Use of fractals for segmentation and recognition of speech | |
| Author | Senevirathne, Tissa R. |
| Call Number | AIT Thesis no. CS-91-10 |
| Subject(s) | Speech perception Fractals |
| Note | A thesis submitted in partial fulfillment of the requirements of the degree of Master of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | In this study a new approach to determine the Fractal Dimension of Speech wave forms is proposed, and compared with existing methods of Fractal Dimension determination. In addition the conceptual incompatibility of Box Counting method in determining the Fractal Dimension of Speech wave forms is discussed. Speech segmentation using Fractal Dimension is studied at both word level and phoneme level. Phoneme Recognition is studied in detail. Phonemes are first clustered using Fractal Dimensions and then individual phoneme group in each cluster is identified using eigen values of the covariance matrix of the Iterated Function System parameters c, d, f. Behavior of eigen values of the covariance matrix of the Iterated Function System parameters c,d,f is studied, and reasons are given for almost zero third eiegn value. Parameter dependency of Iterated Function Systems of speech wave forms are investigated and presented as a new concept for speech concatenation. |
| Year | 1991 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Other Field of Studies (No Department) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Bohez, Erik L.J. |
| Examination Committee(s) | Huynh, Ngoc Phien ;Surapong Auwatanamongkol ;Hall, Michel A |
| Scholarship Donor(s) | Asian Development Bank (ADB), Japan |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1991 |