Dataset · pain
1000 hours of Arabic YouTube speech
pain/MASC
Multidialectal Arabic audio corpus of 1000 hours for training speech recognition (ASR) models.
What's inside
Contains 1000 hours of 16 kHz Arabic audio recordings. Materials are taken from over 700 YouTube channels, covering various regions, genres, and dialects. The dataset was created specifically for advancing speech technologies and Arabic speech recognition.
What you can do with it
- Training or fine-tuning ASR models for the Arabic language
- Testing speech recognition across different Arabic dialects
- Tasks
- Automatic speech recognition
- Languages
- Arabic
- Downloads this month
- 1.5K
- Likes
- 12
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("pain/MASC")