Improving weakly supervised sound event detection with self-supervised auxiliary tasks

TL;DR

This paper introduces a novel shared encoder architecture with self-supervised auxiliary tasks and a two-step attention pooling mechanism to enhance weakly supervised sound event detection in noisy, low-data environments without pretraining.

Contribution

It proposes a new framework combining self-supervised auxiliary tasks and a two-step attention pooling for improved sound event detection without pretraining.

Findings

Outperforms benchmarks by up to 22.3% in noisy conditions

Effective in low SNR scenarios (0, 10, 20 dB)

Ablation confirms the auxiliary task and attention pooling benefits

Abstract

While multitask and transfer learning has shown to improve the performance of neural networks in limited data settings, they require pretraining of the model on large datasets beforehand. In this paper, we focus on improving the performance of weakly supervised sound event detection in low data and noisy settings simultaneously without requiring any pretraining task. To that extent, we propose a shared encoder architecture with sound event detection as a primary task and an additional secondary decoder for a self-supervised auxiliary task. We empirically evaluate the proposed framework for weakly supervised sound event detection on a remix dataset of the DCASE 2019 task 1 acoustic scene data with DCASE 2018 Task 2 sounds event data under 0, 10 and 20 dB SNR. To ensure we retain the localisation information of multiple sound events, we propose a two-step attention pooling mechanism that provides a time-frequency localisation of multiple audio events in the clip. The proposed framework with two-step attention outperforms existing benchmark models by 22.3%, 12.8%, 5.9% on 0, 10 and 20 dB SNR respectively. We carry out an ablation study to determine the contribution of the auxiliary task and two-step attention pooling to the SED performance improvement.