Views
No views yet
1git clone --recursive https://github.com/FunAudioLLM/CosyVoice.git
2# If you failed to clone submodule due to network failures, please run following command until success
3cd CosyVoice
4git submodule update --init --recursive1conda create -n cosyvoice python=3.10
2conda activate cosyvoice
3# pynini is required by WeTextProcessing, use conda to install it as it can be executed on all platform.
4conda install -y -c conda-forge pynini==2.1.5
5pip install -r requirements.txt -i https://mirrors.aliyun.com/pypi/simple/ --trusted-host=mirrors.aliyun.com
6
7# If you encounter sox compatibility issues
8# ubuntu
9sudo apt-get install sox libsox-dev
10# centos
11sudo yum install sox sox-develCosyVoice2-0.5B for better performance.
Follow code below for detailed usage of each model.1import sys
2sys.path.append('third_party/Matcha-TTS')
3from cosyvoice.cli.cosyvoice import CosyVoice, CosyVoice2
4from cosyvoice.utils.file_utils import load_wav
5import torchaudio1cosyvoice = CosyVoice2('ASLP-lab/Cosyvoice2-Yue', load_jit=False, load_trt=False, fp16=False)
2
3# NOTE if you want to reproduce the results on https://funaudiollm.github.io/cosyvoice2, please add text_frontend=False during inference
4# zero_shot usage
5prompt_speech_16k = load_wav('zero_shot_prompt.wav', 16000)
6
7# instruct usage
8for i, j in enumerate(cosyvoice.inference_instruct2('收到朋友从远方寄嚟嘅生日礼物，呢份意外嘅惊喜同埋满满嘅祝福令我内心充满咗甜蜜嘅快乐，个笑容就好似花咁咧盛开住。', '用粤语说这句话', prompt_speech_16k, stream=False)):
9 torchaudio.save('instruct_{}.wav'.format(i), j['tts_speech'], cosyvoice.sample_rate)