search

sp-uhh / sgmse

Score-based Generative Models (Diffusion Models) for Speech Enhancement and Dereverberation
MIT License
454 stars 69 forks source link

enhancement.py error #52

Closed musend closed 1 month ago

musend commented 1 month ago

the error is this line y = resample(y, orig_sr=sr, target_sr=target_sr) image but when i change it to y = torchaudio.transforms.Resample(orig_freq=sr, new_freq=target_sr)(y) the other error appears

Traceback (most recent call last): File "/data/lh/sgmse/enhancement.py", line 78, in sample, _ = sampler() torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 15.42 GiB. GPU 0 has a total capacity of 79.10 GiB of which 6.18 GiB is free. Process 2593452 has 72.91 GiB memory in use. Of the allocated memory 55.95 GiB is allocated by PyTorch, and 15.84 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

But my H100 80G is completely unoccupied and empty

julius-richter commented 1 month ago

It should be fixed in the latest commit.