Continuous Arabic Speech Segmentation using FFT Spectrogram

Author

Awais, M.M. ; WaqasAhmad ; Masud, S. ; Shamail, S.

Author_Institution

Dept. of Comput. Sci., Lahore Univ. of Manage. Sci.

fYear

2006

fDate

Nov. 2006

Firstpage

1

Lastpage

6

Abstract

This paper describes a phoneme segmentation algorithm that uses fast Fourier transform (FFT) spectrogram. The algorithm has been implemented and tested for utterances of continuous Arabic speech of 10 male speakers that contain almost 2346 phonemes in total. The recognition system determines the phoneme boundaries and identifies them as pauses, vowels and consonants. The system uses intensity and phoneme duration for separating pauses from consonants. Intensity in particular is used to detect two specific consonants (/r/, /hf) when they are not detected through the spectrographic information. Segmentation accuracy of 95.39% for the overall system has been achieved

Keywords

fast Fourier transforms; natural languages; spectroscopy; speech recognition; continuous Arabic speech segmentation; fast Fourier transform spectrogram; phoneme boundaries; phoneme segmentation; recognition system; Computer science; Fast Fourier transforms; Frequency domain analysis; Spectrogram; Speech analysis; Speech processing; Speech recognition; Speech synthesis; Testing; Time domain analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Innovations in Information Technology, 2006

Conference_Location

Dubai

Print_ISBN

1-4244-0674-9

Electronic_ISBN

1-4244-0674-9

Type

conf

DOI

10.1109/INNOVATIONS.2006.301939

Filename

4085454