|
Document Type |
: |
Thesis |
|
Document Title |
: |
Visual Lip-Reading for Arabic Alphabets and Quranic Words using Deep Learning قراءة الشفاه المرئية للأبجدية العربية والكلمات القرآنية باستخدام التعلم العميق |
|
Subject |
: |
Faculty of Computing and Information Technology |
|
Document Language |
: |
Arabic |
|
Abstract |
: |
The continuing advances in deep learning have paved the way for several challenging ideas. One such idea is visual lip-reading, which has recently drawn many research interests. Lip-reading is often referred to as visual speech recognition, and it is the ability to understand and predict spoken speech based solely on lip movements without using sounds. Due to the lack of research studies on visual speech recog- nition for the Arabic language in general, and its absence in the Quranic research, this thesis aims to fill this gap. The work in this thesis introduces a new publicly available Arabic lip-reading dataset comprising 10490 videos captured from multi- ple viewpoints and data samples in two letter levels and in the word level based on the content and context of a Quranic study aid given in Al-Qaida Al-Noorania book. Furthermore, this work uses visual speech recognition of spoken Arabic letters (Arabic alphabets), Quranic disjoined letters, and Quranic words, mainly phonetic, as they are recited in the Holy Quran according to Al-Qaida Al-Noorania. It could further validate the correctness of pronunciation and, subsequently, assist people in correctly reciting the Quran. The new proposed dataset is used to train an effective pre-trained deep learning CNN model throughout transfer learning for lip-reading, achieving an average accuracy of 83% for all dataset categories. This work attempts to exploit the systematic nature of the Quranic recitation and intonation (the art of proper pronunciation of letters) by modeling them in the visual speech recognition task. Finally, the results of the most common view angles for lip-reading are com- pared from different aspects, and dataset collection consistency and challenges are discussed and concluded with several new promising trends for future work. |
|
Supervisor |
: |
Dr. Emad Sami Jaha |
|
Thesis Type |
: |
Master Thesis |
|
Publishing Year |
: |
1445 AH
2023 AD |
|
Added Date |
: |
Monday, November 13, 2023 |
|
Researchers
| ندأ فيصل الجهني | Aljohani, Nada Faisal | Researcher | Master | |
|