# Moonshine Voice

> Open source on-device speech to text and text to speech

Steal this: Live on-device transcription that starts working while the person is still talking.

- Category: Agents & dev (`agents`)
- Tags: `open-source`
- Official: https://moonshine.ai
- GitHub: https://github.com/moonshine-ai/moonshine
- Wunderlust: https://wunderlust.xyz/r/moonshine-voice
- Image: https://wunderlust.xyz/media/moonshine-voice-poster.jpg
- Video preview: https://wunderlust.xyz/media/moonshine-voice.webm
- Added: 2026-11-10

## What it is

An open source toolkit for building real-time voice agents. It transcribes speech and generates voice on the device, with no account or API key, and one library for Python, JavaScript, iOS, Android and desktop.

## What it’s for

For adding voice to an app or agent without sending audio to the cloud.

## Benefits

- Runs on the device, so it stays private
- One library across many platforms
- Models from 1 MB up to higher accuracy

## Caveats

- According to its maker, its models beat Whisper Large V3 on accuracy
- Code and models are MIT; only legacy non-English models carry a non-commercial license

## Similar on Wunderlust

- [Morphic](https://wunderlust.xyz/r/morphic.md): Open source AI search engine with cited answers and a generated interface
- [LitServe](https://wunderlust.xyz/r/litserve.md): Python framework for serving AI models as APIs with batching and scaling
- [Postiz](https://wunderlust.xyz/r/postiz.md): Open-source agentic scheduling tool for social media posts
- [Ternary Bonsai 2](https://wunderlust.xyz/r/ternary-bonsai-2.md): Family of 2B, 4B and 9B ternary language models for local use
- [Ollama](https://wunderlust.xyz/r/ollama.md): Runtime for running language models on your own machine
- [Paperclip](https://wunderlust.xyz/r/paperclip.md): Control center for agents organized like a company