Arabic Text Diacritization on CATT
Metric: WER (%) (lower is better)
Metric: WER (%) (lower is better)
| # | Model↕ | WER (%)▲ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | CATT ED | 34.191 | Yes | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 2 | CATT EO | 35.597 | Yes | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 3 | GPT-4 | 38.311 | Yes | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 4 | CBHG | 42.68 | No | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 5 | Command R+ | 48.518 | Yes | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 6 | Deep Diacritization (D2) | 49.417 | No | Deep Diacritization: Efficient Hierarchical Recu... | 2020-11-01 | Code |
| 7 | Shakkala | 50.387 | No | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 8 | Alkhalil | 53.413 | No | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 9 | Sakhr | 56.661 | No | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 10 | Multilevel Diacritizer | 60.844 | No | CATT: Character-based Arabic Tashkeel Transformer | 2024-07-03 | Code |
| 11 | Farasa Diacritization | 65.783 | No | - | - | - |
| 12 | Deep Diacritization (D3) | 98.71 | No | Deep Diacritization: Efficient Hierarchical Recu... | 2020-11-01 | Code |