Spaces:
Running
error:samplerate must be specified?
I ran all the cells, but I get an error in the last cell
I also included the .raw file.
Inference starts and after a few seconds an error occurs. The error name is: TypeError: samplerate must be specified.
How can I fix this error?
The ‘./raw’ in the title of the cell is not a file.....It just indicates that the audio files should be placed in the raw folder under the current path. You should choose an audio file as the input audio instead of a .raw file.
I tried with a .wav file. I ran the inference, but the GPU usage reached 100% and the inference failed.
Error:torch.cuda.outofmemoryerror: cuda out of memory. tried to allocate 6.93 gib (gpu 0; 39.56 gib total capacity; 32.65 gib already allocated; 2.96 gib free; 34.75 gib reserved in total by pytorch) if reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. see documentation for memory management and pytorch_cuda_alloc_conf
I'm doing it on a colab pro so running out of memory doesn't make sense…
I tried with a .wav file. I ran the inference, but the GPU usage reached 100% and the inference failed.
Error:torch.cuda.outofmemoryerror: cuda out of memory. tried to allocate 6.93 gib (gpu 0; 39.56 gib total capacity; 32.65 gib already allocated; 2.96 gib free; 34.75 gib reserved in total by pytorch) if reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. see documentation for memory management and pytorch_cuda_alloc_conf
I'm doing it on a colab pro so running out of memory doesn't make sense…
The program will automatically search for non-silent segments of the audio and split the entire audio into multiple segments for separate inference. If there is a ‘cuda out of memory’ error, it means that a segment lasts too long. In this case, you need to split it into smaller segments manually. But usually, it’s not necessary. Did you directly upload a song with BGM as input audio?
There was no bgm, only vocals, but I think it's because it's a 4-minute audio. I split it into 2 minutes and uploaded it, and it works. Thank you!😀