An Anchor-Free Detector for Continuous Speech Keyword Spotting

TL;DR

This paper introduces AF-KWS, an anchor-free deep learning model for continuous speech keyword spotting, treating it as a one-dimensional object detection problem, and demonstrates its superior performance on new benchmark datasets.

Contribution

The paper proposes a novel anchor-free detection approach for CSKWS, including an auxiliary class for background exclusion, and establishes new benchmark datasets for evaluation.

Findings

AF-KWS outperforms existing methods significantly.

Introduces two new benchmark datasets for CSKWS.

Achieves high accuracy in continuous speech keyword detection.

Abstract

Continuous Speech Keyword Spotting (CSKWS) is a task to detect predefined keywords in a continuous speech. In this paper, we regard CSKWS as a one-dimensional object detection task and propose a novel anchor-free detector, named AF-KWS, to solve the problem. AF-KWS directly regresses the center locations and lengths of the keywords through a single-stage deep neural network. In particular, AF-KWS is tailored for this speech task as we introduce an auxiliary unknown class to exclude other words from non-speech or silent background. We have built two benchmark datasets named LibriTop-20 and continuous meeting analysis keywords (CMAK) dataset for CSKWS. Evaluations on these two datasets show that our proposed AF-KWS outperforms reference schemes by a large margin, and therefore provides a decent baseline for future research.