🎙️VibeVoice
microsoft/VibeVoice · homepage ↗
Open-Source Frontier Voice AI
⭐ Hugely popular: 53k stars, gaining about 367 a week
View on GitHub ↗repo profile
momentum
durability
bus factor = how many people it takes to cover more than half the commits (6 months). 1 is a solo project; higher means the work is spread across a team. top-author share is the single busiest author's slice of those commits.
since we covered it
why it's a big deal
- Ships a single open-source family that covers both speech synthesis and transcription, so teams can build voice pipelines without stitching together separate closed vendors.
- Handles long-form audio in one pass, with TTS up to 90 minutes and ASR up to 60 minutes, which removes the chunk-and-stitch workarounds most tools require.
- Offers a 0.5B real-time variant with about 300ms first-audible latency for interactive use and a 7B ASR model spanning over 50 languages with diarization and timestamps.
under the hood
- Written in Python and built on a next-token diffusion framework, pairing a large language model for textual and dialogue context with a diffusion head for acoustic detail.
- Uses continuous acoustic and semantic speech tokenizers running at a 7.5 Hz frame rate, which keeps token counts low enough to model very long sequences.
- Ships three sized variants, ASR at 7B, TTS at 1.5B, and Realtime at 0.5B, and is released under the MIT license with a stated caution against commercial deployment without further testing.
Radar summary, generated from the project's public sources
star history
- PR#16 6k 2025-09-03
- now 53k + 47k since first covered
curve is sampled from GitHub's star history, plus our own daily readings since we covered it; the dashed stretch is before we first covered it, the solid line since. figures at coverage are the numbers we printed then (approx.), current count is live.
understory
Better known than its recent output, coasting a little on attention.
- output, commits & releases
- clout, star velocity
output = commits & releases; clout = star velocity, both 0 to 100 monthly indices; the gap where output runs above clout is the understory. The understory →
covered in
-
Long-Form Conversational TTS
similar projects
compare these →- 💡 PaddleOCR
Turn any PDF or image document into structured data for your AI. A powerful, lightweight OCR toolkit that bridges the gap between images/PDFs and LLMs. Supports 100+ languages.
88k ACTIVE - ⚡ Unsloth
Local UI to run and train LLMs and diffusion models, including Qwen3.8, Kimi K3, MiniMax-H3, Gemma 4, DeepSeek-V4, FLUX and more.
72k ACTIVE - 🎙️ Real-Time-Voice-Cloning
Clone a voice in 5 seconds to generate arbitrary speech in real-time
60k ACTIVE
comments
Sign in with GitHub to add your blip on VibeVoice.