Fact · Languages and accents
The MGB-3 Arabic challenge data totals 16 hours of video, split evenly across seven genres.
Checked against its source on October 17, 2026
In the source's own words
“A total of 16 hours of videos, split evenly across the different genres”
Speech Recognition Challenge in the Wild: Arabic MGB-3, Ahmed Ali et al., arXiv, 2017, Abstract.
Open the original- What it measures
- The seven genres are comedy, cooking, family and kids, fashion, drama, sports and science. The paper divides the 16 hours into adaptation, development and evaluation sets.
- What it does not show
- Sixteen hours is small compared with the training data of modern speech models, which is why this set is useful for testing rather than for training.