Grapheme-level pronunciation quality detection on AraVoiceL2 (test)
61.72PrecisionAraVoiceL2-Train Baseline
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AraVoiceL2-Train BaselineTraining Type=Baseline, Dataset Level=Grapheme2022.11 | 61.72 | 53.78 | 55.37 | |
| SpeechBlender-AugAugmentation Strategy=SpeechBlender-Aug, Base Training Dataset=AraVoiceL2-Train2022.11 | 60.33 | 59.69 | 60 | |
| Text-augmentationAugmentation Strategy=Text-augmentation, Base Training Dataset=AraVoiceL2-Train2022.11 | 60.08 | 56.57 | 58.08 | |
| GOP-augmentationAugmentation Strategy=GOP-augmentation, Base Training Dataset=AraVoiceL2-Train2022.11 | 59.26 | 57.98 | 58.56 |