Graduate Project

Voice recognition system based on audio fingerprinting

Voice recognition is the ability of a machine or a program to receive and interpret dictation, or to understand the spoken words. In the recent years, an interest has been received by systems for audio fingerprinting based voice recognition systems which enable automatic content-based identification by extracting the audio signatures from the signal and matching them to the fingerprint of the word to be recognized. This project discusses the implementation of a hardware based, real-time voice recognition system which is capable of storing the fingerprints and later recognizing them for the three words of the user's choice. This is achieved by implementing band pass filters in the assembly language with fixed-point arithmetic onto to AT Mega32 microcontroller. The outputs of the filters are squared and accumulated. This approach helps in saving a lot of time for the speech sample to be processed for the frequency spectrum before the next samples comes in. The analysis of the voice is made by calculating the Euclidean distance between the saved and the current fingerprints of the words. This technique provides high success rate in recognizing the word. The algorithm for the voice recognition system is written in assembly and C programming language. The successful demonstration of the voice recognition system was presented on AT Mega16 + 232 kit with AT Mega32 microcontroller.

Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.