System login authentication using voice recognition and ai methods / Metrius Benedict Sunggoh
This thesis will show how the voice recognition technique can be combined with template matching technique (Vector Quantization) to validate a login process. This system will be divided into two modules; Enrollment and Verification During the Enrollment process,the person offers a 'live sample&...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2003
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/9130/ http://studentsrepo.um.edu.my/9130/1/System_login_authentication_using_voice_recogniyion.pdf |
| Summary: | This thesis will show how the voice recognition technique can be combined with template matching technique (Vector Quantization) to validate a login process. This system will be divided into two modules; Enrollment and Verification During the Enrollment process,the person offers a 'live sample' of his/her voice.The feature of this voice will be extracted using MFCC technique which involves Frame Blocking, Windowing, Fast Fourier Transform, Mel-frequency Wrapping and Cepstrum. The feature extracted by the MFCC will be processed and stored as codebooks. A codebook is created for each user. To confirm the identity at a future time, the individual will present the 'live sample', which will be matched
against the stored template (codebook). Vector Quantization technique will be used to compare the codebook and the feature from the voice provided by the user during login session. |
|---|