Domain Adaptation Method and Modality Gap Impact in Audio-Text Models for Prototypical Sound Classification

TL;DR

This paper investigates the impact of background noise on audio-text models for sound classification, proposing a domain adaptation method that improves accuracy by accounting for background sources and reducing modality gap effects.

Contribution

It introduces a novel domain adaptation technique that quantifies background contributions and narrows the audio-text modality gap, enhancing zero-shot sound classification performance.

Findings

Performance drops with increasing background SNR levels

The proposed method improves accuracy across diverse backgrounds

Reducing modality gap enhances classification results

Abstract

Audio-text models are widely used in zero-shot environmental sound classification as they alleviate the need for annotated data. However, we show that their performance severely drops in the presence of background sound sources. Our analysis reveals that this degradation is primarily driven by SNR levels of background soundscapes, and independent of background type. To address this, we propose a novel method that quantifies and integrates the contribution of background sources into the classification process, improving performance without requiring model retraining. Our domain adaptation technique enhances accuracy across various backgrounds and SNR conditions. Moreover, we analyze the modality gap between audio and text embeddings, showing that narrowing this gap improves classification performance. The method generalizes effectively across state-of-the-art prototypical approaches, showcasing its scalability and robustness for diverse environments.