Domain Adaptation Method and Modality Gap Impact in Audio-Text Models for Prototypical Sound Classification
This addresses a practical issue for users of audio-text models in noisy environments, offering an incremental improvement through domain adaptation.
The paper tackled the problem of audio-text models performing poorly in zero-shot environmental sound classification when background sounds are present, showing that performance degradation is primarily driven by SNR levels of background soundscapes. The proposed method quantifies and integrates background contributions into classification, improving accuracy across various backgrounds and SNR conditions without requiring model retraining.
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.