All tasks / Models
VibeVoice-ASR
Microsoft's open-weights long-form speech recognition model with speaker diarization.
- Microsoft
- Open weights
- Released 2 Mar 2026
Where it ranks
- BTranscriptionModels · Local · 8.2% on Hugging Face Open ASR Leaderboard (English, short-form average WER) (secondary benchmark)
Facts
- Vendor
- Microsoft
- Family
- VibeVoice-ASR
- Released
- 2 Mar 2026
- Weights
- Open: download and run it yourself
- License
- MIT
- Parameters
- 8.33B
- GPU memory to run locally
- ~19 GB (Derived: about 8.33B parameters is roughly 16.7 GB of bf16 weights plus runtime overhead; not an official requirement.)
- Access
- Free (open weights)
- Last verified
- 19 Oct 2026
Benchmark results
| Benchmark | Result | Entry on the leaderboard | As of | Source |
|---|---|---|---|---|
| Hugging Face Open ASR Leaderboard (English, short-form average WER) | 8.2% | microsoft/VibeVoice-ASR-HFAverage WER over the board's default datasets (English short-form, version 17-10-2026); rank 46 of 56; read from the Space's /config table data. | 17 Oct 2026 | huggingface.co |
Where you can use it
No tracked tool lists this model yet; as open weights it runs in local runtimes.