Few-Shot Keyword Spotting With Prototypical Networks
Abstract
Recognizing a particular command or a keyword, keyword spotting has been widely used in many voice interfaces such as Amazon’s Alexa and Google Home. In order to recognize a set of keywords, most of the recent deep learning based approaches use a neural network trained with a large number of samples to identify certain pre-defined keywords. This restricts the system from recognizing new, user-defined keywords. Therefore, we first formulate this problem as a few-shot keyword spotting and approach it using metric learning. To enable this research, we also synthesize and publish a Few-shot Google Speech Commands dataset. We then propose a solution to the few-shot keyword spotting problem using temporal and dilated convolutions on prototypical networks. Our comparative experimental results demonstrate keyword spotting of new keywords using just a small number of samples.
中文速览
关键词识别（Keyword Spotting）是智能音箱、语音助手等设备的核心功能，但现有深度学习方案需要大量训练样本，无法识别用户自定义的新词。为此，本文将关键词识别重新定义为一个"少样本学习"问题，借助度量学习框架——原型网络（Prototypical Networks）——让系统仅凭每个新关键词的极少量样本（1到5个）就能完成识别。在网络结构上，作者提出了结合时序卷积与空洞卷积的嵌入网络TD-ResNet7，并基于谷歌语音命令数据集整理发布了一套专门用于少样本关键词识别的基准数据集，涵盖背景噪声、未知词和静音等真实场景。实验结果表明，在2分类、每类5个样本的设定下，该方案准确率可达约94%，且在绝大多数场景下优于各基线方法，为在边缘设备上实现个性化、低资源的关键词识别提供了切实可行的新思路。
原文 arXiv:2007.14463;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2007.14463v1