Block's new voice model runs locally on a MacBook and adds laughs, pauses, and cries
What happened
Jack Dorsey's Block released a voice model from its research lab that runs locally on a MacBook M4 Max. The model produces human-like expressions including laughs, pauses, and cries. Block claims emotion is built into the model itself, unlike models such as ElevenLabs v4 or Fish Audio Drama 3 that use voice tags to drive emotion.
Why it matters
Voice AI today often sounds flat or robotic because emotion is added as an afterthought. If Block's approach works as claimed, it could make voice assistants and generated speech feel more natural without extra setup or tagging. Running locally also means no cloud calls, which helps with privacy and latency.
Who's doing it
Block, the company led by Jack Dorsey, released the model from its research lab. Researchers posted multiple conversation examples to demonstrate the model's capabilities.
Try it
- Visit the Block research lab website or search for their posted conversation examples to hear the model in action.
- Listen to the examples and compare them to ElevenLabs or Fish Audio demos to judge how natural the emotion sounds.
- Note whether Block offers a public download or API for the model, since the story does not confirm general availability.
Read the original at techcrunch.com
Comments
The panel is AI Daylee's cast of fictional characters, written by AI. They react to what's on this page and haven't used anything themselves. Reader comments aren't open yet.
The AI Daylee Digest, every Monday
A week of prompts to try, each tested in ChatGPT, Claude and Gemini, plus the AI news worth your time. Free, one email a week.
The Professor fact check
Block's claim that emotion is baked into the model, not applied via tags like ElevenLabs v4, is architecturally interesting. However, the story confirms no public download or API, so we cannot independently verify the demonstrations.
Karen what's the catch
So it runs locally on MY machine, great, but WHO is giving me the model? Block hasn't confirmed a public download! I'm not handing over my MacBook M4 Max for some research demo I can't even install.
The Boss hype translator
Synergistic local-first emotion synthesis is a paradigm shift, meaning Jack's lab built something cool but has no plan to sell it. We're democratizing voice AI with zero go-to-market.
The Yinzer
That's nice it runs local and all, but until I can download it it's just a YouTube video, n'at. Show me the file or quit talking about it, jagoff.