Performance Based Cost Functions for End-to-End Speech Separation

TL;DR

This paper introduces new perceptually motivated loss functions for end-to-end speech separation, improving performance over traditional MSE by aligning better with human auditory perception.

Contribution

It proposes and evaluates novel loss functions based on perceptual metrics like SDR, SIR, SAR, and STOI for neural speech separation models.

Findings

Proposed loss functions outperform MSE in subjective listening tests.

Combining different perceptual metrics yields better separation results.

Flexible loss function design adapts to specific separation tasks.

Abstract

Recent neural network strategies for source separation attempt to model audio signals by processing their waveforms directly. Mean squared error (MSE) that measures the Euclidean distance between waveforms of denoised speech and the ground-truth speech, has been a natural cost-function for these approaches. However, MSE is not a perceptually motivated measure and may result in large perceptual discrepancies. In this paper, we propose and experiment with new loss functions for end-to-end source separation. These loss functions are motivated by BSS\_Eval and perceptual metrics like source to distortion ratio (SDR), source to interference ratio (SIR), source to artifact ratio (SAR) and short-time objective intelligibility ratio (STOI). This enables the flexibility to mix and match these loss functions depending upon the requirements of the task. Subjective listening tests reveal that combinations of the proposed cost functions help achieve superior separation performance as compared to stand-alone MSE and SDR costs.