# microsoft/VibeVoice

Canonical: https://openpick.ing/r/vibevoice
As of: 2026-11-21 20:30 UTC


VibeVoice is an open source tool in Image and video generation with 57,462 GitHub stars (+282 in the last 7 days), MIT license, slow maintenance, as of 21 Nov 2026.

## Answer
VibeVoice is a family of open-source voice models for long-form speech recognition and text-to-speech, including real-time streaming.

## Takeaways
- ASR model handles 60-minute audio in one pass
- Transcriptions show who spoke, when, and what they said
- TTS models can generate long multi-speaker speech

## Use cases
- Transcribing long meetings with speaker labels and timestamps
- Building podcasts or dialogue audio with multiple speakers
- Real-time speech-to-text as audio streams in
- Generating speech with low first-audio delay using a small model

## Fit for
- Teams needing long-form transcription with speaker identification
- Developers building speech apps who want open models

## Not for
- Users who want a finished app with no coding
- Workloads needing guaranteed accurate or unbiased output

## Setup difficulty
hard

## Quick answers
### Is VibeVoice free to use at work?
Yes. MIT: Free to use at work. Keep the copyright notice. Not legal advice.

### Is VibeVoice still maintained?
Slow. 21 commits in the last 90 days.

### How popular is VibeVoice?
57,462 GitHub stars. +282 in the last 7 days. As of 21 Nov 2026.

## Stats
- Category: Image and video generation
- 57,462 stars, as of 2026-11-21 20:30 UTC
- +282 stars in 7 days
- License: MIT (Yes: Free to use at work. Keep the copyright notice.). Not legal advice. License: https://github.com/microsoft/VibeVoice
- GitHub: https://github.com/microsoft/VibeVoice

## Health
Maintenance: Slow
Health: 45/100 (as of 2026-11-18)
- Issues: 133 open / 116 closed
- Pull requests: 73 open / 45 merged
- Commits in the last 90 days: 21
- Latest release: none (0 releases)

Cite as: VibeVoice on OpenPick, https://openpick.ing/r/vibevoice, data as of 21 Nov 2026.

AI-generated content. Content updated 2026-11-17 11:32 UTC.

See Methodology: https://openpick.ing/methodology · Disclaimer: https://openpick.ing/disclaimer