Regresar

A new technique for assessing glottal dynamics in speech and singing by means of optical-flow computation

Abstract:

The use of high-speed videoendoscopy (HSV) in combination with image-processing techniques is the most promising approach to investigate vocal-folds vibration and laryngeal dynamics in speech and singing. The current challenge is to provide facilitative and informative playbacks for clinical and research purposes. We present three new facilitative playbacks using an optical-flow framework (OF). Optical-flow techniques are widely used in the field of computer vision for tracking unidentified moving objects in video sequences. The application of OF computation to HSV images is investigated. The advantages, drawbacks, and the complementarity to existing methods are discussed. The method has been tested on a database of 60 HSV sequences which covers different voice qualities for spoken and sung vowels. The new data representations have been compared with commonly-used facilitative playbacks. They provide additional information on the temporal dynamics of glottal vibratory movements during glottal closing and opening phases.