Question for other voices....

[madey83]

Is it possible to use voices from this page:
http://openslr.org/94/

To train model and use it in piper, if so could it be possible to provide how to do that?

Many thanks for guidance 😃

[synesthesiam]

Yes, I've trained several voices from that dataset. Unfortunately, they perform very poorly on short sentences because the dataset comes from audio books (where short sentences are less common).

Most of the multi-speaker models I've attempted to train from the MLS dataset have also sounded very bad 🙁

[madey83]

Could it be possible for you to provide me step-by-step instruction how to train MPL Polish model with high quality?

I have found on internet some polish AI generated voice for some tv shows or movies that sound very good but do not know how these guys do it....