Automatic Speech Recognition on WSJ (test)
0.01WEROracle Ocp
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Oracle OcpOracle Type=Compositional2024.05 | 0.01 | — | |
| ClozeGERBackbone=SpeechGPT, Source Speech=true, LoRA=true, Logits Calibration=true, Post-processing=true2024.05 | 0.024 | 42.9 | |
| ClozeGERBackbone=LLaMA-2-7b, Source Speech=false, LoRA=true, Logits Calibration=true, Post-processing=true2024.05 | 0.025 | 40.5 | |
| GERBackbone=SpeechGPT, Source Speech=true, LoRA=true2024.05 | 0.027 | 35.7 | |
| Oracle OnbOracle Type=N-best2024.05 | 0.027 | — | |
| GERBackbone=LLaMA-2-7b, Source Speech=false, LoRA=true2024.05 | 0.028 | 33.3 | |
| GERBackbone=SpeechGPT, Source Speech=false, LoRA=true2024.05 | 0.029 | 31 | |
| ClozeGERBackbone=SpeechGPT, Source Speech=true, LoRA=true, Logits Calibration=true, Post-processing=false2024.05 | 0.033 | 21.4 | |
| ClozeGERBackbone=LLaMA-2-7b, Source Speech=false, LoRA=true, Logits Calibration=true, Post-processing=false2024.05 | 0.033 | 21.4 | |
| ClozeGERBackbone=LLaMA-2-7b, Source Speech=false, LoRA=true, Logits Calibration=false, Post-processing=false2024.05 | 0.037 | 11.9 | |
| ClozeGERBackbone=SpeechGPT, Source Speech=true, LoRA=true, Logits Calibration=false, Post-processing=false2024.05 | 0.038 | 9.5 | |
| Whisper Baseline2024.05 | 0.042 | — |