FastSpeech
FastSpeech copied to clipboard
The target is Log-Mel Spectrogram?
Hi,
I am wondering what are your parameters for extracting the mel spectrograms. And also, if you use the log-mel spectrogram as target for training the model?