Loud-loss: A Perceptually Motivated Loss Function for Speech Enhancement Based on Equal-Loudness Contours

TL;DR

This paper introduces a perceptually-weighted loss function for speech enhancement that uses equal-loudness contours to better align with human auditory perception, improving perceptual quality over traditional MSE.

Contribution

It proposes a novel psychoacoustically motivated loss function based on equal-loudness contours, enhancing speech enhancement models' perceptual performance.

Findings

Significant WB-PESQ score improvement from 2.17 to 2.93.

Loss function is model-agnostic and flexible.

Enhanced perceptual quality in speech enhancement.

Abstract

The mean squared error (MSE) is a ubiquitous loss function for speech enhancement, but its problem is that the error cannot reflect the auditory perception quality. This is because MSE causes models to over-emphasize low-frequency components which has high energy, leading to the inadequate modeling of perceptually important high-frequency information. To overcome this limitation, we propose a perceptually-weighted loss function grounded in psychoacoustic principles. Specifically, it leverages equal-loudness contours to assign frequency-dependent weights to the reconstruction error, thereby penalizing deviations in a way aligning with human auditory sensitivity. The proposed loss is model-agnostic and flexible, demonstrating strong generality. Experiments on the VoiceBank+DEMAND dataset show that replacing MSE with our loss in a GTCRN model elevates the WB-PESQ score from 2.17 to 2.93-a significant improvement in perceptual quality.