Photo Photo Photo Photo Photo Photo
Computer Science: Speech Recognition Based on Fractal Dimension


Speech Recognition Based on Fractal Dimension

Mohammed Sahib Mahdi

Collage of Science, Al-Nuhrain University. Baghdad-Iraq


In this paper, the concepts of fractal geometry were applied to extract the fractal dimension of the speech waveform or a partition of it By studying the behavior of computed fractal dimension for different speech frames, it is found that the fractal dimension takes a steady values for each relatively small set of samples "frames" which may belong to a syllable speech portion or a part of it. Therefore, a speech recognition system can be achieved depending on resulting successive values of fractal dimensions. A supervised classification technique was employed to distinguish the resulting fractal dimensions. The application program gave an encouraging classification results.



S5 Box




Fields marked with an asterisk (*) are required.