search

dansuh17 / segan-pytorch

SEGAN pytorch implementation https://arxiv.org/abs/1703.09452
GNU General Public License v3.0
105 stars 32 forks source link

Improving other noisy data instead of Voice bank #28

Open asadullah797 opened 5 years ago

asadullah797 commented 5 years ago

Hi, I have used GAN for speech enhancement using AMI speech dataset but it does not improve the noisy far-field speech data samples. Although it works well on Voice bank dataset as mentioned in the paper. Can you please give any comments. Thanks

dansuh17 commented 4 years ago

Sorry for the late reply. Haven't experimented with AMI dataset. Perhaps the far-field speech data samples have low SNRs than VCTK's?

asadullah797 commented 4 years ago

Thank you for your reply. If I mix noises to the librispeech and consider librispeech as clean data and noisy librispeech as noisy data. Will this GAN work in that case?

dansuh17 commented 4 years ago

Yes. There is no specific reason for this model to not work for other datasets than VCTK, since it includes no VCTK-specific treatments to work.