Language Independent Speaker Recognition based on pitch measurements

-The significant progress in speaker recognition is by the use of pitch parameters. The accuracy of speaker recognition to a known degree of confidence for forensic tasks is given by the pitch. The experiments performed discriminates an individual and gives the model of a speaker. The analysis is performed on the TIMIT database and the results are 87% effective and are verified by using PRAAT Tool. The ultimate aim of the present work is to develop the speaker model consisting of pitch and statistical measurement results to identify and verify the speaker. We evaluate these features for speaker identi£cation using TIMIT. Further the work is in progress to improve the speaker recognition accuracy by computing spectral moments for each spectrum. Keywords-Pitch, Pitch contour, Speaker recognition