README.md

metadata

language:
  - ar
license: cc-by-4.0
task_categories:
  - automatic-speech-recognition
  - text-to-speech
pretty_name: Mantooq
size_categories:
  - 1M<n<10M
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/part-*.parquet
dataset_info:
  features:
    - name: audio_id
      dtype: string
    - name: audio
      dtype: audio
    - name: text
      dtype: string
    - name: duration
      dtype: int32
    - name: start_ms
      dtype: int32
    - name: end_ms
      dtype: int32
    - name: content_type
      dtype: string
    - name: tone
      dtype: string
    - name: speaker_id
      dtype: string
    - name: gender
      dtype: string
    - name: dialect
      dtype: string
    - name: book_name
      dtype: string
    - name: chapter
      dtype: string

Mantooq — Arabic Audiobook Dataset

~8,300 hours of validated Arabic audiobook audio across 3,940,000 segments from 19,737 chapters.

Dataset Details

• Language: Arabic (Modern Standard + dialects)
• Audio: 16kHz mono FLAC
• Segments: ~10 seconds each
• Validation: Each file validated with Whisper (mlx-whisper large-v3-turbo) against Gemini transcriptions

Access

This dataset is gated. Request access to download.