Comparative layer-wise analysis of self-supervised speech models
This work addresses the limited understanding of model properties for researchers in speech processing, offering insights for more efficient use of pre-trained models, though it is incremental as it builds on existing analysis methods.
The study analyzed intermediate representations of various self-supervised speech models to understand how acoustic, phonetic, and word-level properties evolve across layers, finding that these trends relate to pre-training objectives and can guide layer selection for downstream tasks, with single-layer performance often matching or improving upon using all layers.
Many self-supervised speech models, varying in their pre-training objective, input modality, and pre-training data, have been proposed in the last few years. Despite impressive successes on downstream tasks, we still have a limited understanding of the properties encoded by the models and the differences across models. In this work, we examine the intermediate representations for a variety of recent models. Specifically, we measure acoustic, phonetic, and word-level properties encoded in individual layers, using a lightweight analysis tool based on canonical correlation analysis (CCA). We find that these properties evolve across layers differently depending on the model, and the variations relate to the choice of pre-training objective. We further investigate the utility of our analyses for downstream tasks by comparing the property trends with performance on speech recognition and spoken language understanding tasks. We discover that CCA trends provide reliable guidance to choose layers of interest for downstream tasks and that single-layer performance often matches or improves upon using all layers, suggesting implications for more efficient use of pre-trained models.