search

facebookresearch / svoice

We provide a PyTorch implementation of the paper Voice Separation with an Unknown Number of Multiple Speakers In which, we present a new method for separating a mixed audio sequence, in which multiple voices speak simultaneously. The new method employs gated neural networks that are trained to separate the voices at multiple processing steps, while maintaining the speaker in each output channel fixed. A different model is trained for every number of possible speakers, and the model with the largest number of speakers is employed to select the actual number of speakers in a given sample. Our method greatly outperforms the current state of the art, which, as we show, is not competitive for more than two speakers.
Other
1.23k stars 178 forks source link

sisnr nan #76

Closed srdfjy closed 1 year ago

srdfjy commented 2 years ago

Hi, I have the following problem while training 2 or 3 speaker models，generated samples using make_dataset.py, parallelized the generation process, snr(15-30).

This problem does not occur in small samples，what could be the possible reasons for this problem?

out_sig = out_sig - np.mean(out_sig, axis=1).reshape(B, 1) 2022/06/17 09:52:23 /svoice/svoice/evaluate.py:135: RuntimeWarning: invalid value encountered in subtract Summary | Epoch 1 | Train 7.01209 | Valid 10.36696 | Best 10.36696 | Sisnr nan | Pesq 0.00000 | Stoi 0.00000

srdfjy commented 2 years ago

Hi @adiyoss ，What could be the possible cause of this problem?

If the created samples are all 4 seconds long and the separated wav is greater than 4 seconds, will it matter?

Abhishek577 commented 5 days ago

hi @srdfjy Is this problem solved..What is the possible soltuion