DONUT: CTC-based Query-by-Example Keyword Spotting

TL;DR

DONUT is a low-resource, CTC-based query-by-example keyword spotting system that enables personalized wakeword detection on embedded devices without cloud data upload.

Contribution

It introduces a novel CTC-based algorithm for online query-by-example keyword spotting that combines user adaptation with low computational requirements.

Findings

Effective personalized wakeword detection with few training examples

Operates efficiently on embedded systems without cloud data

Achieves high interpretability and generalization

Abstract

Keyword spotting--or wakeword detection--is an essential feature for hands-free operation of modern voice-controlled devices. With such devices becoming ubiquitous, users might want to choose a personalized custom wakeword. In this work, we present DONUT, a CTC-based algorithm for online query-by-example keyword spotting that enables custom wakeword detection. The algorithm works by recording a small number of training examples from the user, generating a set of label sequence hypotheses from these training examples, and detecting the wakeword by aggregating the scores of all the hypotheses given a new audio recording. Our method combines the generalization and interpretability of CTC-based keyword spotting with the user-adaptation and convenience of a conventional query-by-example system. DONUT has low computational requirements and is well-suited for both learning and inference on embedded systems without requiring private user data to be uploaded to the cloud.