Effective Arabic Dialect Classification Using Diverse Phonotactic Models

This work investigates dialect-specific and cross-dialectal phonotactic models, and achieves 2% average equal error rate for pairwise classification.