Fisher
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Conditional Dialogue Generation | Fisher (test) | GPT-4o Score7.3 | 64 | |
| Speech-to-speech translation | Fisher Spanish-English (test) | BLEU (Speech Input)90.5 | 55 | |
| Speech-to-speech translation | Fisher Spanish-English (dev) | BLEU (Speech)88.5 | 48 | |
| Speech-to-speech translation | Fisher Spanish-English (dev2) | ASR BLEU89.4 | 36 | |
| Unconditional Dialogue Generation | Fisher (test) | GPT-4o Score9.33 | 32 | |
| Speech Translation | Fisher Monolingual (test) | BLEU35.87 | 11 | |
| Speech Translation | Fisher Code-Switching (test) | BLEU37.51 | 11 | |
| Speaker-Attributed Automatic Speech Recognition | Fisher (test) | WDER0.9 | 11 | |
| Speech-to-Speech Translation | Fisher Es→En (test) | ASR chrF70.2 | 10 | |
| Speech-to-Speech Translation | Fisher Es→En (dev) | ASR chrF69.5 | 10 | |
| Conditional Turn-taking Evaluation | Fisher (test) | Occurrence Proportion58 | 7 | |
| Unconditional Turn-taking Evaluation | Fisher (test) | Occurrence Rate60 | 7 | |
| Backchanneling | Fisher | Init Rate97.8 | 5 | |
| Window-level Turn-taking | Fisher | Onset MAE0.69 | 5 | |
| Dialogue Generation | Fisher | M-MOS4.25 | 4 | |
| ES-to-EN AST | Fisher (test) | BLEU64.7 | 4 | |
| Speaker-Attributed Automatic Speech Recognition | Fisher Global Meeting-level | DER15.21 | 4 | |
| Speaker-Attributed Automatic Speech Recognition | Fisher (local setting) | DER8.18 | 4 | |
| Cross-lexical backchannel similarity | Fisher | Proportion of Correct Selections66.3 | 3 | |
| Prosodic backchannel similarity | Fisher | Proportion Correct Selections69.7 | 3 | |
| Fine-grained Score Accuracy | Fisher | Exact Accuracy64.76 | 1 | |
| Binary classification (Human vs Machine speech) | Fisher (Human-Human) OOD (test) | Accuracy98.44 | 1 |
Showing 22 of 22 rows