Open jacobtang opened 2 weeks ago
Truncated audio with XTTS can sometimes be a training issue. I think it's also dependent on the reference wave file, if I use voice references that contain pauses I feel this occurs more often.
Elevenlabs does support realtime voice cloning. You can add voices with their api and then realtime synthesize text with the added voice either with elevenlabs python library or with RealtimeTTS.
Hi @KoljaB When using XTTS for real-time voice cloning, the voice sounds unnatural, and some parts of the text are always ignored,could you please let me know if there are any other solutions that support real-time voice cloning, such as ElevenLabs? Thank you.