whatmodel.sh / stt / SenseVoice Small
SenseVoice Small
↗ share
+ contribute
SenseVoice Small is an open-weights speech-to-text model from Alibaba (FunAudioLLM), released July 2024. Its published size is 0.234B parameters. Licence: funasr-model-license.
Catalogue identity notes
FunAudioLLM/SenseVoiceSmall (SenseVoice speech foundation model: ASR, language ID, emotion and audio-event detection; card). transcribe.cpp port pins upstream 3eb3b4e (2024-07-31); later commits are card edits. Licence: card 'license: other, license_name: model-license' linking FunASR MODEL_LICENSE.
Model information sources
Release date ↗ · Parameter count ↗
transcribe.cpp catalog 'params' 234,000,287 for its converted port of FunAudioLLM/SenseVoiceSmall@3eb3b4e (no safetensors index on the Hugging Face repo).
Score within the Open ASR English short-form WER (March 2026) comparison set only.
$ how it compares
The speech-to-text models ranked either side of SenseVoice Small, out of 176 with a score, then more from Alibaba (FunAudioLLM).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 131 |  STT En FastConformer-Transducer Large | 46±3 | local | ||
| 132 |  Moonshine Streaming Small | 46±3 | local | ||
| 133 | SenseVoice Small | 46±3 | local | ||
| 134 | STT En FastConformer-CTC Large | 45±3 | local | ||
| 135 |  ESPnet wav2vec 2.0 Large + Conformer + Transformer LM | 45±4 | local | ||
| more from Alibaba (FunAudioLLM) | |||||
| 63 | Fun-ASR-MLT-Nano-2512 | 57±4 | local | ||
| 80 | Fun-ASR-Nano-2512 | 55±3 | local | ||
see all speech-to-text models · find the ones that fit your machine
$ what it's good at
other tests5 not in a task · 97% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LibriSpeech test-clean WER +1 | 3.1% ✎ | ▁···▁▁▂·▄▄██ beats 9/56 | 36% |
| LibriSpeech test-other WER | 6.8% ✎ | ███·····██·█ beats 0/5 | 18% |
| FLEURS English test WER +1 | 7.1% ✎ | ▁··▁▁▁▁▁▃▃█▁ beats 11/56 | 17% |
| FLEURS Mandarin test error rate +1 | 10.1% ✎ | ▂▂··▂▄▂▂·▆██ beats 9/17 | 17% |
| VoxPopuli English test WER | 8.6% ✎ | ▄··▄··▄···█▄ beats 0/5 | 8.5% |
$ where to run it
Other ways to run it
- Run it as published · needs ~0.25 GB · from github.com/handy-computer
git clone https://github.com/handy-computer/transcribe.cpp.git && cd transcribe.cpp && cmake -B build && cmake --build build && hf download handy-computer/SenseVoiceSmall-gguf SenseVoiceSmall-Q8_0.gguf --local-dir models/sensevoice-small && build/bin/transcribe-cli --language en -m models/sensevoice-small/SenseVoiceSmall-Q8_0.gguf audio.wav
Official weights and documentation: huggingface.co/FunAudioLLM ↗
$ how the score was worked out · transcribe.cpp model catalog… 55%, microsoft/VibeVoice-ASR-BitNe… 42%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 transcribe.cpp model catalog · LibriSpeech test-clean WER of validated full-precision ports | 3 | 55% |
 microsoft/VibeVoice-ASR-BitNet model card (Accuracy table) | 5 | 42% |
| prior · pull toward the middle | 3% | |
| weighted average | = 45.5 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json