Using time and frequency domain information for the development of adaptive voice control devices

Abstract

A simple voice recognizer was implemented with the aim of building voice control devices by utilizing a cascade LPC with adaptive fuzzy interface system (FIS). The accuracy of the voice recognizer depends on the complexity of the voice commands, the number of frequency components used, and the weights that are utilized in FIS. The quality of the voice recognizer depends both on the time and the frequency resolution of the digitizing hardware. With a limited set of sounds the technique discussed in this paper can be used to produce a speaker dependent voice control system. The performance of the system improves with time due to the adaptive nature of the system. The highly flexible backend (fuzzy interface) enables usage of the device in noisy industrial environments.