Application of signal processing utilities to speech physiology data
William L. Henke, Joseph S. Perkell · The Journal of the Acoustical Society of America · 1982
A coherent family of command level signal processing utilities is being used to process time-continuous signals from experiments on the physiology of speech production. The utilities' wide range of functionality includes the following: simultaneous sampling of up to 16 analog channels at different rates, demultiplexing and separation of channels into groups of signals at different bandwidths with time registration across bandwidth groups, highly flexible signal processing to aid in data analysis, and interactive audition, viewing, marking and labeling to record timing of events and magnitudes of signals for subsequent analysis. Signal stream processing algorithms are specified by the user as graphical block diagrams which can support all the customary forms of signal processing. Useful examples include running average of EMG signals, differentiation of movement signals to obtain velocity and aceleration, integration to provide a measure of muscular “effort,” and filtering. Fourier analysis types of processing include computation of spectra (speech and movement signals) and cross-correlation functions to examine phase relationships between EMG signals and moment parameters. More specialized processing includes ensemble averaging and algorithmic event labeling. Examples will be presented. [Work supported inpart by N.I.H. Grant No. NS04332.]