Online Keyword Spotting with a Character-Level Recurrent Neural Network

TL;DR

This paper introduces a character-level RNN for real-time keyword spotting in continuous speech, eliminating the need for phonetic transcriptions and enabling easy keyword updates, with superior performance and low latency.

Contribution

It presents a novel end-to-end context-aware keyword spotting model using a character-level RNN trained with CTC, allowing flexible keyword addition without retraining.

Findings

Outperforms DNN and HMM-based models in accuracy

Operates with low latency on streaming audio

Requires less computation than traditional models

Abstract

In this paper, we propose a context-aware keyword spotting model employing a character-level recurrent neural network (RNN) for spoken term detection in continuous speech. The RNN is end-to-end trained with connectionist temporal classification (CTC) to generate the probabilities of character and word-boundary labels. There is no need for the phonetic transcription, senone modeling, or system dictionary in training and testing. Also, keywords can easily be added and modified by editing the text based keyword list without retraining the RNN. Moreover, the unidirectional RNN processes an infinitely long input audio streams without pre-segmentation and keywords are detected with low-latency before the utterance is finished. Experimental results show that the proposed keyword spotter significantly outperforms the deep neural network (DNN) and hidden Markov model (HMM) based keyword-filler model even with less computations.