Speech Recognition On Wsj Eval92
평가 지표
Word Error Rate (WER)
평가 결과
이 벤치마크에서 각 모델의 성능 결과
비교 표
| 모델 이름 | Word Error Rate (WER) |
|---|---|
| end-to-end-speech-recognition-using-lattice | 3.0 |
| espresso-a-fast-end-to-end-neural-speech | 3.4 |
| bigssl-exploring-the-frontier-of-large-scale | 1.3 |
| 모델 4 | 5.6 |
| purely-sequence-trained-neural-networks-for | 2.32 |
| efficient-neural-architecture-search-for-end | 2.77 |
| jasper-an-end-to-end-convolutional-neural | 6.9 |
| deep-recurrent-neural-networks-for-acoustic | 3.5 |
| speechstew-simply-mix-all-available-speech | 1.3 |
| 모델 10 | 3.6 |
| fully-convolutional-speech-recognition | 3.5 |
| cat-a-ctc-crf-based-asr-toolkit-bridging-the | 3.2 |
| crf-based-single-stage-acoustic-modeling-with | 3.79 |
| relaxed-attention-a-simple-method-to-boost | 3.19 |
| deep-speech-2-end-to-end-speech-recognition | 3.60 |
| generative-speech-recognition-error | 2.11 |
| it-s-never-too-late-fusing-acoustic | 2.2 |