
Moonshine
Speech to text, intent, and speech back, with the pause kept short.
Content & MediaJul 20, 2026
- Category
- Content & Media
- Website
- github.com
- Language
- English
- Listed
- Jul 20, 2026
- Stars
- 11,104
- Forks
- 614
moonshine-ai/moonshine ↗, C++, Last commit 18d ago
Moonshine is built for developers wiring up voice agents, and latency is the axis it optimises. Speech to text, intent recognition and text to speech are all provided with very low latency, so the gap between a person finishing a sentence and the system reacting stays short. That gap is what makes a virtual assistant or a support bot feel wrong once it grows, which is why the whole loop rather than one stage is covered. Three pieces on one path: hear, interpret, answer.
Like this product?