Voice disorders are a serious threat to a teacher's career. Nearly 60% of all teachers in the U.S. will suffer from a voice disorder during their lifetime and between 20-43% of student teachers report voice problems. A pivotal point is that teachers with a history of voice problems during student teaching have an almost nine-fold greater chance of developing a voice disorder. Consequently, early detection is essential to address this public health issue. However, while vocal effort and fatigue emerge during teaching, signs of these symptoms are often elusive during assessments. The larger issue at hand is that systematic research on laryngeal muscular function lining the trajectory from normal vocal function to chronic muscle tension dysphonia continues to be sparse. Our project will take advantage of existing ambulatory monitoring devices for surface EMG (sEMG) to examine extralaryngeal muscle activity. A key challenge is the development of adequate analysis algorithms and to determine clinical utility. Our interdisciplinary team has developed an innovative pattern recognition algorithm named Hierarchical Guided Underdetermined Source Signal Separation (HiGUSSS) that has shown high accuracies in preliminary tests classifying isolated speech and non-speech gestures and distinguishing between normal and simulated pressed phonations within and across speakers. Our proposed Specific Aims are to (1) establish a reference dataset of extralaryngeal activity to determine the validity and reliability f a new sEMG pattern recognition system to detect vocal fatigue and to (2) examine the accuracy of a novel sEMG pattern recognition system to detect signs of vocal fatigue in student teachers. In the first study, we will collect sEMG data from 40 vocally healthy females ages 21-29 years and 20 matched newly practicing teachers with vocal fatigue selected using the Vocal Fatigue Index. During data collection in a laboratory setting, participants will produce repetitions of vowels and repetitions of a spoken sentence that elicits hard glottal attacks. The data will serve as training data for our HiGUSSS framework that extracts features of interest. The algorithm will be refined to a benchmark of at least 90% accuracy for differentiating between extralaryngeal function with and without vocal fatigue. In the second study, 22 vocally healthy female student teachers will be tested before and during the peak period of student teaching with identical tasks. These data will be analyzed to determine the sensitivity and specificity of the algorithm to identify individuals with vocal fatigue or emerging vocal fatigue. The central hypothesis is that the automated detection of extralaryngeal clinical and preclinical signs of vocal fatigue is possible. The project will provide insights into the muscular signatures underlying vocal fatigue and establish the foundation for a novel diagnostic tool for the early identification of potential voice disorders. In an exploratory manner, we will also test the real-time processing of student teachers' extralaryngeal activity in the classroom. The outcomes will inform prevention strategies and be relevant for a variety of individuals where vocal fatigue is a marker of vocal decline, such as the aging voice.
Voice disorders can end a teacher's career, which makes the early detection of voice problems in student teachers a priority to address this public health issue. We propose a novel pattern recognition system for signals from the neck muscles of the voice box (surface electromyography) that will help to identify vocal fatigue stemming from muscular over activity. The findings will help to improve the screening, early detection, and prevention of voice problems in student teachers transitioning to the teaching job.
