Interpretable Representation Learning for Speech and Audio Signals Based on Relevance Weighting

TL;DR

This paper introduces a relevance weighting scheme within neural networks that enhances interpretability and improves performance in speech recognition and sound classification tasks by enabling feature selection during model training.

Contribution

The work presents a novel relevance weighting approach integrated into CNNs for interpretable speech and audio representation learning, applicable to noisy and reverberant environments.

Findings

Relevance weights capture meaningful speech/audio content.

Improved accuracy in speech recognition on noisy datasets.

Enhanced sound classification performance.

Abstract

The learning of interpretable representations from raw data presents significant challenges for time series data like speech. In this work, we propose a relevance weighting scheme that allows the interpretation of the speech representations during the forward propagation of the model itself. The relevance weighting is achieved using a sub-network approach that performs the task of feature selection. A relevance sub-network, applied on the output of first layer of a convolutional neural network model operating on raw speech signals, acts as an acoustic filterbank (FB) layer with relevance weighting. A similar relevance sub-network applied on the second convolutional layer performs modulation filterbank learning with relevance weighting. The full acoustic model consisting of relevance sub-networks, convolutional layers and feed-forward layers is trained for a speech recognition task on noisy and reverberant speech in the Aurora-4, CHiME-3 and VOiCES datasets. The proposed representation learning framework is also applied for the task of sound classification in the UrbanSound8K dataset. A detailed analysis of the relevance weights learned by the model reveals that the relevance weights capture information regarding the underlying speech/audio content. In addition, speech recognition and sound classification experiments reveal that the incorporation of relevance weighting in the neural network architecture improves the performance significantly.