Request to add StyleTTS 2 model

[Asten-Valentinus]

Looking around, I've discovered the StyleTTS 2 model.

It seems to be of a much higher quality than other Voice Cloning TTS models, so I think it would be nice to support.

From the statistics mentioned in it's page, it seems to be a little more optimised than YourTTS, which is already supported by SpeechNote.

Alike to what SpeechNote does, it seems to work well when generated on a sentence-by-sentence basis. However, the quality degrades on smaller segments of text.

There is a sweet spot to maximize quality.

You can find the source here.

Thanks for the awesome program!

[mkiol]

Thanks for the recommendation. Adding to my TO-DO list.

BTW, Have you tested the Coqui XTTS model? In my opinion it is great, much better than YourTTS. Especially in version 2.0.2.

[Asten-Valentinus]

Thanks for the recommendation. Adding to my TO-DO list.

BTW, Have you tested the Coqui XTTS model? In my opinion it is great, much better than YourTTS. Especially in version 2.0.2.

Yeah, I've tried it. It is certainly much better! I think Style TTS is higher quality than that though, the demos show it as pretty good.