LibriSpeech test-other
Multimodal
The test-other split of LibriSpeech, 2,939 utterances of read English audiobook speech from LibriVox drawn from the corpus's harder-to-recognize speakers. Scores are corpus word error rate in percent after text normalization (the Hugging Face Open ASR Leaderboard applies the Whisper English normalizer); lower is better.
Top models (lower is better)
| Model | Score |
|---|---|
| LFM2.5-Audio-1.5B | 4.3 |
| LFM2-Audio-1.5B | 4.4 |