Implementation of automatic speaker recognition on TMS320C6713 using MFCC

Speech recognition is a major topic in speech signal processing. Speech recognition is considered as one of the most popular and reliable biometric technologies used in automatic personal identification systems. Speech recognition systems are used for variety of applications mainly where security is very important. For a reliable and high accuracy of speech recognition, simple and efficient representation methods are required. In this paper, incoming speech signal from speaker is blocks in to the frames. Then spectrum of each frame is obtained by using FFT and then MFCCs are calculated from spectrum. After computing the MFCCs, these coefficients are written in an array which represent trained vector of the speaker. To identify the speaker, the Euclidean distance between the trained vectors and the MFCCs is computed for each trained vector. The trained vector that produces the smallest Euclidean distance is identified as speaker.