daocodedao / fish-speech-ext

项目

原项目： https://github.com/fishaudio/fish-speech.git

说明： https://speech.fish.audio/inference/#_2

安装

环境

python3.10 -m venv venv_fish source venv_fish/bin/activate

cuda 121

#pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu121

cuda 118

pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu118

pip3 install ninja && MAX_JOBS=4 pip3 install flash-attn==2.4.2 --no-build-isolation

以下方法没法获得模型

https://hf-mirror.com/
HF_ENDPOINT=https://hf-mirror.com huggingface-cli login
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

git config --global http.proxy "http://192.168.0.77:18808"
git config --global https.proxy "http://192.168.0.77:18808"


git config --global --unset https.proxy
git config --global --unset http.proxy


huggingface-cli download fishaudio/speech-lm-v1 vqgan-v1.pth --local-dir checkpoints
huggingface-cli download fishaudio/speech-lm-v1 text2semantic-400m-v0.2-4k.pth --local-dir checkpoints

改为clone

#s机房，非 77 机器上
git config --global http.proxy "http://192.168.0.77:18808"
git config --global https.proxy "http://192.168.0.77:18808"

git lfs install
git clone https://huggingface.co/fishaudio/speech-lm-v1
用户名： 
daocode
秘钥：
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


# clone 完成后
git config --global --unset https.proxy
git config --global --unset http.proxy

mkdir -p fish-speech/checkpoints/
cp speech-lm-v1/vqgan-v1.pth  fish-speech/checkpoints/
cp speech-lm-v1/text2semantic-400m-v0.2-4k.pth fish-speech/checkpoints/

对外端口

cat /data/work/frp/frpc.ini 
vim /data/work/frp/frpc.ini 

[ssh-fish5001]
type = tcp
local_ip = 127.0.0.1
local_port = 5001
remote_port = 5001
use_encryption = false
use_compression = false


[ssh-fish8000]
type = tcp
local_ip = 127.0.0.1
local_port = 8000
remote_port = 5002
use_encryption = false
use_compression = false


# 重启frp
systemctl restart  supervisor
supervisorctl reload

启动

start.sh

source venv_fish/bin/activate
cd /data/work/fish-speech && python -m zibai tools.api_server:app --listen 127.0.0.1:8000
cd /data/work/fish-speech && python fish_speech/webui/app.py

查看GPU

nvidia-smi dmon