Block Releases Dessa Speech-Only Full-Duplex AI Voice Model Trained on Human Conversation Recordings
What happened
On October 23, 2026, Jack Dorsey's Block introduced Dessa, a full-duplex speech-to-speech AI model trained entirely on human voice recordings without text tokens or pretrained base models.
The Context
* Zero text reliance: Dessa skips traditional speech-to-text pipelines entirely, learning conversational cadence and emotional distress cues the way infants acquire language before reading. * True bidirectional flow: The architecture processes simultaneous listening and speaking in real time, handling natural interruptions, pauses, and voice inflections as they occur. * Local operational viability: Early reports demonstrate the system running local inference on Apple M4 Max hardware, opening possibilities for private, on-device voice processing.
The Number
NVIDIA GB200 NVL72 rack trained the entire foundational model, demonstrating that lean computing footprints can bypass standard multi-billion-token text datasets to yield functional conversational intelligence.
Takeaway
As Block restricts early deployment to AI safety researchers via waitlist, expect enterprise voice tooling to pivot away from delayed transcription layers toward direct-audio foundation models.
Curated tech headlines including AI stories.
"Influential aggregator surfacing the day’s top tech/AI links."
— A47 Editor
Block unveils Dessa, a "true speech-to-speech" AI model trained on "nothing but human conversation" that it says runs locally on a MacBook Pro M4 Max or newer (Jay Peters/The Verge)
Block has introduced Dessa, a new AI model designed for speech-to-speech interactions, which has been trained exclusively on human conversations. This model is capable of running locally on a MacBook Pro M4 Max or newer, showcasing advancements in AI...
Daily AI news: models, tools, and policy.
"Independent outlet tracking the fast pace of AI."
— A47 Editor
Block's new AI model Dessa was trained entirely on human speech to catch emotions text-based models miss
Block has unveiled Dessa, a new AI model that processes audio directly to recognize emotions, sarcasm, and pauses in real time, trained entirely on human speech. This model operates locally on a MacBook Pro M4 Max but is not yet publicly available.
Research, news, and analysis on blockchain startups, DeFi, and regulations.
"Crypto Briefing provides research, news, and analysis on blockchain startups, DeFi, and crypto regulations with investor-focused coverage."
— A47 Editor
Jack Dorsey’s Block unveils Dessa, an AI voice model trained entirely on human conversations
Jack Dorsey's Block has introduced Dessa, an AI voice model that has been trained exclusively on human conversations, showcasing advancements in artificial intelligence aimed at enhancing human-like interactions. This development raises significant e...
Consumer tech and culture with frequent AI coverage.
"Influential tech outlet covering AI products and policy."
— A47 Editor
Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend
Jack Dorsey's Block has introduced an AI model named Dessa, which is designed to engage in conversations that mimic the awkwardness of human interactions. The model has been trained exclusively on human dialogue, showcasing its capabilities through s...
Tech news, reviews, and analysis of consumer electronics, science, art, and culture.
"The Verge is a technology-focused media outlet known for in-depth reporting, product reviews, and coverage of the intersection between technology and culture."
— A47 Editor
Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend
Jack Dorsey's Block has introduced an AI model named Dessa, which is designed to engage in conversations that mimic the awkwardness of human interactions. The model has been trained exclusively on human dialogue, showcasing its capabilities through s...