Performance improvement of spatial semantic segmentation with enriched audio features and agent-based error correction for DCASE 2025 Challenge Task 4

TL;DR

This paper enhances spatial semantic segmentation of sound scenes by integrating enriched audio features and an agent-based error correction system, leading to significant performance improvements in the DCASE 2025 Challenge Task 4.

Contribution

It introduces the use of spectral roll-off and chroma features along with an agent-based label correction to improve audio tagging accuracy in spatial sound scene analysis.

Findings

CA-SDRi improved by up to 14.7%

Enhanced classification of low-performing classes

Effective reduction of false positives

Abstract

This technical report presents submission systems for Task 4 of the DCASE 2025 Challenge. This model incorporates additional audio features (spectral roll-off and chroma features) into the embedding feature extracted from the mel-spectral feature to im-prove the classification capabilities of an audio-tagging model in the spatial semantic segmentation of sound scenes (S5) system. This approach is motivated by the fact that mixed audio often contains subtle cues that are difficult to capture with mel-spectrograms alone. Thus, these additional features offer alterna-tive perspectives for the model. Second, an agent-based label correction system is applied to the outputs processed by the S5 system. This system reduces false positives, improving the final class-aware signal-to-distortion ratio improvement (CA-SDRi) metric. Finally, we refine the training dataset to enhance the classi-fication accuracy of low-performing classes by removing irrele-vant samples and incorporating external data. That is, audio mix-tures are generated from a limited number of data points; thus, even a small number of out-of-class data points could degrade model performance. The experiments demonstrate that the submit-ted systems employing these approaches relatively improve CA-SDRi by up to 14.7% compared to the baseline of DCASE 2025 Challenge Task 4.