Automatic Speech Recognition on SlideSpeech
14WERTE2SL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TE2SLSource=LibriSpeech2026.05 | 14 | 57.3 | |
| Upsample-and-MaskSource=LibriSpeech2026.05 | 16.3 | 51 | |
| Soft PromptSource=LibriSpeech2026.05 | 16.4 | 50.7 | |
| TASU2Train Data (Text)=Libri, Conditioning=None2026.04 | 16.41 | — | |
| BaselineSource=LibriSpeech2026.05 | 17 | 50.8 | |
| TASU2Train Data (Text)=Libri+Slide, Conditioning=WER-binned2026.04 | 17.31 | — | |
| TASU2Train Data (Text)=Libri, Conditioning=WER-binned2026.04 | 17.67 | — | |
| SLAM-CTCTrain Data (Audio)=Libri2026.04 | 18.59 | — | |
| TASUTrain Data (Text)=Libri+Slide2026.04 | 18.7 | — | |
| TASUTrain Data (Text)=Libri2026.04 | 24.07 | — |