Block's new voice model sounds eerily human
Block said that the voice model can run locally on a MacBook M4 Max
TL;DR
- Block's research lab has released a new voice model named Dessa.
- Dessa runs locally and can mimic human emotions like laughs, pauses, and cries.
- It is a full-duplex model, capable of talking and listening concurrently.
- The model is trained on human speech and its knowledge comes from conversational data.
- Block claims Dessa's emotion is built-in, unlike models that use voice tags.
- The company believes current voice AI is limited because it converts speech to text, losing emotional information.
- Dessa can run on consumer-grade computers like a MacBook M4 Max or better.
- Block is seeking safety researchers to test the model, potentially for misuse cases.
- Block's existing business includes using voice AI for services like order taking.