Improving Natural-Language-based Audio Retrieval with Transfer Learning and Audio & Text Augmentations

TL;DR

This paper explores transfer learning and data augmentation techniques to enhance natural-language-based audio retrieval, demonstrating improved performance and reduced overfitting in a challenging task.

Contribution

It introduces a system that combines pre-trained embeddings with augmentation strategies, systematically optimized for better audio retrieval with natural language queries.

Findings

Augmentation strategies reduce overfitting.

Performance improvements in retrieval accuracy.

Systematic hyperparameter tuning enhances results.

Abstract

The absence of large labeled datasets remains a significant challenge in many application areas of deep learning. Researchers and practitioners typically resort to transfer learning and data augmentation to alleviate this issue. We study these strategies in the context of audio retrieval with natural language queries (Task 6b of the DCASE 2022 Challenge). Our proposed system uses pre-trained embedding models to project recordings and textual descriptions into a shared audio-caption space in which related examples from different modalities are close. We employ various data augmentation techniques on audio and text inputs and systematically tune their corresponding hyperparameters with sequential model-based optimization. Our results show that the used augmentations strategies reduce overfitting and improve retrieval performance.