L2-ARCTIC
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Accent Normalization | L2-ARCTIC (test) | NAT63.85 | 9 | |
| Automatic Speech Recognition | L2-ARCTIC (test) | WER7.2 | 9 | |
| Phonetic Perception | L2-ARCTIC | PFER8 | 8 | |
| Suggestion Generation | L2-Arctic-plus (test) | BLEU-220.4 | 8 | |
| Mispronunciation Detection | L2-Arctic-plus (test) | Precision53.8 | 8 | |
| Automatic Speech Recognition | L2-Arctic Arabic and Vietnamese accents (test) | WER (ARB)20.3 | 8 | |
| Automatic Speech Recognition | L2-Arctic Chinese and Vietnamese accents (test) | WER (CHN)25.3 | 8 | |
| Automatic Speech Recognition | L2-Arctic Arabic and Chinese accents (test) | WER (Arabic)20.3 | 8 | |
| Automatic Speech Recognition | L2-ARCTIC created subset | Error Rate (ZH)11.67 | 8 | |
| Speech Quality Assessment | L2-ARCTIC Indian English speakers | NISQA-MOS4.84 | 6 | |
| Accent Neutralization | L2-ARCTIC Indian English speakers | CNA100 | 6 | |
| Speech Recognition | L2-Arctic | Arabic WER22.92 | 4 | |
| Pronunciation Training Feedback Generation | L2-Arctic-plus Human Evaluation (12 samples) | SR (Suggestion Relevance)3.8 | 4 | |
| Speaker Similarity Analysis | L2-ARCTIC Indian English speakers | SpkSim0.86 | 3 | |
| Feedback Generation | L2-Arctic-plus | Average Score2.328 | 3 | |
| Mispronunciation Detection and Suggestion Generation | L2-Arctic-plus | Win Rate96.55 | 2 |
Showing 16 of 16 rows