Speech-to-Text Translation on MuST-C EN->DE
Metric: Case-sensitive sacreBLEU (higher is better)
Metric: Case-sensitive sacreBLEU (higher is better)
| # | Model↕ | Case-sensitive sacreBLEU▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | Task Modulation + Multitask Learning(ASR/MT) + Data Augmentation | 28.88 | Yes | - | - | - |
| 2 | Wav2Vec2.0+mBART+Adaptors | 28.22 | Yes | End-to-End Speech Translation with Pre-trained M... | 2021-05-10 | Code |
| 3 | Transformer + Meta Learning(ASR/MT) + Data Augmentation | 27.51 | Yes | - | - | - |
| 4 | Transformer with Adapters | 24.63 | No | Lightweight Adapter Tuning for Multilingual Spee... | 2021-06-02 | Code |
| 5 | Dual-decoder Transformer | 23.63 | No | Dual-decoder Transformer for Joint Automatic Spe... | 2020-11-02 | Code |
| 6 | Speechformer | 23.6 | No | Speechformer: Reducing Information Loss in Direc... | 2021-09-09 | Code |
| 7 | Transformer + ASR Pretrain | 22.8 | No | NeurST: Neural Speech Translation Toolkit | 2020-12-18 | Code |
| 8 | Transformer + ASR Pretrain | 22.7 | No | fairseq S2T: Fast Speech-to-Text Modeling with f... | 2020-10-11 | Code |