Domain-Aware Mispronunciation Detection and Diagnosis Using Language-Specific Statistical Graphs

In three linesMethod for mispronunciation detection and diagnosis using language-specific statistical graphs. Approach captures phoneme confusion patterns and systematic pronunciation differences across native language (L1) backgrounds. F1-score of 59.52% on L2-ARCTIC benchmark.

Summary generated by Claude — human-verified