Speech is the vital natural form of speaker communication speech is the one of the most important signal in which major information’s about speaker embedded in it. These signals have an abundant multi-variant spectral variation that convey gender, intonation, expression, words, emotional state and many other health condition of speaker. In speech analysis resonant frequency are usually extracted from the input speech signals. In this paper we perform Linear Predictive Coefficient (LPC) based algorithm for the estimation of fundamental frequency i.e., pitch and its successive frequency from speech signal which are considered as acoustic features of the speaker. Our point of concern is oriented among the five different vowels of Indian English i.e., /a/,/e/,/i/,/o/ and /u/ by plotting onto the spectrogram. From the Spectrogram different resonant components which are extracted from the major speakers from the southern part of the India. Vowel utterance made by speaker under consideration for 20 times, variations in obtained results for normal conditions of speaker. The author propose to investigate the variability as an parameter for speech recognition process the complete process. The complete process has been executed using Matlab.
Volume 11 | 02-Special Issue
Pages: 811-817