Main Page
About FCIT
History
Strategy
Administration>
Current Administration
Prevouis Administration
Organization Strucutre
Industrial Advisory Board
PhotoAlbum
Lab Guides
Departments
Computer Science
Information Technology
Information Systems
Academics
Bachelor Programs
Graduate Programs
Executive Programs
Academic Calendar
Admission
Bachelor Degree & Transferring
Admission from the Foundation Year
Transferring to the Faculty
Graduate Studies
Graduate Programs
Executive Programs
Scientific Research
Groups and Units
Research Groups
Research Interests
Distinguished Scientists Program
Faculty Journal
Faculty and Staff
Faculty
CS Department
IT Department
IS Department
Staff
Accreditation Integration & Management System (AIM
Development and Quality Unit
Work at FCIT
Capabilities Under the Spotlight
Code of Ethics
Students
Bachelor
ِAcademic Services
Preparatory Year Courses
Students' Guide
Academic Advising
Laboratories and Facilities
Student rights and duties
Graduate
Polices and Regulations
Students' Guide
Student's Handbook
New Student Orientation
Templates of proposals and theses for masters and
Courses
CS Program
IT Program
IS Program
Alumni Registration
Students Activities
Entrepreneurship Club
Cybersecurity Club
Data Science Club
Programming Club
Community
Industrial partnerships
Cisco Academy
Microsoft Academy
Oracle Academy
Files
Researches
Contact Us
عربي
English
About
Admission
Academic
Research and Innovations
University Life
E-Services
Search
Faculty of Computing and Information Technology
Document Details
Document Type
:
Article In Conference
Document Title
:
PROCESSING OF ARABIC SPEECH USING MULTI-LEVEL
تحليل الكلام العربي باستخدام التحول المويجي متعدد المستويات
Document Language
:
English
Abstract
:
In this paper the effect of using the coefficients of the low frequency filters at different levels of the wavelet transform of an Arabic speech samples is presented. Specifically, the discrete wavelet transform is computed to approximate a speech signal at different wavelet levels in preparation for speech processing, or transmission. The low frequency wavelet coefficients at each level of the decomposition are used as a representation for the Arabic speech signal in the wavelet domain. Then, the auto-correlation of the resulted low-frequency coefficients is computed to find out the effect of the wavelet level on the peak of the correlation of the processed Arabic speech signal. Using this method, cross-correlating the entire unknown speech sequence with a known speech sample, in order to recognize the spoken speech, is avoided. Also, the time taken for recognizing or transmitting a speech sample can be decreased due to the reduction in the sequence length that represents the speech sample. The results showed that Arabic speech samples are identified satisfactorily to 1/6.25 resolution. The quality of the Arabic speech samples is affected by the level of the wavelet transformation. The experiments showed that the wavelet transform can be carried out to the third level of the wavelet decomposition with an acceptable audition quality.
Conference Name
:
The Third Symposium on Scientific Research and Technological Development Outlook in The Arab World
Duration
:
From : 21 صفر 1424 AH - To : 24 صفر 1424 AH
From : 11 أبريل -2004 AD - To : 14 أبريل -2004 AD
Publishing Year
:
1425 AH
2004 AD
Number Of Pages
:
11
Article Type
:
Article
Conference Place
:
Riyadh, Saudi Arabia.
Organizing Body
:
KING ABDULAZIZ City for Science and Technology, Riyadh
Added Date
:
Saturday, June 23, 2012
Researchers
Researcher Name (Arabic)
Researcher Name (English)
Researcher Type
Dr Grade
Email
عبدالله احمد باسهيل
Basuhail, Abdullah Ahmad
Investigator
Doctorate
abasuhail@yahoo.com
يوسف العتيبي
Al-Otaibi, Youssef
Researcher
Files
File Name
Type
Description
33728.pdf
pdf
Back To Researches Page