VibeVoice-ASR 7B
VibeVoice-ASR 7B is a speech-to-text model (9B total) from Microsoft, released under the MIT license. The reviewed source lists 51 languages (51 explicitly enumerated in our record), and the model is designed for batch-first transcription. It produces speaker-attributed transcripts natively. Among the common setups we test, it ranks highest for meetings with speaker labels (#1 of 26 in the matcher).
Long-form model marketed as VibeVoice-ASR-7B; the current Hugging Face artifact reports 9B total parameters. Accepts up to 60 minutes in one pass and emits speaker-attributed, timestamped Who/When/What transcripts across 51 listed languages. MIT licensed; benchmark on your own meetings before production.
Key facts
| Provider | Microsoft |
|---|---|
| Model family | VibeVoice-ASR |
| Parameters | 9B total |
| License | MIT · Permissive — license text ↗ |
| Language coverage | 51 listed languages · 51 enumerated in our record |
| Streaming | Batch-first (no streaming path in the reviewed source) |
| Timestamps | Yes — in the model or its runtime family |
| Speaker diarization | Yes — native speaker-attributed output |
| Speech translation | No |
| Documented runtimes | microsoft/VibeVoice, vLLM, Hugging Face Transformers — how to run it |
| Primary source | huggingface.co/microsoft/VibeVoice-ASR ↗ |
| Last reviewed |
Hardware and workload fit
These are Speech Model Match's conservative 1–5 editorial ratings, relative to the other 19 reviewed models. They come from model size, official memory guidance and documented runtimes — not from a universal benchmark. How the ratings work.
| Dimension | Rating | Scale |
|---|---|---|
| CPU-only fit | 1/5 | |
| Apple Silicon fit | 2/5 | |
| NVIDIA GPU fit | 4/5 | |
| Speed | 2/5 | |
| Accuracy (relative) | 4/5 | |
| Low memory use | 1/5 | |
| Long recordings | 5/5 | |
| Live / low latency | 1/5 | |
| Subtitle work | 5/5 | |
| Edge / low-power devices | 1/5 |
Where it is strong
- Native speaker-attributed transcription (diarization)
- NVIDIA GPU fit: rated 4/5
- Accuracy (relative): rated 4/5
- Long recordings: rated 5/5
- Subtitle work: rated 5/5
- Timestamps are available in the model or its runtime family
What to check first
- No streaming path in the reviewed source — plan for file/batch transcription
- CPU-only fit: rated 1/5
- Apple Silicon fit: rated 2/5
- Speed: rated 2/5
- Low memory use: rated 1/5
- Live / low latency: rated 1/5
- Edge / low-power devices: rated 1/5
How it ranks in common setups
Ranks come from the same deterministic matcher that runs on the homepage, applied to all 26 reviewed models. They are decision guidance, not accuracy claims.
| Setup | Rank | Fit score | Top match |
|---|---|---|---|
| Long recordings on an 8–12 GB NVIDIA GPU, balanced | #14 of 26 | 68/100 · Possible fit | Whisper turbo |
| Live captions or voice input (native streaming + timestamps, speed first) | #25 of 26 | 50/100 · Weak fit | Nemotron 3.5 ASR Streaming 0.6B |
| Subtitles with timing, accuracy first | #9 of 26 | 79/100 · Strong fit | Whisper medium |
| Meetings with speaker labels (diarization + timestamps) | #1 of 26 | 86/100 · Excellent fit | This model |
| CPU-only machine, low memory use | #22 of 26 | 51/100 · Weak fit | VibeVoice-ASR-BitNet |
| Apple Silicon with 8–16 GB unified memory, long recordings | #21 of 26 | 59/100 · Possible fit | Whisper small |
| Japanese live transcription (native streaming, speed first) | #17 of 26 | 44/100 · Weak fit | Nemotron 3.5 ASR Streaming 0.6B |
Supported languages
- Afar
aa - Afrikaans
af - Arabic
ar - Armenian
hy - Bulgarian
bg - Catalan
ca - Croatian
hr - Czech
cs - Danish
da - Dutch
nl - English
en - Estonian
et - Filipino
fil - Finnish
fi - French
fr - German
de - Greek
el - Hebrew
he - Hindi
hi - Hungarian
hu - Icelandic
is - Indonesian
id - Italian
it - Japanese
ja - Javanese
jv - Kalaallisut
kl - Korean
ko - Latvian
lv - Lithuanian
lt - Malay
ms - Mandarin Chinese
zh - Mongolian
mn - Nepali
ne - Norwegian
no - Persian
fa - Polish
pl - Portuguese
pt - Romanian
ro - Russian
ru - Serbian
sr - Slovak
sk - Slovenian
sl - Spanish
es - Swahili
sw - Swedish
sv - Thai
th - Turkish
tr - Ukrainian
uk - Urdu
ur - Vietnamese
vi - Yiddish
yi
Compare coverage across every reviewed model in the language support matrix.
Community activity on Hugging Face
- Downloads, last 30 days
- 714,453
- Likes
- 1,328
- Repository last updated
- 27 Jan 2026
- Published on the Hub
- 21 Jan 2026
Figures from the Hugging Face Hub API, 11 Nov 2026. Source: huggingface.co/microsoft/VibeVoice-ASR ↗. Downloads count file downloads, not active users, and do not affect the matcher's scores. Compare all models.
How to run VibeVoice-ASR 7B locally
Microsoft runs VibeVoice-ASR from its VibeVoice repository, with a vLLM launcher for serving. Runtimes below are the ones the official model card or repository documents for this model (checked November 2, 2026).
- microsoft/VibeVoice ↗Official package · reference code and demos
- vLLM ↗Serving engine · Docker launcher
- Hugging Face Transformers ↗Framework · separate VibeVoice-ASR-HF checkpoint
Install command shown in the official github.com page ↗:
git clone https://github.com/microsoft/VibeVoice.git
cd VibeVoice
pip install -e .
Audio input: Accepts up to 60 minutes of continuous audio in one pass (64K-token context).
Official hardware guidance: Microsoft's setup uses an NVIDIA GPU container. No CPU-only path is documented for this model; the BitNet build is the CPU option. Source ↗
Whatever runtime you choose, measure speed and accuracy on your own audio with the benchmark tool; results differ a lot between runtimes, precision and hardware.
Before you deploy VibeVoice-ASR 7B
A printable checklist built from the facts on this page. Tick it off before putting the model into production.
- Language. Confirm your language is in the 51-language list on this page, and test any accents or code-switching you expect.
- License. MIT: Include the copyright notice and the permission notice in all copies or substantial portions. Provided as is, without warranty. License guide.
- Install. Start from the official command:
git clone https://github.com/microsoft/VibeVoice.git …(full run notes). - Audio input. Accepts up to 60 minutes of continuous audio in one pass (64K-token context).
- Hardware. Microsoft's setup uses an NVIDIA GPU container. No CPU-only path is documented for this model; the BitNet build is the CPU option.
- Features you rely on. For VibeVoice-ASR 7B, no streaming path is documented; timestamps are documented; speaker attribution is native. Test each one you need with real audio.
- Measure. Run 10–20 representative clips, score them with the WER calculator and record speed in the benchmark workspace before you commit.
Alternatives and related models
Same family: VibeVoice-ASR Streaming 1.5B, VibeVoice-ASR-BitNet, VibeVoice-ASR-Streaming-7B.
Similar fit profile: Cohere Transcribe 03-2026 (Cohere Labs, 2B), Jais-ASR (IFM, 5.7B total), Qwen3-ASR 1.7B (Qwen, 1.7B). Similarity is computed from the ten fit ratings plus streaming and diarization support.
| Fact | VibeVoice-ASR 7B | VibeVoice-ASR Streaming 1.5B | VibeVoice-ASR-BitNet |
|---|---|---|---|
| Provider · size | Microsoft · 9B total | Microsoft · 3B total | Microsoft · 0.3B quantized |
| License | MIT Permissive | MIT Permissive | MIT Permissive |
| Languages | 51 listed languages | 10 listed languages | 7 listed languages |
| Streaming | Batch-first | Native | Via runtime |
| Timestamps | Yes | Limited | Limited |
| Speaker diarization | Native | Native | Separate pipeline |
| CPU · Apple · NVIDIA fit | 1/5 · 2/5 · 4/5 | 1/5 · 2/5 · 4/5 | 5/5 · 4/5 · 3/5 |
| Live · long recordings | 1/5 · 5/5 | 5/5 · 4/5 | 4/5 · 4/5 |
| Main runtime | microsoft/VibeVoice | microsoft/VibeVoice | VibeASR.cpp |
Head-to-head pages: Whisper turbo vs VibeVoice-ASR 7B
Compare VibeVoice-ASR 7B side by side
Guides that cover VibeVoice-ASR 7B
- Best local speech-to-text for meetings and speaker attribution
- Best local speech-to-text models in 2026
- Best local speech-to-text models for CPU-only PCs
- Best local speech-to-text models for streaming
- Best local speech-to-text models for multilingual use
- Best local speech-to-text for subtitles and timestamps
Check VibeVoice-ASR 7B for your setup
Pick your language, hardware and workload to see this model's fit score, the reasons behind it and where it ranks against all reviewed models. Everything runs in your browser.
Common questions
Can I use VibeVoice-ASR 7B commercially?
Permissive license; commercial use is allowed under the listed terms. The listed license is MIT; read the current license text ↗ before shipping a product, because terms can change. The license guide compares the obligations of all 26 reviewed models.
Does VibeVoice-ASR 7B support real-time streaming?
No streaming path is documented in the reviewed source. Treat it as a file/batch transcription model.
Which languages does VibeVoice-ASR 7B transcribe?
51 listed languages. The full list of 51 explicitly enumerated languages is shown above.
Can VibeVoice-ASR 7B tell speakers apart?
Yes. The reviewed source documents native speaker-attributed transcription. Still check accuracy on overlapping speech in your own recordings.
Primary sources
Reviewed against primary sources on September 16, 2026. Model behavior, runtime support, language coverage and licenses can change; re-check the linked primary source before production use. See the full source directory.