Quranic Recitation MFCC Spectrograms for 20-Class Reciter Recognition

This dataset was compiled to support research on automatic Quranic reciter identification using deep learning techniques. It comprises audio recordings of the Holy Quran recited by 20 distinct reciters, covering a diverse range of vocal styles and recitation characteristics. The raw audio files were preprocessed and converted into Mel-Frequency Cepstral Coefficients (MFCC) spectrogram images, enabling the application of image-based deep learning architectures, including Convolutional Neural Networks (CNNs), ensemble methods, and Vision Transformers (ViTs). The dataset contains approximately 11,000 MFCC spectrogram images distributed across 20 classes, with each class corresponding to one reciter. It was used in the experimental evaluation reported in Saber et al. (2024), published in Neural Computing and Applications. This dataset is intended to serve as a benchmark resource for researchers working on Quranic speech processing, speaker recognition, and Islamic audio analysis.