Disentangled Training with Adversarial Examples For Robust Small-footprint Keyword Spotting

TL;DR

This paper introduces a datasource-aware disentangled learning approach with adversarial examples to enhance the robustness of small-footprint keyword spotting models across varying acoustic environments, achieving significant improvements in accuracy and false reject rates.

Contribution

It presents a novel adversarial training method that reduces data mismatch issues in keyword spotting, leading to more robust models for on-device speech recognition.

Findings

40.31% reduction in false reject rate at 1% false accept rate

Achieved 98.06% accuracy on Google Speech Commands V1 dataset

Improved robustness against unseen acoustic conditions

Abstract

A keyword spotting (KWS) engine that is continuously running on device is exposed to various speech signals that are usually unseen before. It is a challenging problem to build a small-footprint and high-performing KWS model with robustness under different acoustic environments. In this paper, we explore how to effectively apply adversarial examples to improve KWS robustness. We propose datasource-aware disentangled learning with adversarial examples to reduce the mismatch between the original and adversarial data as well as the mismatch across original training datasources. The KWS model architecture is based on depth-wise separable convolution and a simple attention module. Experimental results demonstrate that the proposed learning strategy improves false reject rate by $40.31%$ at $1%$ false accept rate on the internal dataset, compared to the strongest baseline without using adversarial examples. Our best-performing system achieves $98.06%$ accuracy on the Google Speech Commands V1 dataset.