204 likesNew · 7d old
Open-source TTS model runs real-time on phones and browsers
What is ZONOS3-Mini?
ZONOS3-Mini is a 100-million-parameter text-to-speech model that can generate natural speech and clone voices. It runs locally on laptops, phones or in the browser via WebGPU, supporting 43 languages and fine-grained control over emotion and pacing.
Who it is for
Developers building games, voice assistants or narration tools can add on-device speech without a cloud service.
How to try it
Open the Zonos Voice Studio demo in a WebGPU-enabled browser.
The story so far
Early Signal first spotted Zyphra/ZONOS3-Mini on November 11, 2026, with 204 likes. On its busiest day it gained 68 likes. The project is 7 days old. Early Signal files it under Voice.
Questions about ZONOS3-Mini
- How do I try ZONOS3-Mini?
- Open the Zonos Voice Studio demo in a WebGPU-enabled browser.
- What do I need to run ZONOS3-Mini?
- WebGPU-compatible browser (e.g., Chrome Canary)