Voila

oila is an open-source voice-language model family by Maitrix.org & labs for low-latency, emotionally rich AI voice role-play, ASR & TTS.

Voila is an open-source AI family designed for real-time, expressive voice interactions. Developed by Maitrix.org, it enables low-latency, emotionally rich role-play, speech recognition, and text-to-speech applications. Unlike traditional systems, Voila uses an end-to-end architecture for fluid, dynamic conversations with a response time of just 195 milliseconds—faster than human reaction. The model combines large language model reasoning with advanced acoustic modeling, allowing users to define speaker identity and tone through simple text instructions. It supports over a million pre-built voices and can create custom voices from short audio samples. Voila also serves as a unified solution for speech recognition, text-to-speech, and multilingual translation. Open-sourced for research, it aims to advance human-machine interaction through natural, persona-aware voice generation.