Feature extraction through spectral analysis for speech recognition | |
| Author | Hasan, A. K. M. Rashidul |
| Call Number | AIT RSPR no. TC-95-14 |
| Subject(s) | Spectral analysis (Phonetics) Speech perception |
| Note | A research study submitted in partial fulfillment of .the requirements for the deg1·ee of Master of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | This study presents a method for extracting speech features through spectral analysis for speech recognition purposes. After a brief overview of all the major spectral analysis techniques available, LPC-Derived Bilinearly Transformed Cepstral Coefficients have been proposed as the working coefficients. A program has been developed in MATLAB environment for implementing the proposed algorithm and using real Thai speech (vowel parts of Thai syllables 'THI' and 'KHA:'), two codevectors are calculated. Through some calculations (e.g., mean or centroid, distance etc.), threshold values have been set for evaluating the performance of the generated codevectors. |
| Year | 1995 |
| Type | Research Study Project Report (RSPR) |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Telecommunications (TC) |
| Chairperson(s) | Makelainen, Kimmo |
| Examination Committee(s) | Ahmed, Kazi Mohiuddin ;Erke, Tapio J. |
| Scholarship Donor(s) | Finnish International Development Agency (FINNIDA), Finland. |
| Degree | Research Studies Project Report (M. Eng.) - Asian Institute of Technology, 1995 |