Small-footprint Keyword Spotting Using Deep Neural Network and Connectionist Temporal Classifier
Abstract
Mainly for the sake of solving the lack of keyword-specific data, we propose one Keyword Spotting (KWS) system using Deep Neural Network (DNN) and Connectionist Temporal Classifier (CTC) on power-constrained small-footprint mobile devices, taking full advantage of general corpus from continuous speech recognition which is of great amount. DNN is to directly predict the posterior of phoneme units of any personally customized key-phrase, and CTC to produce a confidence score of the given phoneme sequence as responsive decision-making mechanism. The CTC-KWS has competitive performance in comparison with purely DNN based keyword specific KWS, but not increasing any computational complexity.
中文速览
关键词唤醒（Keyword Spotting, KWS）系统必须跑在算力极为有限的移动端，而最大的拦路虎是针对特定唤醒词的标注数据严重不足、采集成本高昂。作者提出用深度神经网络（DNN）结合联结主义时序分类器（CTC）来绕开这一瓶颈：先用海量通用连续语音识别语料训练DNN预测音素后验概率，再用CTC对目标唤醒词的音素序列打置信度分数，整个流程完全不依赖关键词专属数据，任何用户自定义的触发词只需查词典拆成音素序列即可使用。在普通话唤醒词实验中，CTC-KWS在参数量和计算量与基线Deep-KWS基本持平的前提下，误报率和漏报率均优于后者，若再用少量唤醒词数据做自适应微调，性能还能进一步提升。这项工作的意义在于：它将唤醒词系统从"数据驱动、专词专练"的高成本模式解放出来，让低功耗设备上的个性化唤醒词部署变得既灵活又经济可行。
原文 arXiv:1709.03665;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1709.03665v1