AN INNOVATIVE APPROACH TO THE AUTOMATIC CONSTRUCTION OF ISLAMIC GEOMETRIC PATTERNS USING COMPUTER GRAPHICS AND GENERATIVE MODELS

This paper presents a multimodal intelligent system that integrates Speech-to-Text (STT) technology into the automatic generation of Islamic geometric patterns. The proposed framework transforms user voice commands into structured parametric representations and synthesizes visual patterns through generative algorithms. The approach is based on the integration of computer graphics and artificial intelligence models, enabling a more intuitive and efficient human–computer interaction. The software implementation is developed in Python using open-source STT and graphics libraries. Experimental evaluation reports an average speech recognition accuracy of WER = 7.8%, while pattern generation latency ranges between 120–250 ms under GPU acceleration. The results confirm that the system maintains high visual consistency and design diversity while achieving real-time performance.