A speaker identification system using wavelet transform power density 3 and neural networks

مقدم أطروحة جامعية

Abu Hilal, Tariq Ali Yaqub

مشرف أطروحة جامعية

al-Shalabi, Riyad

أعضاء اللجنة

Kanan, Ghassan Jaddu
al-Shaykh, Asim A. R.
Muaydi, Hasan

الجامعة

الأكاديمية العربية للعلوم المالية و المصرفية

الكلية

كلية نظم و تكنولوجيا المعلومات

القسم الأكاديمي

قسم نظم المعلومات الحاسوبية

دولة الجامعة

الأردن

الدرجة العلمية

دكتوراه

تاريخ الدرجة العلمية

2010

الملخص الإنجليزي

New Automatic Speaker Recognition (ASR) techniques are presented for speaker feature extraction and classification process, based on experimental and theoretical approaches those addressed this issue.

A combination between the Discrete Wavelet Transform (DWT) and logarithmic Power Spectrum Density (PSD) is used to illustrate the speaker formants, where DWT approximation sub-signals is represented and a low pass filter is utilized, which indicates low frequency signal levels.

This approach produces an additional efficient method in speaker recognition rate, and higher classification accuracy.

Feed Forward Back Propagation Neural Network (FFBPNN) and Correlation Coefficient factor are integrated and presented for features classification to locate the similarity between the processed signals.

The proposed methodology worked with excellent capability for features tracking and recognition for text dependant recordings, the system can be applied in password or PINs identification in any security system, adequate comparison was discussed.

Ultimately, the achieved results show high-quality performance in classification, which reaches about 99 % classification rate.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• الشبكات العصبية(الحاسبات الإلكترونية)

عدد الصفحات

98

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-307009