VibeVoice

Voice-powered AI companion - Atlas demo target project

Open-Source Frontier Voice AI

We introduced VIBEVOICE, a novel framework for long-form and multi-speaker speech generation. By integrating efficient hybrid speech representations from specialized ultra-low frame rate (7.5 Hz) acoustic and semantic tokenizers with an end-to-end LLM-based next-token diffusion framework, VIBEVOICE achieves state-of-the-art performance.

Python / Open-Source Frontier Voice AI