Boosting Tail Neural Network for Realtime Custom Keyword Spotting
This work addresses the industrial problem of efficient keyword spotting with limited computational resources, though it appears incremental in nature.
The paper tackles the challenge of realtime custom keyword spotting by proposing a Boosting Tail Neural Network, which achieves an 18% relative improvement in wakeup rate and false alarm compared to traditional single-classifier methods.
In this paper, we propose a Boosting Tail Neural Network (BTNN) for improving the performance of Realtime Custom Keyword Spotting (RCKS) that is still an industrial challenge for demanding powerful classification ability with limited computation resources. Inspired by Brain Science that a brain is only partly activated for a nerve simulation and numerous machine learning algorithms are developed to use a batch of weak classifiers to resolve arduous problems, which are often proved to be effective. We show that this method is helpful to the RCKS problem. The proposed approach achieve better performances in terms of wakeup rate and false alarm. In our experiments compared with those traditional algorithms that use only one strong classifier, it gets 18\% relative improvement. We also point out that this approach may be promising in future ASR exploration.