The implementation of phoneme recognition system using multi wavelet transform and artificial neural networks

العناوين الأخرى

تطبيق نظام تمييز الألفاظ باستخدام التحويل المويجي و الشبكات العصبية

مقدم أطروحة جامعية

Abu Tabikh, Mayy Abd al-Munim S. Rahma

مشرف أطروحة جامعية

al-Jawhar, Walid Amin Mahmud

أعضاء اللجنة

Khazim, L.
Mahdi, S.

الجامعة

جامعة بغداد

الكلية

كلية العلوم

القسم الأكاديمي

قسم علوم الحاسبات

دولة الجامعة

العراق

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2004

الملخص الإنجليزي

There are several advantages of Phoneme recognition.

It is easier to use speech for data entrance than other tools.

It allows writing userfriendly data entrance programs.

There are several difficulties in speech recognition.

One of these difficulties is noise.

Variability in speech is another problem.

Even the speech of same speaker varies.

The data used in this project are Arabic phonemes stored as 8-bit mono 8000Hz PCM WAVE Sound file.

The most important phase is feature extraction which is done by the Multiwavelet transform ,where the signal is framed into 256 samples each then windowed using the Hamming window finally the feature extraction which results 8 blocks of 128 sample and the extracted features are in the Low-Low block (the first block) .

Two Systems for Phoneme Recognition are proposed the first depends on the Euclidian differences of the coefficients of the Multiwavelet Transform.

The second depends on Artificial Neural Networks as decision making algorithm to find the best match for the tested phonemes.

The research showed that the accuracy of the first proposed system is 85% while the second recognizes the phonemes 98% efficiently.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

عدد الصفحات

66

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-605016