Squeeze-and-Excite ResNet-Conformers for Sound Event Localization, Detection, and Distance Estimation for DCASE 2024 Challenge

TL;DR

This paper introduces an improved ResNet-Conformer architecture with Squeeze-and-Excitation blocks and SALSA features for enhanced sound event localization, detection, and distance estimation in the DCASE 2024 challenge.

Contribution

It proposes novel architectural enhancements and feature representations, along with data augmentation and distance scaling techniques, to improve polyphonic SELD performance.

Findings

Enhanced SELD accuracy on STARSS23 dataset

Effective use of SALSA features over log-mel spectra

Improved distance estimation accuracy

Abstract

This technical report details our systems submitted for Task 3 of the DCASE 2024 Challenge: Audio and Audiovisual Sound Event Localization and Detection (SELD) with Source Distance Estimation (SDE). We address only the audio-only SELD with SDE (SELDDE) task in this report. We propose to improve the existing ResNet-Conformer architectures with Squeeze-and-Excitation blocks in order to introduce additional forms of channel- and spatial-wise attention. In order to improve SELD performance, we also utilize the Spatial Cue-Augmented Log-Spectrogram (SALSA) features over the commonly used log-mel spectra features for polyphonic SELD. We complement the existing Sony-TAu Realistic Spatial Soundscapes 2023 (STARSS23) dataset with the audio channel swapping technique and synthesize additional data using the SpatialScaper generator. We also perform distance scaling in order to prevent large distance errors from contributing more towards the loss function. Finally, we evaluate our approach on the evaluation subset of the STARSS23 dataset.