Abstract
This article presents the comprehensive design and evaluation of a digital storytelling system tailored forchildren aged 4 to 6 in Latin America. This system is based on generative artificial intelligence.Tests were conducted encompassing the system’s functionality, content diversity, generation times,voice evaluation, intonation, speed, and pronunciation quality. The results confirm the system’s correctoperation and intuitive nature. The generated stories exhibit a high degree of diversity, as evidenced bythe calculation of Jaccard indices, with a maximum value of 0.2 observed in the evaluations of 30 analyzedstories. As anticipated, generation times increase with the length of the stories. It was determined thatthe voice best suited for storytelling is Onyx from OpenAI’s TTS. However, pronunciation errors werenoted in all TTS model voices. According to the conducted analysis, the system produces diverse stories that promote values among Spanish-speaking children, thus emphasizing the importance of inclusivity for individuals with different abilities. It is noteworthy that no inappropriate content for children was found in any of the stories
Publisher
Salesian Polytechnic University of Ecuador