ASALA: Authentic Spoken Arabic Language Archive

ASALA consists of Arabic speech-to-text transcriptions collected from publicly accessible conversational sources across the Middle East and North Africa (MENA) region. The dataset was developed to preserve authentic spoken Arabic while providing standardized country and dialect annotations. ASALA is distributed as two complementary resources, ASALA-Raw and ASALA-QA, which differ in their collection strategy, scale, and quality assurance process.