Mispronunciation Detection and Suggestion Generation on L2-Arctic-plus
96.55Win RateWhisper Large + Llama3
Evaluation results
| Method | Links | |
|---|---|---|
| Whisper Large + Llama3Comparison Baseline=Cascaded System (Wav2vec2 Base + Llama 3)2026.01 | 96.55 | |
| Whisper Large + Llama3Comparison Baseline=GPT-4o-Audio2026.01 | 80.78 |