저지연 오디오 생성 LLVC 사용기

https://github.com/KoeAI/LLVC 써봄

CUDA 11.8 버전 쓰고 있어

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

VC 없어서 설치

https://wikidocs.net/168848 참조

FAIRSEQ 관련 오류

https://github.com/facebookresearch/fairseq/issues/2908

pip install editdistance 

설치

No CUDA runtime is found, using CUDA_HOME='C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8'

오류

멀쩡하게 있는데 안됨.

pip install torch==2.1.0+cu118 torchvision==0.16.0+cu118 torchaudio==2.1.0+cu118 --extra-index-url https://download.pytorch.org/whl/cu118

다시 설치

set DISTUTILS_USE_SDK=1

이게 문젠가 설정하고 재도전

Using cached ruamel.yaml.clib-0.2.8-cp311-cp311-win_amd64.whl (118 kB)

Building wheels for collected packages: fairseq

  Building wheel for fairseq (pyproject.toml) ... error

  error: subprocess-exited-with-error

  × Building wheel for fairseq (pyproject.toml) did not run successfully.

  │ exit code: 1

  ╰─> [1829 lines of output]

      No CUDA runtime is found, using CUDA_HOME='C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8'

.............................. 

          match = self._regex.search(version)

                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^

      TypeError: expected string or bytes-like object, got 'NoneType'

      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.

  ERROR: Failed building wheel for fairseq

Failed to build fairseq

ERROR: Could not build wheels for fairseq, which is required to install pyproject.toml-based projects

여기서 안넘어감.

pip install fairseq

로 그냥 설치

tensorboard

scipy

torchmetrics

speechbrain

auraloss

librosa

# fairseq @ git+https://github.com/One-sixth/fairseq.git

fairseq

numpy

# I think these are the changes needed for minimal_rvc to run

gradio==3.36.1

faiss-cpu==1.7.3

pyworld==0.3.2

soundfile==0.12.1

ffmpeg-python==0.2.0

pydub==0.25.1

torchcrepe==0.0.20

fairseq@git+https://github.com/One-sixth/fairseq.git

버전을 포기했더니 넘어는 감.