A System to Support Children in Speech Therapies at Home

Abstract

Voice disorders occur when voice quality, pitch, and volume differ or are inadequate for an individual's age, gender, cultural background, or geographic location. These are due to inherent internal and/or external factors, such as vocal cord damage, brain damage, muscle weakness, or vocal cord paralysis that often damage the vocal folds. Commonly the age range of the patients is 4-6 years old. To overcome these problems, speech therapy is needed, which consists of a set of exercises aiming to stimulate the child's language. A personalized treatment for each patient should be defined in accordance with the patient's specific problems. Since speech exercises, even if they usually are proposed as games, are often boring for the children and their caregivers, this research proposed the system Pronuntia, which supports all the actors involved in the speech therapy. The automatic acquisition and correction of the speech exercises through the system allows real-time feedback to patients and therapists. Moreover, some AI techniques have been implemented to help therapists in tuning the automatic recognition level of the patient's speech according to the disease severity. A user test involving 5 speech therapists and 5 caregivers has been carried out to evaluate the usability of the system.