Soft-Median Choice: An Automatic Feature Smoothing Method for Sound Event Detection

TL;DR

This paper introduces Soft-Median Choice (SMC), a novel feature smoothing method for sound event detection that enables automatic optimization of median filter lengths during training, improving detection accuracy.

Contribution

The paper proposes a differentiable soft-median function and a multi-channel feature smoothing framework, allowing median filter parameters to be learned end-to-end in SED systems.

Findings

Outperforms baseline by over 10% in Event-Based F1 Score

Automatically learns optimal feature smoothing during training

Enhances shared feature extraction across sound event frames

Abstract

In Sound Event Detection (SED) systems, the lengths of median filters for post-processing have never been optimized during training due to several problems. No gradient is received by the lengths so they cannot be learned during back-propagation. The median-filtering inserted in the models also causes block in gradient flowing and the smoothing process misleads the model by ignoring errors. To resolve these problems, we provide different channels of features smoothed to different extents along with the original feature, so the model can optimize the weights while cognizing all the errors. We then use a linear layer to integrate the results and produce a linear combination. We further design the soft-median function to dredge the gradient flow. The proposed framework is called Soft-Median Choice (SMC). Experiments show that the SMC block not only automatically smooths the features based on the training set, but also forces the model to extract common features shared by all the frames of a sound event. The performance of the proposed method outperforms the baseline by over 10% of Event-Based F1 Score (EBFS) in both the validation and the evaluation set, and also slightly outperforms the single model of the state-of-the-art SED system.