dataset
yunzqq opened this issue · comments
QiquanZhang commented
For dataset processing for training, all speech is set to a fixed length (corresponds to the parameter segment in code)?
Pariente Manuel commented
Yes, we extract random fixed segment from the mixtures most of the time.
QiquanZhang commented
Yes, we extract random fixed segment from the mixtures most of the time.
Thank you very much for your response.