TOWARDS INTELLIGENT THERAPY FOR SPEECH SOUND DISORDERS: A REVIEW OF ARTICULA-TION THERAPY APPS AND GPT-BASED ADVANCEMENTS

Abstract

Speech Sound Disorder (SSD) is characterized by difficulties in articulation, leading to mispronunciations that impact clear communication. Early detection and customized therapy are essential for effective treatment. This paper explores multiple methodologies for SSD therapy support, with an emphasis on integrating Generative Pre-trained Transformers (GPT)-based models for phoneme-specific sentence generation. Using a GPT model, sentences containing user-specified phonemes were generated, ensuring grammatical accuracy to support targeted phonetic practice. Experiment results indicate that leveraging AI-based models, such as GPT for phoneme-focused generation of sentences, can significantly enhance therapy outcomes by providing tailored interventions, also reducing the Speech Language Pathologist (SLP’s) workload. Our findings underscore the potential of the technology to support more accessible and effective SSD therapy, particularly with future expansions to include region-specific language support, to aid diverse linguistic populations.