Speech Recognition On Aishell 1
Métriques
Params(M)
Word Error Rate (WER)
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | Params(M) | Word Error Rate (WER) |
|---|---|---|
| unified-streaming-and-non-streaming-two-pass | 47 | 4.72 |
| cr-ctc-consistency-regularization-on-ctc-for | 66.2 | 4.02 |
| funasr-a-fundamental-end-to-end-speech | 46.3 | 4.95 |
| qwen-audio-advancing-universal-audio | - | 1.29 |
| lightweight-transducer-based-on-frame-level | 45.3 | 4.03 |
| funasr-a-fundamental-end-to-end-speech | 220 | 1.95 |
| end-to-end-speech-recognition-with-adaptive | - | 18.7 |
| improving-mandarin-speech-recogntion-with | 46 | 4.1 |
| a-comparative-study-on-transformer-vs-rnn-in | - | 6.7 |
| cat-a-ctc-crf-based-asr-toolkit-bridging-the | - | 6.34 |
| unimodal-aggregation-for-ctc-based-speech | 44.7 | 4.7 |
| mmspeech-multi-modal-multi-task-encoder | - | 1.9 |
| knowledge-transfer-from-pre-trained-language | 47 | 4.1 |
| beyond-universal-transformer-block-reusing | 8.5 | 6.63 |
| fireredasr-open-source-industrial-grade | 1,100 | 0.55 |
| seed-asr-understanding-diverse-speech-and | - | 0.68 |
| lightweight-transducer-based-on-frame-level | 45.3 | 4.31 |
| bat-boundary-aware-transducer-for-memory | 90 | 4.97 |