VibeVoice

Open-source frontier voice AI from Microsoft. Family of models for text-to-speech (TTS), automatic speech recognition (ASR), and real-time streaming synthesis.

GitHub: microsoft/VibeVoice

Overview

VibeVoice is Microsoft's open-source voice AI platform offering three model sizes: ASR (7B) for transcription with speaker diarization, TTS (1.5B) for multi-speaker long-form synthesis, and Realtime (0.5B) for low-latency streaming. It supports 50+ languages and runs on consumer GPUs.

Instructions

Installation