Block's new AI model Dessa was trained entirely on human speech to catch emotions text-based models miss
Block's new AI model Dessa was trained entirely on human speech to catch emotions text-based models miss
The company has launched an AI voice model Dessa, which is completely trained on human voice and does not use any text data, directly processing audio to capture emotions, sarcasm, laughter, and pauses in real-time, supporting full-duplex communication, and can run locally on MacBook Pro M4 Max, trained by a single NVIDIA GB200 NVL72, and is currently not publicly available, the team is recruiting AI safety researchers for testing.
Same story4 other outlets also reported this
How the accounts differ:The Verge mentions the model has 'occasionally stiff and strange' dialogue; The Decoder and TechCrunch do not mention this.
- The Verge · AI··First
Jack Dorsey’s Block releases Dessa AI model: a voice interaction model trained only on human dialogue
- The Decoder··+18 h 37 min·Reading now
The company has launched a new AI voice model Dessa, which is completely trained on human voice to capture emotions that text models may overlook
Adds:NVIDIA GB200 NVL72 used for training.
- TechCrunch · AI··+20 h 27 min
Block releases Dessa full-duplex speech model
Adds:Supports full-duplex communication.
- ITHome · AI··+30 h 57 min
Block releases full-duplex AI voice model Dessa: full human-like feel, supports local operation on Apple MacBook
Adds:Knowledge comes from real conversation data.
- MarkTechPost··+6 d
Block releases a full-duplex voice model Dessa trained solely on voice data
Adds:1 rack of 72 Blackwell GPUs used for training.
In order of publication · later reports are added as they appear · Compared by AI from each outlet’s title and summary
Block, the company behind Square and Cash App, has unveiled an AI speech model called Dessa.
Block says Dessa is the first AI model trained entirely on human speech, with no text data at all, built from scratch. Instead of converting audio to text first, Dessa processes speech directly as audio. That lets it pick up on emotions, sarcasm, laughter, and pauses in real time.
The model supports full-duplex communication, so it can listen and talk at the same time. It runs locally on a MacBook Pro M4 Max and was trained on a single NVIDIA GB200 NVL72. Block says it built Dessa out of curiosity, a desire for better human-AI communication, and safety concerns, since current voice AI ignores emotional cues. Dessa isn't publicly available yet. The team is first looking for AI security researchers to test the model for unexpected risks. Anyone interested can sign up via a form.
Earlier attempts at emotional voice AI haven't stuck
Several teams have tried to build voice AI that sounds emotionally rich. In March 2024, the startup Hume AI launched EVI (Empathic Voice Interface), a voice AI with "emotional intelligence." Kyutai followed in September 2024 with Moshi. LAION's Open Empathic project has been working since late 2023 on training empathetic speech synthesis models using an open-source audio dataset. OpenAI overhauled ChatGPT's "Advanced Voice Mode" in June 2025 and has continued refining it since then, adding better intonation, pausing, and more realistic expressions of empathy and sarcasm.
But sounding emotional isn't the same as being emotional or intelligent, even if it creates that impression. For advanced speech AI, capabilities like reasoning while responding matter more than how natural the voice sounds, because they actually affect the quality of the answer.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
Source: The Decoder · the-decoder.com