Low volume vocal segments are suppressed

Audio-WestlakeU / FullSubNet

PyTorch implementation of "FullSubNet: A Full-Band and Sub-Band Fusion Model for Real-Time Single-Channel Speech Enhancement."

https://fullsubnet.readthedocs.io/en/latest/

MIT License

553 stars 157 forks source link

Low volume vocal segments are suppressed #63

Open SEMLLYCAT opened 1 year ago

SEMLLYCAT commented 1 year ago

Dear author, when I tried to optimize the model, I found a phenomenon that when the model processing includes a small volume of human voice segment, the segment will be suppressed or even completely removed. I'm not sure if this is related to the model or target_dB_FS, or do you have a suggested solution? . I look forward to your reply. Thank you

haoxiangsnr commented 1 year ago

Hi, I don't have a very satisfactory suggestion, you can try to include clips with changing volume when simulation.

SEMLLYCAT commented 1 year ago

Ok, I am trying. Thank you very much for your reply