Self-supervised Flemish Dutch Speech Models vs. Fine-tuned English Models in Robustness to Noise and Accent Variation

Self-supervised pre-trained speech models have strongly improved speech recognition, yet they are still sensitive to domain shifts and accented or atypical speech. Many of these models rely on quantisation or clustering to learn discrete acoustic units. We propose to correct the discovered discrete units for accented speech back to a standard pronunciation in an unsupervised manner. A masked language model is trained on discrete units from a standard accent and iteratively corrects an accented token sequence by masking unexpected cluster sequences and predicting their common variant. Small acc

Research goal: How do self-supervised speech models pre-trained on low-resource Flemish Dutch compare to fine-tuned English models in terms of robustness to noise and accent variation on standard ASR test sets?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.7/10.