A Psychoacoustic Approach to Dysphonic Voice Quality Perception

This project targets the millions who suffer from voice disorders. It aims to develop new methods for quantifying voice quality disorders, establish relationships among different voice quality dimensions, develop fundamental scales of voice quality, and develop computational models to predict voice quality perception from a recording.

Sponsor
National Institutes of Health
$1,400,000
Principal investigator
Rahul Shrivastav
Active since
July 2007

Abstract

The proposed project seeks to understand the perception of dysphonic voice quality (VQ) and to develop an innovative approach to quantify VQ through models based on well-established experimental and mathematical approaches, such as those used in psychoacoustics, engineering, and computer sciences. The proposed science will advance our knowledge of the cognitive processes underlying VQ perception as well as allow linking VQ perception to underlying acoustics and vocal fold physiology. The practical outcomes include the development of metrics that can easily be translated to practice and have a direct impact on clinical care and in other non-clinical applications. The development of a universally accepted standard for voice quality measurement is a grand goal, and the proposed project will bring us significantly closer to this possibility. Supported by two prior NIH grants, this project has made significant advances towards addressing this problem. Our research team is highly multidisciplinary combining knowledge from hearing science, psychoacoustics, speech science, speech-language pathology, and computer/electrical engineering.