Functional Validation of a Generative AI-Based Mobile App for Assessing Speech Difficulties in Children

Abstract

Speech and voice development in childhood are essential for academic progress and social participation; difficulties in oral expression may impact learning and emotional health. Generative artificial intelligence (GAI) technologies integrated into interactive mobile applications offer new possibilities for the automated assessment of Spanish-speaking children’s speech. This study functionally validates a mobile application that uses GAI to automatically assess children’s speech fluency, pronunciation, and intonation, providing automated scoring, targeted feedback, and personalized recommendations. This Phase 1 functional validation, based on synthetic data, lays the groundwork for a four-phase framework aimed at guiding cross-national and multilingual research in artificial intelligence (AI)-supported speech evaluation. The methodology focused on a correlational analysis between the scores generated by the application and the acoustic indicators—number of pauses, pitch range, and spectral clarity—obtained from 160 samples. Descriptive and spectrographic analyses revealed mean decibel levels ranging from 15 to 33 dB, pitch ranges around 3.843 Hz, and spectral clarity between 0.033 and 0.036. It is concluded that this tool could contribute to the automated and multidimensional assessment and feedback of speech difficulties in Spanishspeaking children.