Multi-Scale Time-Frequency Attention for Acoustic Event Detection

TL;DR

This paper introduces a Multi-Scale Time-Frequency Attention module for Acoustic Event Detection, enabling models to focus on relevant time and frequency features at multiple resolutions, improving detection of events with varying scales.

Contribution

The paper presents a novel MTFA module that captures multi-scale time-frequency information, addressing limitations of previous attention methods in AED.

Findings

Achieves competitive results on DCASE 2017 datasets.

Effectively captures events with different temporal and frequential scales.

Enhances model focus on relevant time-frequency regions.

Abstract

Most attention-based methods only concentrate along the time axis, which is insufficient for Acoustic Event Detection (AED). Meanwhile, previous methods for AED rarely considered that target events possess distinct temporal and frequential scales. In this work, we propose a Multi-Scale Time-Frequency Attention (MTFA) module for AED. MTFA gathers information at multiple resolutions to generate a time-frequency attention mask which tells the model where to focus along both time and frequency axis. With MTFA, the model could capture the characteristics of target events with different scales. We demonstrate the proposed method on Task 2 of Detection and Classification of Acoustic Scenes and Events (DCASE) 2017 Challenge. Our method achieves competitive results on both development dataset and evaluation dataset.