Is it possible RealtimeTTS support VITS?

KoljaB / RealtimeTTS

Converts text to speech in realtime

1.39k stars 119 forks source link

Is it possible RealtimeTTS support VITS? #92

Closed Rares9999 closed 2 weeks ago

Rares9999 commented 2 weeks ago

Hi, RealtimeTTS running coqui with XTTS, is that possible running coqui with VITS, with streaming features?

If it is possible, how to implement VITS?

KoljaB commented 2 weeks ago

Unlike XTTS VITS does not offer streaming inference out of the box. While it would be possible to offer sentence by sentence inference the latency would not be great. I aim for sub 1 second latency for an engine to be integrated into RealtimeTTS. VITS would very probably not achieve this without input streaming, so I feel it isn't worth the amount of work I would need to put into this.