Brazilian portuguese dataset losed quality

#4
by Awestesr4er - opened

Hello! Since that i use this program to make her voice in my native language - Brazilian Portuguese, i saw that when the text generation completed today losed quality.

For Example, here's the original version using ThalitaMultilingualNeural and Hatsune Miku V2 Dataset that i made for an fictional vlog that i make only for personal use:

Now look this audio that completely lost quality:

so, can you restore the audio to the upper first?

thanks in advance.

Owner

Thanks for the report. Seems like the voice changed upstream... 😅
https://learn.microsoft.com/en-us/answers/questions/5771581/change-in-tts-voice
https://github.com/rany2/edge-tts/issues/461

Since reverting it seems impossible, I added a voice selection feature for now.

Sign up or log in to comment