Dissertation/ Thesis
A Research of Multi-Glottal Micro-motion Signal on SpeakerIdentification
| العنوان: | A Research of Multi-Glottal Micro-motion Signal on SpeakerIdentification |
|---|---|
| Alternate Title: | 多種咽喉微振動訊號在語者識別的探討 |
| المؤلفون: | En-Yi Lin, 林恩義 |
| Thesis Advisors: | Sung-nien Yu, 余松年 |
| سنة النشر: | 2008 |
| المجموعة: | National Digital Library of Theses and Dissertations in Taiwan |
| الوصف: | 96 Speaker identification is a kind of biological authentication technology, which uses personal characteristics to distinguish users. The performance of speaker identification is easily disturbed by environmental noises and personal health and emotional conditions. Traditional speaker identification system must apply recognition technologies in noiseless environments for high performance. This thesis aims to use novel vocal signal to assist traditional speaker identification system. This study stems from the fact that human sound is caused by the vibration of vocal cord. Therefore, we extract both the features from sounds glottal motions and compare their effects in speaker identification. The glottal motion was proven to be a representational signal for human identification. The accuracy of speaker identification based on vocal signals is higher than that of microphone signals. A high accuracy of 100% was obtained by using only vocal signal. The accuracy on vocal signals is better than microphone signal in the text-independent experiments, but becomes lower than microphone signals in experiments that changes the position of the electrodes. The accuracy by using microphone signal becomes lower in noisy environment, which reveals the fact that all signal have their drawbacks. Combining microphone and vocal signals results in superior performance even in noisy environments or when the electrode position has been changed. |
| Original Identifier: | 096CCU05442113 |
| نوع الوثيقة: | 學位論文 ; thesis |
| وصف الملف: | 66 |
| الاتاحة: | http://ndltd.ncl.edu.tw/handle/02850525855912961815 |
| رقم الانضمام: | edsndl.TW.096CCU05442113 |
| قاعدة البيانات: | Networked Digital Library of Theses & Dissertations |
| الوصف غير متاح. |