audio
audio copied to clipboard
Data manipulation and transformation for audio signal processing, powered by PyTorch
CUDA 13 change library and clean code of redundancy. This fix compile on Nvidia 5090 with CUDA 13.0 PLEASE NOTE THAT THE TORCHAUDIO REPOSITORY IS NO LONGER ACTIVELY MONITORED. You...
in CUDA 13 Nvidia clean duplicate function. Need to use standard function. This patch run for make a whl pip package. PLEASE NOTE THAT THE TORCHAUDIO REPOSITORY IS NO LONGER...
In the ComfyUI project , and other projects which are dependent on pytorch/audio, there's a failure to load ffmpeg support due to failures when attempting to load libtorio libraries. Namely...
* Eliminates dependence on `/opt/rocm` and path based heuristics. * Normalizes package finding for Rocm 6.5+ layout. * Workaround cmake >= 4.0 for hiprtc This will help with windows support....
### 🐛 Describe the bug While using torchaudio.functional.bandreject_biquad with cuda tensors for waveform input, I am getting this error ``` RuntimeError: Expected (in.dtype() == torch::kFloat32 || in.dtype() == torch::kFloat64) &&...