Open tilmankamp opened 7 years ago
Training samples should be augmented by different reverberation environments.
Should look at Kaldi examples of doing this[1][2]
Training samples should be augmented by different reverberation environments.