BYOL-S: Learning Self-supervised Speech Representations by Bootstrapping
This work addresses the need for better audio representations in machine learning, though it appears incremental as it builds on existing self-supervised methods.
The authors tackled the problem of developing general-purpose audio representations by extending self-supervised learning methods with bootstrapping and proposing hybrid models combining handcrafted and learned features. Their hybrid model with a convolutional transformer encoder achieved superior performance in most tasks of the HEAR NeurIPS 2021 challenge for auditory scene classification and timestamp detection.
Methods for extracting audio and speech features have been studied since pioneering work on spectrum analysis decades ago. Recent efforts are guided by the ambition to develop general-purpose audio representations. For example, deep neural networks can extract optimal embeddings if they are trained on large audio datasets. This work extends existing methods based on self-supervised learning by bootstrapping, proposes various encoder architectures, and explores the effects of using different pre-training datasets. Lastly, we present a novel training framework to come up with a hybrid audio representation, which combines handcrafted and data-driven learned audio features. All the proposed representations were evaluated within the HEAR NeurIPS 2021 challenge for auditory scene classification and timestamp detection tasks. Our results indicate that the hybrid model with a convolutional transformer as the encoder yields superior performance in most HEAR challenge tasks.