A Wavelet Transform Module for a Speech Recognition Virtual Machine (Abstract Only)
Euisung Kim · 2016
This work explores the trade-offs between time and frequency information during the feature extraction process of an automatic speech recognition system using wavelet transform features instead of Mel-frequency cepstral coefficients. The Speech Recognition Virtual Kitchen toolkit (www.speechkitchen.org) is used as the framework for implementing wavelet modules in a virtual machine loaded with the Kaldi recognition system. The SRVK toolkit is a computing resource that provides virtual machines for a variety of research and education purposes. Results comparing different wavelets and feature extraction approaches will be presented. The resulting virtual machine, which allows straightforward comparisons of signal processing approaches, is freely available for research and educational use.