ASR for Tajweed Rules: Integrated with Self-Learning Environments

Published Online November 2017 in MECS DOI: 10.5815/ijieeb.2017.06.01

The Quran is the Holy book for the Muslims. The Quran contains guidance for life, which has to be applied by the Muslim people. In order to achieve this goal, it is important for the Muslims to understand the Quran clearly so they can be capable of applying it. Recitation is one of the Holy Quran related sciences. Previously, it was necessary to have a teacher of Quran and a student to meet face-to-face for the student to learn the recitation orally. It is the only certified way to guarantee that a certain verse of the Holy Quran is recited correctly.

Nowadays, because of the continuous increase and huge demand of people to learn the Quran, several organizations have started serving the learners by providing an online instructor in order to help them learn how to recite the verses of the Holy Quran correctly according to the rules of intonation (Tajweed science). Prophet Muhammad (peace be upon him) is the founder of the rules of this science, and certain Sahabah (companions of the prophet; may Allah be pleased with them) have learned from him the rules of pronunciation, and then those Sahabah have taught the second generation [1]. The process has continued up in the same manner till now.

The model proposed in this research facilitates teaching the recitation of the Holy Quran online so that students can practice Quran rules through using automatic speech recognition (ASR).

This approach of teaching has several benefits such as the delivery to individuals who cannot attend the Halaqat (sessions) held at the masjids (mosques), and it facilitates Quran teaching styles to receive more than the long-established model. The extreme importance of recitation and memorization of the Quran is due to the numerous benefits for readers and learners, as stated in Quran and Sunnah [2]. The learning of Quran is achieved by a qualified reciter (called sheikh qari) who has elder licensed linked to the transmission chain until it reaches the Messenger of Allah, Prophet Muhammad (peace be upon him). Detailed information about the Holy Quran and its sciences can be found in many resources; for example, see [3].

Due to the extensive use of the Internet and its availability, there is a strong need to develop a system that emulates the traditional way of the Quran teaching. There is some research that focuses on these issues, such as Miqra’ah, which is a server that uses virtually over the Internet.

The Holy Quran is written in the Arabic language, which is considered as a complex morphological language. From the perspective of ASR, the combination of letters is pronounced in the same way or different, depending on the Harakat used in upper and lower-case character [4]. Intrinsic motivation to develop ASR as participation to serve the Holy Quran sciences and its proposed approaches is needed to implement a system to correct the pronunciation mistakes and integrate it within a self-learning environment. Therefore, we suggest the use of the Model View Controller (MVC) as a base structure that helps in massive development such as this research. Phonetic Quran is a special case of Arabic phonemes where there is a guttural letter followed by any other letter. This case is called guttural manifestation.

Gutturalness, in Quran, relates to the quality of being guttural (i.e., producing a particular sound that comes from the back of the throat). The articulation of Quran emphatics affects adjacent vowels.

There are many commercial packages that are available, such as audio applications to recite (Tarteel) the Holy Quran. One among these packages is the Quran Auto Reciter (QAR) [5]; however, this application does not support the rules of Tajweed to verify and validate the Quranic recitation.

The field of speech recognition in Quranic voice recognition is a significant field, where the processing and acoustic model has a relation with Arabic phonemes and articulation of each word; thus, the research in the recitation of Quran could be taken from a different aspect.

In general, and especially in computer science, there are substantial research achieves meant to produce worthy results in the correction of the pronunciation of the Quran words according to the rules of Tajweed. Hassan Tabbal has done research on the topic of automated delimiters, which extracts ayah (verse) from an audio file and then converts verses of Quran into an audio file using the technology of speech recognition tools. The developed system depends on the framework of Sphinx IV [6].

Putra, Atmaja, and Prananto developed a learning system that used speech recognition for the recitation of the Quranic verses to reduce obstacles in learning the Quran and to facilitate the learning process. Their implementation depended on the Gaussian Mixture Model (GMM) and the Mel Frequency Cepstral (MFCC) features. The system produces good results for an effective and flexible learning process. The method of template referencing was used in that research [7]. Noor Jamaliah, in her master thesis in the field of speech recognition, used Mel Frequency Cepstral Coefficients (MFCC) for extracting feature from the input sound, and she used the Hidden Markov Model (HMM) for recognition and training purposes. The engine showed recognition rates that exceeded 86.41% (phonemes), and 91.95% (ayates) [8].

KAPD are available on 3 CDs for researchers [9]. A research on e-Halagat is demonstrated in [10]. Noureddine Aloui with other researchers used Discrete Walsh Hadamard Transform (DWHT), where the original speech is converted into stationary frames, and then applied the DWHT to the output signal. The performance is evaluated by using some objective criteria such as NRMSE, SNR, CR, and PSNR [11].

Nijhawan and Soni used the MFCC for feature extraction to build Speaker Recognition System (SRS) [12]. The training phase was done by calculating MFCC, executing VQ, finding the nearest neighbor using Euclidean distance, and then computing centroid and creating codebook for each speaker. After the completion of the training process, the testing phase is achieved through calculating MFCC, finding the nearest neighbor, finding minimum distance and then decision making.