LaunchAI ModelsNovember 11, 2026

Fish Audio releases Live 3 full-duplex voice model

Read original source →marktechpost.com

Live 3 listens while it speaks, deciding to respond in as little as 186 ms for short backchannels like "mm-hmm." Turn-taking moves from hand-tuned pipeline rules into the speech model, trained on real human conversations.

2 sources

More stories today

Open the live feed