# Dessa learns to speak without converting audio to text

Canonical source : [Dessa learns to speak without converting audio to text](<https://sebastien.rigaux.be/en/tech-watch/links/dessa-learns-to-speak-without-converting-audio-to-text>)

Unlike voice systems that turn speech into text before responding, Dessa processes audio directly, end to end. Trained only on human conversation, this speech-to-speech model runs full-duplex: it can listen and talk at once, interrupt, or punctuate an exchange with laughter and backchannels. Its demos also show how pauses and intonation convey meaning.

Inference is reported to run faster than real time, with local execution on a MacBook Pro M4 Max or better. This is still a research preview; the team is inviting AI safety researchers to assess the risks of a model that perceives emotion in speech.

- Selected on : 2026-10-28

- Entry updated : 2026-11-07T20:32:37+0000

- Format : Article

- Source language : EN

- Credit : Ragavan Thurairatnam

Original resource : [Dessa learns to speak without converting audio to text](<https://block.xyz/dessa>)

Tags : [AI](<https://sebastien.rigaux.be/en/tech-watch/tags/ai>)

Available versions : [Français](<https://sebastien.rigaux.be/veille/liens/dessa-un-modele-vocal-qui-apprend-sans-passer-par-le-texte>) · [English](<https://sebastien.rigaux.be/en/tech-watch/links/dessa-learns-to-speak-without-converting-audio-to-text>)

[Back to tech watch](<https://sebastien.rigaux.be/en/tech-watch>)