Generating Adversarial Samples For Training Wake-up Word Detection Systems Against Confusing Words

TL;DR

This paper introduces methods to generate adversarial confusing samples to improve wake-up word detection systems' robustness against similar-sounding words, addressing a key challenge in real-world applications.

Contribution

It proposes novel techniques for creating adversarial confusing samples and a domain embedding approach, enhancing wake-up word detection robustness without requiring real confusing samples.

Findings

Generated adversarial samples improve detection robustness

The approach performs well in both normal and confusing scenarios

A new confusing words testing database HI-MIA-CW is released

Abstract

Wake-up word detection models are widely used in real life, but suffer from severe performance degradation when encountering adversarial samples. In this paper we discuss the concept of confusing words in adversarial samples. Confusing words are commonly encountered, which are various kinds of words that sound similar to the predefined keywords. To enhance the wake word detection system's robustness against confusing words, we propose several methods to generate the adversarial confusing samples for simulating real confusing words scenarios in which we usually do not have any real confusing samples in the training set. The generated samples include concatenated audio, synthesized data, and partially masked keywords. Moreover, we use a domain embedding concatenated system to improve the performance. Experimental results show that the adversarial samples generated in our approach help improve the system's robustness in both the common scenario and the confusing words scenario. In addition, we release the confusing words testing database called HI-MIA-CW for future research.