Both teaching and research in Speech Sciences differ radically from comparable work in the physical sciences because of the need to link different levels of representation - and analysis. Spectrographic and waveform presentations are useful first tools but the vital phonetic information which they encapsulate is not readily accessible. The present work is aimed at meeting three speech science teaching objectives:
to show how quite simple phonetically relevant features can be derived in real-time from combinations of acoustic and physiological data provide a basic PC workstation giving a platform for the analysis, display and measurement of single and combined speech pattern element sets which are directly related to: pitch; loudness; regularity; frication; and timbre discuss particular examples (taken from English, French, Chinese, Sekgalagadi and speech pathology in the demonstrations).
Special attention is given to establishing links between auditory, articulatory and physical levels of description and to provide for normalisation in display and quantification.