Automatic Posture Evaluation for Professional Voice Users

Abstract

Voice trainings are executed mostly by therapists to their knowledge, experiences and skills. In these therapies, tension spots of a body are evaluated. A body is relieved with guidance of a therapist and physical exercises. However, voice quality evaluation of professional voice users are implemented subjectively. Classical voice evaluations can not be done by an objective approach, yet done with therapists’ intuition. In this study, a measurement system was proposed to evaluate voice quality objectively by using the posture of a patient. Different machine learning algorithms were used to classify objective voice quality and posture quality, yet Artificial Neural Network models were found as best classifiers. Two models were tested using individual test sets and accuracies of voice quality and posture quality estimations were found to be 83.35% and 78.27%.