Brief

Researchers have developed a new method for detecting and diagnosing mispronunciations using language-specific statistical graphs. This approach models phoneme confusion patterns as directed graphs and incorporates strategies to account for pronunciation differences based on a user's native language. Experiments on the L2-ARCTIC benchmark showed this method achieved an F1-score of 59.52%, surpassing existing baseline approaches. AI

Researchers have developed a new method called supervised contrastive learning (SupCon) to improve the robustness of automatic speech recognition (ASR) systems against accent variations. This technique acts as an auxiliary objective during the fine-tuning process, regularizing the model's internal representations without requiring architectural changes or explicit accent labels. Experiments on the L2-ARCTIC benchmark demonstrated significant reductions in word error rates, particularly for unseen accents. AI