Block, the financial technology company led by Jack Dorsey, has released a voice model that laughs, pauses and cries in ways that sound uncannily human. The model, called Dessa, came out of the company’s research lab, and Block says it runs locally on a consumer-grade computer.
Block said Dessa is trained entirely on human speech and a single NVIDIA GPU rack. It is a full-duplex model, which means it can talk and listen at the same time. OpenAI’s GPT-Live model, released earlier this year, works the same way. Most speech models are paired with a large language model to retrieve knowledge. Block said Dessa draws its knowledge from the conversational data it was trained on.
The company argues that today’s voice systems often sound unnatural because they convert speech into text before generating a reply. Many models, including ElevenLabs v4 and Fish Audio Drama 3, use voice tags to control emotion. Block says emotion is built into Dessa itself. It is not clear whether any tags steer the model’s expressiveness.
“Interacting with current voice AI can just plain suck,” the company said in a blog post. “It turns your voice into text and throws away an incredible amount of information text alone can’t carry. The most powerful AIs in the world can’t hear how you feel. We think that’s ridiculous, leads to a worse experience, and is potentially dangerous.”
To show the range of the model, Block’s researchers posted several conversations in which the agent used different voices, cadences and emotions. In one, the bot said it was tired from work, and it sounded tired. In another, it talked about the person who was making it comfortable, and its voice grew emotional, nearly breaking down. A third clip shows a Block team member playing a game with the voice assistant.
Dessa is not publicly available. Block has asked safety researchers to test the model, a process likely to focus on scenarios in which a voice agent could trick people into believing they are speaking with a human.
The model fits into Block’s broader business. The company is in the finance industry and sells its products to small businesses. Square already uses voice AI to handle orders. Block says an on-device voice model could support more business and consumer applications at a lower cost.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.