Added CUDA 12.8 support for RTX 5000 series

README_dev_en.md

@@ -20,22 +20,43 @@ $ conda activate vcclient-dev
 
 ```
 $ git clone https://github.com/w-okada/voice-changer.git
+$ cd voice-changer
 ```
 
 ## For Server Developer
 
 1. Install requirements
 
+1-1. For No GPU
+
+```
+$ python -m pip install -r server/requirements_cpuonly.txt
+```
+
+1-2. For NVIDIA GPUs
+
+Please rewrite the `cu128` part in `server/requirements_nvidia.txt` according to your CUDA version.
+The default `cu128` is for CUDA 12.8.
+
+```
+--extra-index-url https://download.pytorch.org/whl/cu128
+
+torch==2.7.0+cu128
+torchaudio==2.7.0+cu128
+```
+
+Next, execute the following:
+
 ```
-$ cd voice-changer/server
-$ pip install -r requirements.txt
+$ python -m pip install -r server/requirements_nvidia.txt
 ```
 
 2. Run server
 
 Run server with the below command. You can replace the path to each weight.
 
 ```
+$ cd server
 $ python3 MMVCServerSIO.py -p 18888 --https true \
     --content_vec_500 pretrain/checkpoint_best_legacy_500.pt  \
     --content_vec_500_onnx pretrain/content_vec_500.onnx \

README_dev_ja.md

@@ -20,22 +20,43 @@ $ conda activate vcclient-dev
 
 ```
 $ git clone https://github.com/w-okada/voice-changer.git
+$ cd voice-changer
 ```
 
 ## サーバ開発者向け
 
 1. モジュールをインストールする
 
+1-1. GPUなしの場合
+
+```
+$ python -m pip install -r server/requirements_cpuonly.txt
+```
+
+1-2. NVIDIAのGPUを利用する場合
+
+server/requirements_nvidia.txtをCUDAのバージョンに応じて以下のcu128の部分を適宜書き換えてください。
+デフォルトのcu128場合はCUDA12.8向けです。
+
+```
+--extra-index-url https://download.pytorch.org/whl/cu128
+
+torch==2.7.0+cu128
+torchaudio==2.7.0+cu128
+```
+
+次に以下を実行します
+
 ```
-$ cd voice-changer/server
-$ pip install -r requirements.txt
+$ python -m pip install -r server/requirements_nvidia.txt
 ```
 
 2. サーバを起動する
 
 次のコマンドで起動します。各種重みについてのパスは環境に合わせて変えてください。
 
 ```
+$ cd server
 $ python3 MMVCServerSIO.py -p 18888 --https true \
     --content_vec_500 pretrain/checkpoint_best_legacy_500.pt  \
     --content_vec_500_onnx pretrain/content_vec_500.onnx \

README_dev_ko.md

@@ -26,9 +26,28 @@ $ git clone https://github.com/w-okada/voice-changer.git
 
 1. 모듈을 설치한다
 
+1-1. GPU가 없는 경우
+
+```
+$ python -m pip install -r server/requirements_cpuonly.txt
+```
+
+1-2. NVIDIA GPU를 사용하는 경우
+
+`server/requirements_nvidia.txt` 파일에서 CUDA 버전에 따라 아래 `cu128` 부분을 적절히 수정하십시오.
+기본 `cu128`은 CUDA 12.8용입니다.
+
+```
+--extra-index-url https://download.pytorch.org/whl/cu128
+
+torch==2.7.0+cu128
+torchaudio==2.7.0+cu128
+```
+
+다음을 실행하십시오:
+
 ```
-$ cd voice-changer/server
-$ pip install -r requirements.txt
+$ python -m pip install -r server/requirements_nvidia.txt
 ```
 
 2. 서버를 구동한다

README_dev_ru.md

@@ -28,9 +28,28 @@ $ git clone https://github.com/w-okada/voice-changer.git
 
 1. Установите необходимые зависимости:
 
+1-1. Для систем без GPU
+
+```
+$ python -m pip install -r server/requirements_cpuonly.txt
+```
+
+1-2. Для систем с GPU NVIDIA
+
+Пожалуйста, перепишите часть `cu128` в файле `server/requirements_nvidia.txt` в соответствии с вашей версией CUDA.
+Значение по умолчанию `cu128` предназначено для CUDA 12.8.
+
+```
+--extra-index-url https://download.pytorch.org/whl/cu128
+
+torch==2.7.0+cu128
+torchaudio==2.7.0+cu128
+```
+
+Затем выполните следующую команду:
+
 ```
-$ cd voice-changer/server
-$ pip install -r requirements.txt
+$ python -m pip install -r server/requirements_nvidia.txt
 ```
 
 2. Запустите сервер

server/requirements.txt

@@ -3,28 +3,29 @@
 # # wget https://repo.anaconda.com/archive/Anaconda3-2022.10-Linux-x86_64.sh
 # # bash Anaconda3-2022.10-Linux-x86_64.sh
 
-uvicorn==0.21.1
-pyOpenSSL==23.1.1
-numpy==1.23.5
-torch==2.0.1
-torchaudio==2.0.2
-resampy==0.4.2
-python-socketio==5.8.0
-fastapi==0.95.1
-python-multipart==0.0.6
-onnxruntime-gpu==1.13.1
-scipy==1.10.1
-matplotlib==3.7.1
-websockets==11.0.2
-faiss-cpu==1.7.3
-torchcrepe==0.0.18
-librosa==0.9.1
+# fairseq
+git+https://github.com/mallocfree009/fairseq.git@dev_mf1
+
+uvicorn==0.34.2
+pyOpenSSL==25.0.0
+numpy==2.1.2
+resampy==0.4.3
+python-socketio==5.13.0
+fastapi==0.115.12
+python-multipart==0.0.20
+onnxruntime-gpu==1.22.0
+scipy==1.15.3
+matplotlib==3.10.3
+websockets==15.0.1
+faiss-cpu==1.11.0
+torchcrepe==0.0.23
+librosa==0.11.0
 gin==0.1.6
-gin_config==0.5.0
-einops==0.6.0
-local_attention==1.8.5
-websockets==11.0.2
-sounddevice==0.4.6
-dataclasses_json==0.5.7
-onnxsim==0.4.28
-torchfcpe==0.0.3
+gin-config==0.5.0
+einops==0.8.1
+local-attention==1.11.1
+sounddevice==0.5.1
+dataclasses-json==0.6.7
+onnxsim==0.4.36
+torchfcpe==0.0.4
+pyworld==0.3.5

server/requirements_cpuonly.txt

@@ -0,0 +1,4 @@
+torch==2.7.0
+torchaudio==2.7.0
+
+-r requirements.txt

server/requirements_nvidia.txt

@@ -0,0 +1,12 @@
+# As of 2025/05/17, if you do not install typing-extensions first, 
+# an error will occur when installing torch for NVIDIA.
+typing-extensions==4.13.2
+
+# Please set the URL and package version according to your CUDA version environment. 
+# The one listed here is for CUDA 12.8.
+--extra-index-url https://download.pytorch.org/whl/cu128
+
+torch==2.7.0+cu128
+torchaudio==2.7.0+cu128
+
+-r requirements.txt