Exploring ASR-Based Wav2Vec2 for Automated Speech Disorder Assessment: Insights and Analysis

TL;DR

This paper presents the first analysis of this baseline model for speech quality assessment, focusing on intelligibility and severity tasks, and conducts layer-wise analysis to identify key layers and compare different SSL and ASR Wav2Vec2 models based on pretrained data.

Abstract

With the rise of SSL and ASR technologies, the Wav2Vec2 ASR-based model has been fine-tuned for automated speech disorder quality assessment tasks, yielding impressive results and setting a new baseline for Head and Neck Cancer speech contexts. This demonstrates that the ASR dimension from Wav2Vec2 closely aligns with assessment dimensions. Despite its effectiveness, this system remains a black box with no clear interpretation of the connection between the model ASR dimension and clinical assessments. This paper presents the first analysis of this baseline model for speech quality assessment, focusing on intelligibility and severity tasks. We conduct a layer-wise analysis to identify key layers and compare different SSL and ASR Wav2Vec2 models based on pre-trained data. Additionally, post-hoc XAI methods, including Canonical Correlation Analysis (CCA) and visualization techniques, are used to track model evolution and visualize embeddings for enhanced interpretability.

Figures (3)

• Figure 1: Performance comparison of freeze and fine-tuned layer-wise feature extractor training on speech quality assessment tasks
• Figure 2: CCA similarity between fine-tuned feature extractors with pre-trained ASR Wav2Vec2, SSL Wav2vec2 models and phoneme encoder
• Figure 3: 2D t-SNE visualization of the last Wav2Vec2 layer