Block's Jack Dorsey Introduces Dessa, a Speech-to-Speech AI Model Built to Sound Human

Block cofounder Jack Dorsey introduced Dessa on Friday, October 23, calling it "the most authentic full-duplex speech model ever built," according to TheStreet. "Talking with it feels like talking with a person," Dorsey wrote.
Dessa is a true speech-to-speech model, meaning audio goes in and audio comes out without being converted to text. It is also full-duplex, so it listens and talks at the same time. Inference runs faster than real time, and the model runs locally on a MacBook Pro M4 Max or better.
The model comes from a small AI research team at Block led by Ragavan Thurairatnam, founded by former members of Toronto deep learning startup Dessa. The team says it trained the model from scratch on human conversation only, with no books, web pages or pretrained models. It cites curiosity, better understanding between humans and AI, and safety as its reasons for building it.
The TheStreet writer, who listened to the clips on the Dessa page, said the model laughed mid-sentence, asked a speaker to move closer to the mic, and reacted to a surprise anniversary trip to Paris with what sounded like teary gratitude.