Miso One is an 8B-parameter text-to-speech system developed by Miso Labs. It is designed for expressive English conversational speech, voice continuation, and low-latency voice-agent research.
Key Features
- 8B Parameter Model: An open-weights text-to-speech system available for local inference.
- Expressive Speech: Optimized for conversational delivery, emotion, and pacing in English.
- Audio Context Support: Capable of conditioning on prompt audio for voice continuation and one-shot cloning.
- Low-Latency Performance: Designed for voice-agent workflows with a published 110 ms latency benchmark.
- Discrete Audio Coding: Utilizes Mimi audio codes for speech generation.
