Created by: ssteven502tw

Is Baidu Internal English Dataset clean voice? Do you use the following methods in paper? (Additionally, we dynamically augment the dataset by adding unique noise every epoch with an SNR between 0dB and 30dB)