

- Clone the repo

git clone --recursive [https://github.com/FunAudioLLM/CosyVoice.git](https://github.com/FunAudioLLM/CosyVoice.git) # If you failed to clone submodule due to network failures, please run following command until success cd CosyVoice git submodule update --init --recursive

- Install Conda: please see [https://docs.conda.io/en/latest/miniconda.html](https://docs.conda.io/en/latest/miniconda.html)
- Create Conda env:

conda create -n cosyvoice python=3.8 conda activate cosyvoice # pynini is required by WeTextProcessing, use conda to install it as it can be executed on all platform. conda install -y -c conda-forge pynini==2.1.5 pip install -r requirements.txt -i [https://mirrors.aliyun.com/pypi/simple/](https://mirrors.aliyun.com/pypi/simple/) --trusted-host=[mirrors.aliyun.com](http://mirrors.aliyun.com) # If you encounter sox compatibility issues # ubuntu sudo apt-get install sox libsox-dev # centos sudo yum install sox sox-devel

**Model download**

We strongly recommend that you download our pretrained `CosyVoice-300M` `CosyVoice-300M-SFT` `CosyVoice-300M-Instruct` model and `CosyVoice-ttsfrd` resource.

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.

# 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’)

# 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](https://www.modelscope.cn/iic/CosyVoice-300M.git) pretrained\_models/CosyVoice-300M git clone [https://www.modelscope.cn/iic/CosyVoice-300M-SFT.git](https://www.modelscope.cn/iic/CosyVoice-300M-SFT.git) pretrained\_models/CosyVoice-300M-SFT git clone [https://www.modelscope.cn/iic/CosyVoice-300M-Instruct.git](https://www.modelscope.cn/iic/CosyVoice-300M-Instruct.git) pretrained\_models/CosyVoice-300M-Instruct git clone [https://www.modelscope.cn/iic/CosyVoice-ttsfrd.git](https://www.modelscope.cn/iic/CosyVoice-ttsfrd.git) pretrained\_models/CosyVoice-ttsfrd

Optionaly, you can unzip `ttsfrd` resouce and install `ttsfrd` package for better text normalization performance.

Notice that this step is not necessary. If you do not install `ttsfrd` package, we will use WeTextProcessing by default.

cd pretrained\_models/CosyVoice-ttsfrd/ unzip resource.zip -d . pip install ttsfrd-0.3.6-cp38-cp38-linux\_x86\_64.whl

**Basic Usage**

For zero\_shot/cross\_lingual inference, please use `CosyVoice-300M` model. For sft inference, please use `CosyVoice-300M-SFT` model. For instruct inference, please use `CosyVoice-300M-Instruct` model. First, add `third_party/Matcha-TTS` to your `PYTHONPATH`.

export PYTHONPATH=third\_party/Matcha-TTS

from cosyvoice.cli.cosyvoice import CosyVoice from cosyvoice.utils.file\_utils import load\_wav import torchaudio cosyvoice = CosyVoice(‘pretrained\_models/CosyVoice-300M-SFT’) # sft usage print(cosyvoice.list\_avaliable\_spks()) # change stream=True for chunk stream inference for i, j in enumerate(cosyvoice.inference\_sft(‘你好，我是通义生成式语音大模型，请问有什么可以帮您的吗？’, ‘中文女’, stream=False)): torchaudio.save(‘sft\_{}.wav’.format(i), j[‘tts\_speech’], 22050) cosyvoice = CosyVoice(‘pretrained\_models/CosyVoice-300M’) # zero\_shot usage, \<|zh|\>\<|en|\>\<|jp|\>\<|yue|\>\<|ko|\> for Chinese/English/Japanese/Cantonese/Korean prompt\_speech\_16k = load\_wav(‘zero\_shot\_prompt.wav’, 16000) for i, j in enumerate(cosyvoice.inference\_zero\_shot(‘收到好友从远方寄来的生日礼物，那份意外的惊喜与深深的祝福让我心中充满了甜蜜的快乐，笑容如花儿般绽放。’, ‘希望你以后能够做的比我还好呦。’, prompt\_speech\_16k, stream=False)): torchaudio.save(‘zero\_shot\_{}.wav’.format(i), j[‘tts\_speech’], 22050) # cross\_lingual usage prompt\_speech\_16k = load\_wav(‘cross\_lingual\_prompt.wav’, 16000) for i, j in enumerate(cosyvoice.inference\_cross\_lingual(‘\<|en|\>And then later on, fully acquiring that company. So keeping management in line, interest in line with the asset that's coming into the family is a reason why sometimes we don't buy the whole thing.’, prompt\_speech\_16k, stream=False)): torchaudio.save(‘cross\_lingual\_{}.wav’.format(i), j[‘tts\_speech’], 22050) cosyvoice = CosyVoice(‘pretrained\_models/CosyVoice-300M-Instruct’) # instruct usage, support [laughter][breath] for i, j in enumerate(cosyvoice.inference\_instruct(‘在面对挑战时，他展现了非凡的 **勇气** 与 **智慧** 。’, ‘中文男’, ‘Theo 'Crimson', is a fiery, passionate rebel leader. Fights with fervor for justice, but struggles with impulsiveness.’, stream=False)): torchaudio.save(‘instruct\_{}.wav’.format(i), j[‘tts\_speech’], 22050)

**Start web demo**

You can use our web demo page to get familiar with CosyVoice quickly. We support sft/zero\_shot/cross\_lingual/instruct inference in web demo.

Please see the demo website for details.

# change iic/CosyVoice-300M-SFT for sft inference, or iic/CosyVoice-300M-Instruct for instruct inference python3 webui.py --port 50000 --model\_dir pretrained\_models/CosyVoice-300M

**Advanced Usage**

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.

**Build for deployment**

Optionally, if you want to use grpc for service deployment, you can run following steps. Otherwise, you can just ignore this step.

cd runtime/python docker build -t cosyvoice:v1.0 . # change iic/CosyVoice-300M to iic/CosyVoice-300M-Instruct if you want to use instruct inference # for grpc usage docker run -d --runtime=nvidia -p 50000:50000 cosyvoice:v1.0 /bin/bash -c “cd /opt/CosyVoice/CosyVoice/runtime/python/grpc && python3 server.py --port 50000 --max\_conc 4 --model\_dir iic/CosyVoice-300M && sleep infinity” cd grpc && python3 client.py --port 50000 --mode \<sft|zero\_shot|cross\_lingual|instruct\> # for fastapi usage docker run -d --runtime=nvidia -p 50000:50000 cosyvoice:v1.0 /bin/bash -c “cd /opt/CosyVoice/CosyVoice/runtime/python/fastapi && python3 server.py --port 50000 --model\_dir iic/CosyVoice-300M && sleep infinity” cd fastapi && python3 client.py --port 50000 --mode \<sft|zero\_shot|cross\_lingual|instruct\>

## Discussion & Communication

You can directly discuss on [Github Issues](https://github.com/FunAudioLLM/CosyVoice/issues).

You can also scan the QR code to join our official Dingding chat group.

[![](https://github.com/FunAudioLLM/CosyVoice/raw/main/asset/dingding.png)](https://github.com/FunAudioLLM/CosyVoice/blob/main/asset/dingding.png)

## Acknowledge

1. We borrowed a lot of code from [FunASR](https://github.com/modelscope/FunASR).
2. We borrowed a lot of code from [FunCodec](https://github.com/modelscope/FunCodec).
3. We borrowed a lot of code from [Matcha-TTS](https://github.com/shivammehta25/Matcha-TTS).
4. We borrowed a lot of code from [AcademiCodec](https://github.com/yangdongchao/AcademiCodec).
5. We borrowed a lot of code from [WeNet](https://github.com/wenet-e2e/wenet).

## Disclaimer

The content provided above is for academic purposes only and is intended to demonstrate technical capabilities. Some examples are sourced from the internet. If any content infringes on your rights, please contact us to request its removal.

## GitHub:

[https://github.com/FunAudioLLM/CosyVoice](https://github.com/FunAudioLLM/CosyVoice)