Abstracts Submitted: 510
Number of Users: 708
View Abstracts Submitted
Back to home Page
Speaking is the fundamental activity in everybody's life. It becomes a pleasure if spoken with our own voice. But many human beings have problem in speaking and those disorders are termed with many names based on the cause. Few to name are apraxia, dystheria, stammering and so on which are the problems caused because of the malfunctioning of our organs. The sound produced by malfunctioning of the organ is called as Disordered speech, There are cases were the larynx, the main part of our vocal systems is damaged and removed. Artificial Larynx Transducer(ALT), a small hand held device will be used by them and the speech produced by ALT is called Substituition voices. In this paper, an algorithm is developed to improve the quality of ALT speech. Initially the F0 contour, formant frequency deviations are used to categories the ALT speech and HE speech as well as used to train the system. Further subspace filtering technique is used for noise removal. The novelty of the system is to use optimum interpolation and decimation for making Principal Component Analysis to remove the noise. Multidimensional scaling(MDS) is used to estimate the distance vector between the training and test data set.
© Copyright 2017 All Rights Reserved