cosyvoice 安装，多音色的控制 #190

New Issue

2025-12-23T02:21:43+08:00

bigtran commented

2025-12-23 02:21:43 +08:00

1. 基础工具

sudo apt update && sudo apt install -y git build-essential sox libsox-dev ffmpeg unzip curl

2、安装conda

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
conda --version

3、创建隔离环境

conda create -n cosyvoice python=3.10 -y
会提示
conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/main
conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/r
conda activate cosyvoice
安装pynini
conda install -y -c conda-forge pynini==2.1.5

4、拉代码

git clone --recursive https://github.com/FunAudioLLM/CosyVoice.git
cd CosyVoice

sed -i 's|https://github.com|https://ghfast.top/https://github.com|g' .gitmodules
git submodule update --init --recursive

5、安装python依赖

pip config set global.index-url https://mirrors.aliyun.com/pypi/simple
python -m pip install -U pip setuptools wheel
python -m pip install -r requirements.txt

6、下载预训练模型

pip install modelscope

300M和0.5B 的模型共6G的模型

python -c "from modelscope import snapshot_download;\nsnapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M');
snapshot_download('iic/CosyVoice2-0.5B', local_dir='pretrained_models/CosyVoice2-0.5B')"

300M的模型，约400M

python -c "from modelscope import snapshot_download;
snapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M')"

7、试验

python -m cosyvoice.cli.cosyvoice synthesis --text "你好" --spk_id female1
查看spk_id
python -m cosyvoice.cli.cosyvoice list_spks

整体安装非常耗时！

# 1. 基础工具 sudo apt update && sudo apt install -y git build-essential sox libsox-dev ffmpeg unzip curl # 2、安装conda wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh chmod +x Miniconda3-latest-Linux-x86_64.sh bash Miniconda3-latest-Linux-x86_64.sh conda --version # 3、创建隔离环境 conda create -n cosyvoice python=3.10 -y 会提示 conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/main conda tos accept --override-channels --channel https://repo.anaconda.com/pkgs/r conda activate cosyvoice 安装pynini conda install -y -c conda-forge pynini==2.1.5 # 4、拉代码 git clone --recursive https://github.com/FunAudioLLM/CosyVoice.git cd CosyVoice sed -i 's|https://github.com|https://ghfast.top/https://github.com|g' .gitmodules git submodule update --init --recursive # 5、安装python依赖 pip config set global.index-url https://mirrors.aliyun.com/pypi/simple python -m pip install -U pip setuptools wheel python -m pip install -r requirements.txt # 6、下载预训练模型 pip install modelscope ## 300M和0.5B 的模型共6G的模型 python -c "from modelscope import snapshot_download;\\nsnapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M');\ snapshot_download('iic/CosyVoice2-0.5B', local_dir='pretrained_models/CosyVoice2-0.5B')" ## 300M的模型，约400M python -c "from modelscope import snapshot_download; \ snapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M')" # 7、试验 python -m cosyvoice.cli.cosyvoice synthesis --text "你好" --spk_id female1 查看spk_id python -m cosyvoice.cli.cosyvoice list_spks # 整体安装非常耗时！

bigtran commented

2025-12-23 02:27:14 +08:00

查看服务器配置

系统版本

lsb_release -a
uname -r

cpu

lscpu

内存

free -h

显卡

nvidia-smi
lspci | grep -i vga

# 查看服务器配置 ## 系统版本 lsb_release -a uname -r ## cpu lscpu ## 内存 free -h ## 显卡 nvidia-smi lspci | grep -i vga

bigtran commented

2025-12-23 02:44:18 +08:00

#千万不要下载0.5B

#千万不要下载0.5B ![image.png](/attachments/7fe0b87c-b213-420c-ab7e-a9e65c6161a8)

image.png

136 KiB

bigtran added the 4-document label 2025-12-23 03:16:24 +08:00

bigtran commented

2025-12-23 12:51:06 +08:00

在 cosyvoice 环境里

pip install hyperpyyaml
pip install torch
pip install numpy
pip install onnxruntime
pip install whisper
pip install torchaudio
pip install inflect
pip install wetext
pip install regex
pip install transformers
pip install omegaconf
pip install scipy
pip install ttsfrd

# 在 cosyvoice 环境里 pip install hyperpyyaml pip install torch pip install numpy pip install onnxruntime pip install whisper pip install torchaudio pip install inflect pip install wetext pip install regex pip install transformers pip install omegaconf pip install scipy pip install ttsfrd

bigtran closed this issue

2025-12-27 22:37:04 +08:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: temp-tickets/tickets#190

cosyvoice 安装，多音色的控制 #190

1. 基础工具

2、安装conda

3、创建隔离环境

4、拉代码

5、安装python依赖

6、下载预训练模型

300M和0.5B 的 模型 共6G的模型

300M的模型，约400M

7、试验

整体安装非常耗时！

查看服务器配置

系统版本

cpu

内存

显卡

在 cosyvoice 环境里

300M和0.5B 的模型共6G的模型