voice conversion with XTTS_V2

[etlweather]

I could be completely wrong, but from what I read all over, I should be able to do voice-to-voice conversion with other models than freevc24.

I managed to make it work with your_tts by changing the command line parameters. Instead of using source_wav and target_wav, I used reference_wav and speaker_wav and added language_idx.

But this does not work with the XTTS_V2 model.

What am I doing wrong? Or is this not even possible?

[eginhard]

No, this is not possible. XTTS only supports voice cloning, i.e. converting text to the voice of a given target speaker, but not voice-to-voice conversion. But we're planning to add 1-2 other voice conversion models soon.