search

ming024 / FastSpeech2

An implementation of Microsoft's "FastSpeech 2: Fast and High-Quality End-to-End Text to Speech"
MIT License
1.79k stars 528 forks source link

Pre-emphasis #103

Open inconnu11 opened 3 years ago

inconnu11 commented 3 years ago

Hi, in your code, you did not pre emphasis the wav before extracting the mel spectrogram?

CiE-XinYuChen commented 2 years ago

I do not understand this phoneme embedding.Did you find it or solve your problem?