lipi
Small, fast, on-device English speech recognition. Streaming-capable, 4-bit weights, 13.6 MB, built for CPU-only runtimes.
Results
Word error rate (%), lower is better. Greedy decoding, no language model, full context, English only.
Measured from the exported 4-bit weights with 8-bit activations. Test data: the hf-audio/open-asr-leaderboard test configurations, using the cleaned versions where the leaderboard provides them.
| Benchmark (test set) | WER |
|---|---|
| LibriSpeech test-clean | 4.55 |
| LibriSpeech test-other | 11.75 |
| Earnings-22 (cleaned, call level) | 17.50 |
| GigaSpeech (cleaned) | 14.80 |
| SPGISpeech | 3.82 |
| AMI (cleaned) | 14.78 |
| VoxPopuli (cleaned) | 6.39 |
| Mean of the seven | 10.51 |
Also: Earnings-22 raw test 20.49, Common Voice (English) 29.38. Streaming (640 ms chunks) costs accuracy: on LibriSpeech dev-clean 4.46 full context versus 5.86 streaming.
Scoring: our own scorer with the Whisper English text normaliser applied to the references; Common Voice audio is resampled from 48 kHz to 16 kHz. This is not the leaderboard's own harness, so an official run may differ slightly. Known weak spots: accented and noisy speech (Common Voice) and phone-quality business calls (Earnings-22).
Files
lipi-w4.lipithe 4-bit model (13.6 MB)tokenizer.modelSentencePiece tokenizer (1024 pieces)
License
Creative Commons Attribution-NonCommercial-ShareAlike 4.0 (CC BY-NC-SA 4.0) for the model weights and these documents: attribution required, non-commercial use only, and adaptations must be shared under the same licence. Full text in LICENSE.
Training data and attribution
Trained on public English speech corpora: LibriSpeech, GigaSpeech, SPGISpeech, VoxPopuli, AMI and TED-LIUM, plus pseudo-labelled data:
- YODAS-Granary English audio (CC-BY-3.0), transcribed with NVIDIA Parakeet-TDT 0.6B v2 (CC-BY-4.0) and kept where the two transcripts agree;
- Earnings-22 earnings-call audio (CC-BY-4.0), excluding every call used in the evaluation sets, transcribed with the same teacher.
Some of the corpora above are distributed under research-only or non-commercial terms (GigaSpeech, SPGISpeech, TED-LIUM), which is why the model is released for non-commercial use only.