Arabic Speech Processing

We develop tools and datasets for Arabic speech, including automatic speech recognition, speech synthesis, and diacritic restoration. We work on different variants of Arabic: Classical Arabic, Modern Standard Arabic (MSA), dialects, and code-switching.

Screening for Developmental Language Disorder

In collaboration with speech-language pathologists, we investigate the feasibility of automatic early screening of developmental language disorder (DLD) using video games and speech processing tools.

Stuttering Assessment

In collaboration with speech-language pathologists, we develop clinical annotation schemes and models for automatic stuttering severity assessment, and collect/annotate audiovisual data in multiple languages.

Spoken Emirati Archive

We ethically collect diverse Emirati speech data from various regions and age groups by engaging directly with the public. We develop tools, dictionaries, and AI models for effective processing of Emirati dialects.