1Assistant Professor, Department of Electronics and Communication Engineering, Cambridge Institute of Technology, Ranchi-835103, Jharkhand, India
2HOD, Department of Electronics and Communication Engineering, Jaipur National University, Jaipur-302017, Rajasthan, India
3HOD, Department of Electronics and Communication Engineering, Cambridge Institute of Technology, Ranchi-835103, Jharkhand, India
4Research Scholar, Department of Electronics and Communication Engineering, Jaipur National University, Jaipur-302017, Rajasthan, India
*Corresponding author email id: shri.kant.yay@gmail.com
Online published on 2 August, 2016.
Speech is one means of interacting for a human species in this Earth. In addition, to make interaction efficient, robust system must have high rate of recognition capability under adverse acoustic condition. In this paper, we will first give a brief overview of speech verification and identification, and then some of the important extraction techniques that will help to get it out its feature. The different model that will help in term of algorithm analysis will be discussed. We will study and try to compare the various extraction techniques that will help to make the recognition process efficient which includes perceptual linear predictive, linear predictive cepstral coefficient, mel-frequency cepstral coefficient, linear predictive coding, Fast Fourier transform technique, mel-frequency cepstrum co-efficient and relative spectral transform, and also try to identify the suitable technique for practical or real-time applications and suggest the suitable technique for different voice recognition patterns and its applications.
Various feature extraction method, Gaussian mixture model, Speaker verification identification, Acoustic condition and Behaviour
