What Does a Pathological Speech Assessment Model Know about Acoustic Features? A Case Study on Oral and Oropharyngeal Cancer Patients
This paper investigates the interpretability of a Wav2Vec 2.0-based speech intelligibility assessment model for oral and oropharyngeal cancer patients.
This work offers a novel approach to interpreting the acoustic information learned by a Wav2Vec 2.0-based speech intelligibility assessment model.
Before reading this…
Applications
- →Speech assessment for oral and oropharyngeal cancer patients
To understand this paper, make sure you know these concepts first:
- Knowledge of speech processing and machine learningfind papers →
Abstract
More Like ThisThis work investigates the interpretability of a Wav2Vec 2.0based speech intelligibility assessment model for oral and oropharyngeal cancer patients through canonical correlation analysis. By measuring the correlation between the model embeddings and eGeMAPS low-level descriptors (LLDs) as an interpretable reference, we analyze how acoustic information is encoded across the model layers. The analysis is conducted at two levels: individual LLDs layer-wise, and group-level: prosodic, spectral, and voice quality. Results show that the learned representations are most strongly correlated with spectral and prosodic features, with the first MFCC coefficient yielding the highest correlations across all layers. At the group level, spectral and prosodic groups achieve correlations of 0.77 and 0.71 respectively, while voice quality reaches 0.65. Beyond model interpretability, this work also offers practical guidance on acoustic feature selection for pathological speech assessment.