%0 Conference Paper
%A Zulkifly, M.A.A.
%A Yahya, N.
%D 2017
%F scholars:8038
%I Institute of Electrical and Electronics Engineers Inc.
%K Eigenvalues and eigenfunctions; Extraction; Feature extraction; Manufacture; Robotics; Signal analysis; Singular value decomposition; Speech; Speech communication; Speech recognition, Eigen decomposition; Malay words; Oriented energy; Principal vectors; Rasta-plp; STFT, Audio signal processing
%P 1-5
%R 10.1109/ROMA.2017.8231833
%T Relative spectral-perceptual linear prediction (RASTA-PLP) speech signals analysis using singular value decomposition (SVD)
%U https://khub.utp.edu.my/scholars/8038/
%V 2017-D
%X Speech recognition system has application in many areas such as customer call centers and as a medium in helping those with learning disabilities. There are three main stages in speech recognition which are signal analysis, feature extraction and modeling. Feature extraction plays an important role in speech recognition system and good speech feature extraction technique will allow the systems to accurately identify spoken words. In this work, we present the analysis of speech signals extracted by RASTA-PLP technique using eigenvalue decomposition. Using some Malay words having one, two and three syllables, we analyze the distance of RASTA-PLP speech signals between two same words and two different words. The distance between two words for RASTA-PLP was calculated using distance between principal eigenvectors. Results from this work shows that compared to STFT, RASTA-PLP's main signal features can be represented by smaller number of vectors. This characteristic gives an advantage of lower computational burden for RASTA-PLP. © 2017 IEEE.
%Z cited By 8; Conference of 3rd IEEE International Symposium in Robotics and Manufacturing Automation, ROMA 2017 ; Conference Date: 19 September 2017 Through 21 September 2017; Conference Code:134001