Exploring the Best Loss Function for DNN-Based Low-latency Speech Enhancement with Temporal Convolutional Networks

TL;DR

This paper compares different loss functions and methods for DNN-based speech enhancement, proposing a new STFT-based approach with PASE features and a low-latency TasNet variant, demonstrating superior performance on multiple datasets.

Contribution

It introduces a novel STFT-based loss function with PASE features and a low-latency TasNet model, advancing DNN-based speech enhancement techniques.

Findings

Proposed STFT-based method with PASE features improves subjective quality on small datasets.

Low-latency TasNet achieves excellent performance in the DNS Challenge.

Methods compare favorably to state-of-the-art approaches.

Abstract

Recently, deep neural networks (DNNs) have been successfully used for speech enhancement, and DNN-based speech enhancement is becoming an attractive research area. While time-frequency masking based on the short-time Fourier transform (STFT) has been widely used for DNN-based speech enhancement over the last years, time domain methods such as the time-domain audio separation network (TasNet) have also been proposed. The most suitable method depends on the scale of the dataset and the type of task. In this paper, we explore the best speech enhancement algorithm on two different datasets. We propose a STFT-based method and a loss function using problem-agnostic speech encoder (PASE) features to improve subjective quality for the smaller dataset. Our proposed methods are effective on the Voice Bank + DEMAND dataset and compare favorably to other state-of-the-art methods. We also implement a low-latency version of TasNet, which we submitted to the DNS Challenge and made public by open-sourcing it. Our model achieves excellent performance on the DNS Challenge dataset.