Automatic Speech Recognition on English DementiaBank Pitt 1.0 (dev)
27.41WER (Par.)Enc&Dec Prompts
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Enc&Dec PromptsModel=Whisper (LoRA), Online=false, SD Param.=0.75M, RTF=4.132026.06 | 27.41 | 12.05 | |
| Enc Only PromptsModel=Whisper (LoRA), Online=false, SD Param.=0.60M, RTF=4.032026.06 | 27.53 | 12.44 | |
| Audio-Textual PromptsModel=Whisper (LoRA), Online=true, RTF=0.422026.06 | 28.05 | 12.51 | |
| Audio-Only PromptsModel=Whisper (LoRA), Online=true, RTF=0.402026.06 | 28.46 | 13.08 | |
| Whisper (LoRA)Model=Whisper (LoRA), Online=false, RTF=0.242026.06 | 28.79 | 12.76 | |
| ECAPA-TDNNModel=Whisper (LoRA), Online=true, RTF=0.272026.06 | 29.01 | 13.85 | |
| i-vectorModel=Whisper (LoRA), Online=true, RTF=0.272026.06 | 29.32 | 13.13 | |
| RABModel=Whisper (LoRA), Online=false, SD Param.=0.53M, RTF=4.152026.06 | 29.54 | 13.14 | |
| x-vectorModel=Whisper (LoRA), Online=true, RTF=0.272026.06 | 31.49 | 14.96 | |
| Conformer Transducer with Only Speech ContextsModel=Conformer Transducer, Online=Not specified2026.06 | 34.95 | 14.99 |