Buraaq/quran-md-ayahs

Quran-MD - Ayah Level

This dataset is part of the complete Quran-MD dataset available here: Complete Quran-MD

Paper

Quran-MD: A Fine-Grained Multimodal Dataset of the Quran (Accepted at: 5th Muslims in ML Workshop co-located with NeurIPS 2025)

📄 Paper Link: quran-md-paper

Abstract

We present Quran-MD, a comprehensive multimodal dataset of the Qur’an that integrates textual, linguistic, and audio dimensions at the verse and word levels. For each verse (ayah), the dataset provides its original Arabic text, English translation, and phonetic transliteration. To capture the rich oral tradition of Qur’anic recitation, we include verse-level audio from 30 distinct reciters, reflecting diverse recitation styles and dialectical nuances. At the word level, each token is paired with its corresponding Arabic script, English translation, transliteration, and an aligned audio recording, allowing fine-grained analysis of pronunciation, phonology, and semantic context. This dataset supports various applications, including natural language processing, speech recognition, text-to-speech synthesis, linguistic analysis, and digital Islamic studies. Bridging text and audio modalities across multiple reciters, this dataset provides a unique resource to advance computational approaches to Qur’anic recitation and study. Beyond enabling tasks such as ASR, tajweed detection, and Qur’anic TTS, it lays the foundation for multimodal embeddings, semantic retrieval, style transfer, and personalized tutoring systems that can support both research and community applications.

Dataset Description

This dataset contains 187,080 verse-level audio recordings of the complete Quran with multiple recitations. Each sample includes a complete ayah (verse) recitation along with Arabic text, English translation, and transliteration metadata.

Dataset Summary

• 🎵 Audio Samples: 187,080 MP3 files (complete verse recitations)
• 🎤 Reciters: 30 renowned Quranic reciters
• 📚 Coverage: Complete Quran (114 surahs, 6,236 unique ayahs)
• 🔢 Reciter Coverage: Each ayah is recited by all 30 reciters (complete coverage).
• 🌐 Languages: Arabic (original), English (translation), Transliteration
• 🎧 Audio Quality: Various bitrates (32kbps to 192kbps)
• ⏱️ Duration: Approximately 450+ hours of verse-level audio

Supported Tasks

• 🎙️ Automatic Speech Recognition (ASR): Arabic speech recognition training
• 🗣️ Text-to-Speech (TTS): Arabic text-to-speech synthesis
• 👤 Speaker Recognition: Multi-speaker voice identification (30 reciters)
• 📖 Religious Studies: Quranic recitation research and analysis
• 🎓 Language Learning: Arabic pronunciation and recitation learning
• 📊 Audio Analysis: Verse-level prosody and intonation studies

Languages

• Arabic (Classical Arabic - Quranic Arabic)
• English (translations)
• Transliteration (romanized Arabic)

Dataset Structure

Each sample contains complete verse-level information:

{
    "surah_id": int32,           # Surah (chapter) number (1-114)
    "ayah_id": int32,            # Ayah (verse) number within surah
    "surah_name_ar": string,     # Surah name in Arabic
    "surah_name_en": string,     # Surah name in English  
    "surah_name_tr": string,     # Surah name transliterated
    "ayah_count": int32,         # Total ayahs in this surah
    "ayah_ar": string,           # Complete ayah text in Arabic
    "ayah_en": string,           # Ayah translation in English
    "ayah_tr": string,           # Ayah transliteration
    "reciter_id": string,        # Unique reciter identifier
    "reciter_name": string,      # Reciter's name and audio quality info
    "audio": Audio(),            # Audio file (MP3 format)
}

Sample Data Example

{
    "surah_id": 1,
    "ayah_id": 1, 
    "surah_name_ar": "الفاتحة",
    "surah_name_en": "Al-Fatiha",
    "surah_name_tr": "Al-Faatiha",
    "ayah_count": 7,
    "ayah_ar": "بِسْمِ اللَّهِ الرَّحْمَٰنِ الرَّحِيمِ",
    "ayah_en": "In the name of Allah, the Entirely Merciful, the Especially Merciful.",
    "ayah_tr": "Bismillaahi ar-rahmaani ar-raheem",
    "reciter_id": "alafasy",
    "reciter_name": "Mishary Alafasy - 64kbps",
    "audio": <Audio object>
}

Featured Reciters

The dataset includes complete Quran recitations from 30 renowned reciters:

Reciter	Style	Samples
Abdul Basit	Murattal	6,236
Abdullaah 3awwaad Al-Juhaynee	Murattal / Standard	6,236
Abdullah Basfar	Murattal / Standard	6,236
Abdul Samad	Murattal / Standard	6,236
Abdurrahmaan As-Sudais	Murattal / Standard	6,236
Abu Bakr Ash-Shaatree	Murattal / Standard	6,236
Alafasy	Murattal / Standard	6,236
Ali Jaber	Murattal / Standard	6,236
Ayman Sowaid	Murattal / Standard	6,236
Banna	Murattal / Standard	6,236
Fares Abbad	Murattal / Standard	6,236
Ghamadi	Murattal / Standard	6,236
Hani Rifai	Murattal / Standard	6,236
Hudhaify	Murattal / Standard	6,236
Husary	Murattal / Standard	6,236
Husary (Mujawwad)	Mujawwad	6,236
Husary (Teacher)	Teacher Recitation	6,236
Husary (Warsh)	Warsh	6,236
Ibrahim Akhdar	Murattal / Standard	6,236
Minshawy (Mujawwad)	Mujawwad	6,236
Minshawy (Murattal)	Murattal	6,236
Minshawy (Teacher)	Teacher Recitation	6,236
Mostafa Ismail	Murattal / Standard	6,236
Muhammad Jibreel	Murattal / Standard	6,236
Muhsin Al-Qasim	Murattal / Standard	6,236
Nasser Alqatami	Murattal	6,236
Saood Ash-Shuraym	Murattal / Standard	6,236
Tunaiji	Murattal / Standard	6,236
Yassin (Warsh)	Warsh	6,236
Yasser Ad-Dussary	Murattal	6,236

Total: 30 reciters × 6,236 ayahs = 187,080 audio samples

Usage Examples

Loading the Dataset

from datasets import load_dataset

# Load the Ayah-level dataset
dataset = load_dataset("Buraaq/quran-md-ayahs")

print(f"Dataset size: {len(dataset):,} ayah recordings")
print(f"Sample keys: {list(dataset[0].keys())}")

# Access first sample
first_sample = dataset[0]
print(f"Surah: {first_sample['surah_name_en']} ({first_sample['surah_name_ar']})")
print(f"Ayah: {first_sample['ayah_ar']}")
print(f"Translation: {first_sample['ayah_en']}")
print(f"Reciter: {first_sample['reciter_name']}")

Play Audio Sample

import IPython.display as ipd

# Play first ayah audio
first_ayah = dataset[0]
ipd.Audio(first_ayah["audio"]["array"], rate=first_ayah["audio"]["sampling_rate"])

Filter by Reciter

# Get all recitations by Mishary Alafasy
alafasy_recitations = dataset.filter(lambda x: x["reciter_id"] == "alafasy")
print(f"Alafasy has {len(alafasy_recitations):,} ayah recordings")

# Get unique reciters in dataset
reciters = set(dataset["reciter_id"])
print(f"Available reciters: {len(reciters)}")
for reciter in sorted(reciters):
    count = len(dataset.filter(lambda x: x["reciter_id"] == reciter))
    print(f"  {reciter}: {count:,} recordings")

Filter by Surah (Chapter)

# Get all ayahs from Surah Al-Fatiha (Chapter 1) across all reciters
al_fatiha = dataset.filter(lambda x: x["surah_id"] == 1)
print(f"Al-Fatiha: {len(al_fatiha)} recordings across {len(set(al_fatiha['reciter_id']))} reciters")

# Get specific surah with specific reciter
al_fatiha_alafasy = dataset.filter(
    lambda x: x["surah_id"] == 1 and x["reciter_id"] == "alafasy"
)
print(f"Al-Fatiha by Alafasy: {len(al_fatiha_alafasy)} ayahs")

Audio Duration Analysis

import librosa

# Analyze audio durations
durations = []
sample_size = 1000  # Analyze first 1000 samples for performance

for i in range(min(sample_size, len(dataset))):
    audio_array = dataset[i]["audio"]["array"]
    sample_rate = dataset[i]["audio"]["sampling_rate"]
    duration = len(audio_array) / sample_rate
    durations.append(duration)

print(f"Average duration: {sum(durations)/len(durations):.2f} seconds")
print(f"Min duration: {min(durations):.2f} seconds")
print(f"Max duration: {max(durations):.2f} seconds")

Create Training/Validation Splits

# Split by surah for train/validation
train_surahs = list(range(1, 100))  # Surahs 1-99 for training
val_surahs = list(range(100, 115))   # Surahs 100-114 for validation

train_data = dataset.filter(lambda x: x["surah_id"] in train_surahs)
val_data = dataset.filter(lambda x: x["surah_id"] in val_surahs)

print(f"Training samples: {len(train_data):,}")
print(f"Validation samples: {len(val_data):,}")

Citation

@misc{salman2026quranmdfinegrainedmultilingualmultimodal,
      title={Quran-MD: A Fine-Grained Multilingual Multimodal Dataset of the Quran}, 
      author={Muhammad Umar Salman and Mohammad Areeb Qazi and Mohammed Talha Alam},
      year={2026},
      eprint={2601.17880},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2601.17880}, 
}

Related Datasets

• 📝 Quran-MD - Word: Individual word pronunciations (77,429 samples)
• 🏠 Complete Quran-MD Overview: Full dataset collection documentation

Contact

• Hugging Face Profile: umarsalman
• GitHub Profile: umar1997
• Dataset Issues: Please open an issue on the repository.

Feel free to reach out for dataset issues, research collaborations, or community contributions.

---

"And We have certainly made the Quran easy for remembrance, so is there any who will remember?" - Quran 54:17