Speech & Audio
Nigerian English Speech Dataset
Catalogue entry for read Nigerian English. Audio files are not hosted.
Dataset description
Describes a 50-hour speech collection that does not exist as audio in this preview. The task runner uses on-device synthetic speech so workers can practise transcription without uploading anyone's voice.
- Creator
- AI HUB Data Team (demo)
- Records
- 50
- Format
- WAV + TXT (not hosted)
- Collection period
- Not collected
- Updated
- 2026-11-07
- Downloads of samples
- 0
Quality score 89/100
₦240,000 licence (demo price)
At 20% platform fee: creator ₦192,000, platform ₦48,000. This is a quote. Paystack is not connected.
Samples use a 10-minute grant. Full files are not hosted.
Quality breakdown
20% completeness + 18% accuracy + 14% consistency + 12% annotation agreement + 14% documentation + 12% duplicate health + 10% metadata.
Completeness
90%
Accuracy
92%
Consistency
88%
Duplicate rate
0.4%
Annotation agreement
90%
Documentation
84%
Metadata
80%
Data preview
Safe sample only. Restricted fields are not shown.
| utterance_id | transcript | region | duration_s |
|---|---|---|---|
| demo-1 | The maize market opens before sunrise. | Lagos | 4.2 |
Data card
- Purpose
- Automatic speech recognition for Nigerian English.
- Collection method
- Planned consented read-speech. Not started.
- Annotation method
- Orthographic transcript plus speaker region, when collected.
- Source
- Placeholder catalogue card.
- Size
- 50 catalogue target. Sample only is hosted.
- Fields
- utterance_id, transcript, region, duration_s
- Language
- Nigerian English
- Geography
- Nigeria
- Limitations
- No audio bytes are stored. Do not treat the hour count as inventory.
- Potential bias
- A real collection would need more than one region and age band.
- Recommended uses
- Product walkthrough of a speech listing.
- Prohibited uses
- Claiming a downloadable 50-hour corpus.
- Licence
- commercial
Versions
v0.9 · 2026-02-20
Card only. Audio storage not connected.
+0 / −0 · Documentation partial · Declared, no files
Report dataset
Use this for privacy, copyright, harmful content or a misleading card. Do not attach more personal data than the report needs.
You can also use the report desk.