myshell-ai / OpenVoice

Instant voice cloning by MIT and MyShell.
https://research.myshell.ai/open-voice
MIT License
29.93k stars 2.95k forks source link

RuntimeError: cuFFT error: CUFFT_INTERNAL_ERROR #92

Open JIers001 opened 10 months ago

JIers001 commented 10 months ago

RuntimeError: cuFFT error: CUFFT_INTERNAL_ERROR My cuda is 11.8.
How can solve it if I don't want to reinstall my cuda? (Other virtual environments rely on cuda11.8.) More information: Traceback (most recent call last): File "/home/km/OpenVoice/OpenVoice-main/openvoice.py", line 33, in tone_color_converter.convert( File "/home/km/OpenVoice/OpenVoice-main/api.py", line 149, in convert spec = spectrogram_torch(y, hps.data.filter_length, File "/home/km/OpenVoice/OpenVoice-main/mel_processing.py", line 61, in spectrogram_torch spec = torch.stft( File "/home/km/miniconda3/envs/openvoice/lib/python3.9/site-packages/torch/functional.py", line 632, in stft return _VF.stft(input, n_fft, hop_length, win_length, window, # type: ignore[attr-defined] RuntimeError: cuFFT error: CUFFT_INTERNAL_ERROR

rlenain commented 10 months ago

Hi there -- I get the same issue

rlenain commented 10 months ago

Fixed with

pip install torch==2.0.1
pip install wavmark==0.0.3
JIers001 commented 10 months ago

Thanks. The error above was addressed. But a new issue arises. My cuda is 11.2 (11.8 is wrong). Cudnn is 8.1. Is it possible for me to fix this problem without reinstalling cuda and cudnn? Error information: RuntimeError: cuDNN version incompatibility: PyTorch was compiled against (8, 5, 0) but found runtime version (8, 1, 0). PyTorch already comes bundled with cuDNN. One option to resolving this error is to ensure PyTorch can find the bundled cuDNN.Looks like your LD_LIBRARY_PATH contains incompatible version of cudnnPlease either remove it from the path or install cudnn (8, 5, 0) Thanks again.

naseerfaheem commented 10 months ago

@rlenain I tried that, now it's complaining about Silero :

OSError Traceback (most recent call last) File ~/anaconda3/envs/openvoice/lib/python3.9/site-packages/whisper_timestamped/transcribe.py:1885, in get_vad_segments(audio, output_sample, min_speech_duration, min_silence_duration, dilatation, method) 1884 try: -> 1885 _silero_vad_model, utils = torch.hub.load(repo_or_dir=repo_or_dir, model="silero_vad", onnx=onnx, source=source) ... -> 1889 raise RuntimeError(f"Problem when installing silero with version {version}. Check versions here: https://github.com/snakers4/silero-vad/wiki/Version-history-and-Available-Models") from err 1890 finally: 1891 if need_folder_hack:

RuntimeError: Problem when installing silero with version None. Check versions here: https://github.com/snakers4/silero-vad/wiki/Version-history-and-Available-Models

jayavanth commented 10 months ago

This worked for me

conda install pytorch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 pytorch-cuda=12.1 -c pytorch -c nvidia

I have 12.1 installed so I just bumped the version