Jack Dorsey's Block unveils Dessa, a full-duplex voice AI trained only on human speech

What happened? Block announced Dessa on 23 October 2026, a speech-to-speech AI model that its makers say was trained from scratch on nothing but recorded human conversation: no text and no pretrained models. Co-founder and chairman Jack Dorsey called it "the most authentic full-duplex speech model ever built" in a post viewed more than 6.7 million times.

How it works: Dessa takes audio in and gives audio out, with no transcription step in between. It is full-duplex, meaning it listens and talks at the same time, so it can laugh mid-sentence, cut in or backchannel ("yeah", "got it") the way people do. Block says inference runs faster than real time and the model runs locally on a MacBook Pro M4 Max or better. Its argument is that today's voice assistants turn speech into text and lose emotion, timing and tone, which it calls both a worse experience and a safety problem.

Who built it? A very small research lab inside Block, founded by former members of Dessa, the Toronto deep learning startup, and led by Ragavan Thurairatnam. The team says it has built generative speech models for more than eight years and reports directly to Dorsey.

Caveats: Dessa is a research preview behind a waitlist, with early access for a small group of AI safety researchers. Block has not published a parameter count, API, pricing, weights or benchmark results, so the evidence so far is its own demo recordings, including one conversation with OpenAI's GPT-Live-1. The Verge found the demos mostly natural but occasionally awkward; in one, the model mistakes the speaker for someone named "Ronaldo".

Sources: Block: Dessa · Jack Dorsey on X · The Verge · The Indian Express

Image credit: Block