AQQD: Annotated Quranic Recitations from 308 Reciters

AQQD (Annotated Quranic Qira’at Dataset) is an open research dataset of Quranic recitations annotated across multiple canonical Qira’at styles. Version 1 includes 23,111 audio files from 308 reciters covering 70 selected Quranic Surahs, with Version 2 adding 1,072 files from a single reciter for controlled comparison. The dataset was created by Linda Smail and last updated on 2026-04-18.

Use Cases

• Train speech recognition models for Quranic Arabic based on audio files from multiple reciters.
• Analyze phonetic and linguistic variations across different Qira’at styles using the annotated metadata.
• Conduct controlled acoustic comparisons using the single-reciter dataset (R000) to minimize speaker variability.
• Develop models for fine-grained audio segmentation based on recordings of specific words or phrases where Qira’at differences occur.

Strengths

• Contains 23,111 audio files in its initial version, providing a substantial corpus for analysis.
• Includes recordings from 308 reciters, offering diversity in vocal characteristics.
• Version 2 adds a controlled dataset of 1,072 files from a single reciter, enabling isolated style comparison.
• Covers 70 selected Quranic Surahs with structured metadata linking recordings to reciter, style, and verse.

Limitations

• Row count and total size are unknown, which may limit suitability assessment for large-scale projects.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's geographic and demographic representativeness of reciters is not specified.

Provenance

Source

Linda Smail AQQD- Annotated Quranic Qira’at Dataset; Version 2 includes data sourced from Taibah University (Madinah, Saudi Arabia).

Collection Method

Audio recordings of Quranic recitations, annotated with structured metadata.

Freshness

Last updated 2026-04-18 09:23:50; freshness should be verified.

Geography

Likely global, with specific contribution from Madinah, Saudi Arabia for the R000 dataset.