GoKawiilBlock's Square unit unveils Dessa, a voice-native AI that skips text transcription
Block, led by Jack Dorsey, has released an AI model called Dessa that processes spoken language directly rather than converting speech to text first. The company says this approach lets the model respond to vocal tone and timing more naturally, similar to how infants learn language before learning to read or write.
GoKawiil's interpretation of the reporting above, not reported fact.
Most voice assistants today translate speech into text, which can strip out emotional cues like distress or urgency in a speaker's voice, according to Block. By building a model that reportedly processes audio natively, Block suggests it could better detect emotional states such as pain, potentially making voice AI more responsive in situations where tone carries critical meaning. This positions Block as pursuing a different technical path for voice interfaces, though it remains unclear how widely Dessa will be deployed or tested outside the company's own framing.
- Block has built an AI model called Dessa that works directly with speech rather than converting it to text first.
- The company frames the technology as addressing a 'safety problem' where voice AI fails to detect emotional distress in speech.
- Block compares the approach to how babies learn spoken language before learning written text.
Source: theverge.com — Jay Peters, 2026-10-23
Published there as: “Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.