CorentinJ / Real-Time-Voice-Cloning

Clone a voice in 5 seconds to generate arbitrary speech in real-time
Other
52.28k stars 8.75k forks source link

Synthesizer alignment #1053

Open Alterbort opened 2 years ago

Alterbort commented 2 years ago

Hi, when I followed the steps inside the training guide to train the code, I found some problems with the alignment of the synthesizer. The dataset I used was LibriSpeech and I followed the instructions exactly for each step and also downloaded the alignment file. I don't know what's wrong, any help would be greatly appreciated. Looking forward to your reply : tt tt1

raccoonML commented 2 years ago

It's possible this will be resolved with more training steps. You can also try restarting the training with a higher reduction factor to make it easier to learn attention. Once learned, the reduction factor can be decreased.