Speech and Language Technology for Disabled Persons

Stockholm, Sweden
May 31 - June 2, 1993

CASPER: Computer Assisted Speech Evaluation Expert System

James A. Till

Speech Research Laboratory (126), Veterans Affairs Medical Center, Long Beach, California, USA

Existing computer technologies and laboratory knowledge are not used routinely for diagnostic speech evaluation. There are no universally accepted standard tasks or measures for the evaluation of speech. Speech evaluation methods can vary markedly among clinics and rely heavily on subjective impressions. Existing general purpose signal analysis systems are cumbersome for routine clinical use. We have developed a hardware\software system which addresses some of these problems. The system, CASPER (Computer Assisted Speech Evaluation and Rehabilitation), was developed with funding from the Department of Veterans Affairs, USA. It uses a microphone and other specialized instruments to sample the speech signal and test related physiological functions. Four of CASPER's 11 protocols provide measures related to voice, prosody and nasal resonance. Depending on the nature of the observed speech deviance, additional protocols are recommended to assess the physiological integrity of the pulmonary, laryngeal, velopharyngeal, and upper airway articulatory system. A user specified protocol allows custom tasks to be created and analyzed using 1-4 channels of input. This paper describes the system and illustrates selected analysis features.

Full Paper

Bibliographic reference.  Till, James A. (1993): "CASPER: computer assisted speech evaluation expert system", In SLTDP-93, 141-144.