Automatic Speech Recognition on LibriCSS (test)
0.25Ins. ErrorConformer-transducer (IHM+)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Conformer-transducer (IHM+)Setting=IHM+, Backbone=Conformer-transducer2022.12 | 0.25 | 0.22 | 1.74 | 2.21 | |
| GPU-accelerated GSSSetting=GSS, Backbone=Conformer-transducer, Channels=72022.12 | 0.31 | 0.89 | 2.14 | 3.34 | |
| Conformer-transducer (SDM)Setting=SDM, Backbone=Conformer-transducer2022.12 | 1.06 | 3.12 | 6.59 | 10.77 |