The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
Darija Consent-Verified Speech Sample
Overview
A small, high-quality sample of Moroccan Darija (Arabic dialect) speech audio paired with clean, human-reviewed transcriptions. Released as a preview of a larger, growing corpus available for commercial licensing.
Why this dataset is different
Most Darija audio data on the Hub is either scraped without clear consent, mislabeled, or corrupted (broken uploads, empty splits). This sample is:
- Consent-verified: every speaker explicitly agreed to have their voice used for AI training purposes.
- Human-reviewed transcription: not raw ASR output — corrected by a native Darija speaker.
- Metadata-rich: gender, approximate region, and recording context included where available.
Dataset details
- Duration: ~Coming soon
- Speakers: Coming soon
- Format: WAV audio + text transcription (CSV/JSONL)
Intended use
Evaluating Darija ASR/TTS model performance, benchmarking, and as a preview before licensing the full dataset.
Full dataset access
This dataset card is live ahead of the first audio upload. Sample audio + transcriptions will be added within the coming days. This is a limited preview. The full corpus (growing weekly, consent documentation available on request) is available for licensing.
Contact: hello@darijadata.com Website: darijadata.com
- Downloads last month
- 6