E-Ink News Daily

← Back to list

Block’s new voice model sounds eerily human

Block, Jack Dorsey's company, released Dessa, a new full-duplex voice model trained entirely on human speech that runs locally on a single NVIDIA GPU rack. The model reportedly captures natural human expressions like laughs, pauses, and cries without relying on voice tags or an external LLM for knowledge retrieval.

Background

Block (formerly Square) has been investing in AI research through its engineering team, positioning voice as the most natural interface for human-AI interaction. The announcement follows growing competition in the AI voice space from companies like OpenAI, ElevenLabs, and others.

Source
TechCrunch
Published
Oct 25, 2026 at 12:00 AM
Score
7.0 / 10