CosyVoice2 install on windows 10

[PaulYBChiang]

first of all, I will explain my current approach on how to install CosyVoice2 on Windows10 as the official requirement file is targeting Linux. then I will show the problems.

To install CosyVoice2 for windows, we need to install TensorRT for Windows first:

• visit https://developer.nvidia.com/nvidia-tensorrt-8x-download
• Download "TensorRT 8.6 GA for Windows 10 and CUDA 11.0, 11.1, 11.2, 11.3, 11.4, 11.5, 11.6, 11.7 and 11.8 ZIP Package"
• Extract TensorRT-8.6.1.6\onnx_graphsurgeon\onnx_graphsurgeon-0.3.12-py2.py3-none-any.whl
• pip install onnx_graphsurgeon-0.3.12-py2.py3-none-any.whl

requirement.txt for windows:
--extra-index-url https://download.pytorch.org/whl/cu121
--extra-index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
conformer==0.3.2
deepspeed>=0.14.2
diffusers==0.29.0
gdown==5.1.0
gradio==5.4.0
grpcio==1.57.0
grpcio-tools==1.57.0
hydra-core==1.3.2
HyperPyYAML==1.2.2
inflect==7.3.1
librosa==0.10.2
lightning==2.2.4
matplotlib==3.7.5
modelscope==1.15.0
networkx==3.1
omegaconf==2.3.0
onnx==1.16.0
onnxruntime==1.18.0; sys_platform == 'win32' # Use ONNX Runtime for Windows
onnxruntime-gpu==1.18.0; sys_platform == 'win32' # Use ONNX Runtime GPU for Windows
openai-whisper==20231117
protobuf==4.25.0
pydantic==2.7.0
pyworld==0.3.4
rich==13.7.1
soundfile==0.12.1
tensorboard==2.14.0

TensorRT dependencies (manual installation required):

torch==2.3.1
torchaudio==2.3.1
transformers==4.40.1
uvicorn==0.30.0
wget==3.2
fastapi>=0.115.6
fastapi-cli>=0.0.0
WeTextProcessing==1.0.0

With the above two changes, I managed to install the CosyVoice2 on windows 10 and the sample py script works, However I got the following warning message:

miniconda3\envs\cosyvoice2\lib\site-packages\onnxruntime\capi\onnxruntime_inference_collection.py:69:
UserWarning: Specified provider 'CUDAExecutionProvider' is not in available provider names.
Available providers: 'AzureExecutionProvider, CPUExecutionProvider'

[aluminumbox]

tensorrt is not needed if you only want to use pytorch inference