Moshi is a speech-text foundation model and full-duplex spoken dialogue framework. It uses Mimi, a state-of-the-art streaming neural audio codec.
Cost / License
- Free
- Open Source (Apache-2.0)
Since
- Since 2024
Platforms
- Mac
- Linux
- Self-Hosted
- Rust
- Python
- Docker
- PyTorch
France
EU
