Speaker identification on compactRIO

In order to prevent imposters from fraudulently acquiring the access privilege of an authorized person, reliable means of user authenticity are required. The biometric information in general and one's voice in particular are of such means; therefore a text-independent speaker identification system was built for this concern. What characterizes this paper is its implementation using LabVIEW on National Instruments CompactRIO, a control and acquisition system powered by reconfigurable I/O field-programmable gate array (FPGA) technology. The Mel-Frequency Cepstral Coefficients (MFCC) technique is used for feature extraction from the voice, Vector Quantization (VQ) technique based on the Linde-Buzo-Gray (LBG) algorithm for feature modeling, and the minimum distance classifier for the feature matching. The developed system identifies the user in runtime with high accuracy.