High-Speed Videolaryngoscopy in Pathological Phonetics
* Presenting author
Abstract:
High-speed videos of vocal fold vibration have been used in research on pathological phonetics for the past few decades. However, the relevance of such videos still continues to increase, basically because of two reasons. First, more and more data is collected. Our current database contains data of 200 human subjects with various voice pathologies, giving rise to the clinical relevance of the size of the database. Second, video analyses become increasingly sophisticated. A brief overview of current attempts to characterize varieties of pathological voice qualities via video analyses is given. Vocal fold vibration is quantified via graphical segmentation of the glottal gap. We focus on model-based analyses of vocal fold vibration. In particular, a combination of lumped and distributed elements enables modelling of phase differences in the anterior-to-posterior direction along the vocal folds. A model that outputs synthetic videokymograms of the vocal folds is capable of modelling left-to-right phase differences of vocal fold vibration. This model fits natural data without the need for graphical segmentation. Finally, glottal area waveforms involved in vocal fry and diplophonia are modelled. Glottal area waveforms appear to be relevant for relating voice production to perception.