Neural ODE with Temporal Convolution and Time Delay Neural Networks for Small-Footprint Keyword Spotting
Audio and Speech Processing
2020-09-08 v2 Machine Learning
Sound
Abstract
In this paper, we propose neural network models based on the neural ordinary differential equation (NODE) for small-footprint keyword spotting (KWS). We present techniques to apply NODE to KWS that make it possible to adopt Batch Normalization to NODE-based network and to reduce the number of computations during inference. Finally, we show that the number of model parameters of the proposed model is smaller by 68% than that of the conventional KWS model.
Keywords
Cite
@article{arxiv.2008.00209,
title = {Neural ODE with Temporal Convolution and Time Delay Neural Networks for Small-Footprint Keyword Spotting},
author = {Hiroshi Fuketa and Yukinori Morita},
journal= {arXiv preprint arXiv:2008.00209},
year = {2020}
}
Comments
5 pages, 5 figures