search

yuhogun0908 / MISOnet

Unofficial Multi-microphone complex spectral mapping for utterance-wise and continuous speech separation(MISO-BF-MISO)
MIT License
51 stars 9 forks source link

RIR #8

Open yunzqq opened 1 year ago

yunzqq commented 1 year ago

Where is the RIR convolution operation for speech generation?

yuhogun0908 commented 1 year ago

The RIR convolution operation worked in the same way as the dataset of the existing paper.

If you want to do this, plz visit https://github.com/fgnt/sms_wsj.

yunzqq commented 1 year ago

The RIR convolution operation worked in the same way as the dataset of the existing paper.

If you want to do this, plz visit https://github.com/fgnt/sms_wsj.

Thank you! And I also want to know that is there compression technique is used for complex spectrum for training ?

yuhogun0908 commented 1 year ago

I'm glad it helped! We didn't use the compression technique such as using Mel-filter bank

yunzqq commented 1 year ago

I'm glad it helped! We didn't use the compression technique such as using Mel-filter bank

I want to consult that if are there any normalization technique is used for processing audio files?