The present design demonstrates effective communication in educational environments which face challenges related to the lack of personalised emotional interaction between teachers and students. This project aimed to design, develop and evaluate an educational robotic head equipped with emotional recognition and expressive voice synthesis to improve such communication. An applied experimental methodology was used, integrating hardware (Arduino Uno, 1080p camera, microphones, servo motors) and software (Pictoblox programming and multimodal deep learning models). The system was tested in real classrooms, showing an average accuracy of 80% in emotional recognition and a positive evaluation of the naturalness and empathy of the synthesised voice. The results indicated significant improvements in motivation, participation and learning environment. It is concluded that the integration of emotional intelligence in educational robots has great potential to transform communication and promote inclusive and affective learning.