Enhancing Qur'anic recitation through machine learning: a predictive approach to Tajweed optimization

Abstract

The human voice is a powerful medium for conveying emotion, identity, and intellect. Arabic, as the language of the Qur'an, holds deep spiritual and linguistic importance. Reciting the Qur'an correctly involves following Tajweed rules, which ensure phonetic precision and aesthetic quality. However, mastering these rules is challenging due to complex pronunciation and articulation variations, often requiring expert guidance. Traditional learning methods lack personalized feedback, making it difficult for learners to identify and correct errors. With the rise of machine learning, new opportunities have emerged to support Qur’anic recitation through intelligent analysis of Tajweed patterns and error prediction. This study presents a predictive model that identifies Qur’an reciters using ensemble learning techniques. By incorporating deep learning models like gated recurrent units (GRUs), long short-term memory (LSTM), and recurrent neural network (RNN), the system effectively captures the vocal features unique to each reciter. The model achieves an accuracy rate of 88.57%, demonstrating its potential to support Qur’anic learning and preservation. Nonetheless, its performance may be affected by audio quality and limited training data diversity. To improve adaptability and robustness, future work will focus on enriching the dataset and optimizing the model to generalize better across a broader range of reciters.