seamless_communication icon indicating copy to clipboard operation
seamless_communication copied to clipboard
verified publisher icon facebookresearch

Can the speech to speech translation preserve the original speaker's voice?

Open drishyakarki opened this issue 2 years ago • 5 comments

drishyakarki avatar Oct 19 '23 10:10 drishyakarki

This is really good option. I need this too.

laion01 avatar Oct 19 '23 12:10 laion01

+1

Lidong1211 avatar Oct 20 '23 08:10 Lidong1211

I was wondering if we could use the units extracted from audio_to_units.py for this. Is it possible?

drishyakarki avatar Oct 21 '23 01:10 drishyakarki

+1

Lenstack avatar Oct 28 '23 00:10 Lenstack

+1

masoudpz avatar Dec 01 '23 14:12 masoudpz

The SeamlessExpressive model released in November does preserve many properties of the source voice. I guess, it fulfills what this issue is asking about.

avidale avatar Mar 14 '24 15:03 avidale