Building a Contextualized Arabic Voice Command Corpus for Industrial Automation Systems
Naouar Laaidi, Abderrahim Ezzine, Meryam Telmem, Mohamed Lamrini, Hassan Satori · 2025
Voice command systems are increasingly critical for enhancing human-machine interaction, particularly in industrial environments shaped by Industry 4.0. However, deploying such systems in Arabic-Speaking contexts presents unique challenges due to the morphological complexity of the Arabic language, dialectal variations, and the lack of standardized industrial command corpora. This study addresses these gaps by introducing a specialized Arabic voice command corpus tailored for industrial applications. Data was collected from major stakeholders in sectors such as energy production, railway operations, and chemical processing, culminating in a corpus of 35 distinct commands. A comprehensive syllabic analysis revealed dominant patterns in CV and CVC structures, essential for optimizing speech recognition systems. Cross-Sector validation was conducted with industry professionals from prominent Moroccan organizations, including ONE, ONCF, and OCP, ensuring alignment with real-world operational requirements. The resulting corpus covers critical industrial functions such as process control, safety protocols, maintenance, and inventory management. By integrating linguistic insights with practical applications, this research lays the groundwork for developing robust, context-aware Arabic automatic speech recognition (ASR) systems. The study contributes to the advancement of Arabic natural language processing and enhances accessibility to Industry 4.0 technologies in Arabic-Speaking regions.