IDENTIFY: A program for speech recognition, identification, and categorization experiments on PC compatible computers

Abstract

IDENTIFY is a program written in Turbo Pascal that accepts any combination of up to 100 speech waveform files as stimuli and up to 25 response categories. The waveform files can be from either ILS or C‐Speech, or any compatible format. The program presents the stimuli in block‐randomized order and waits for the subject to respond on the touch sensitive surface of a small digitizing tablet (Koala Pad). The tablet can be configured to reflect any number and pattern of desired responses. The most common case is when the number of stimuli and responses are the same, producing a standard confusion matrix. In this special case the program outputs an additional data file that conforms to the format for the Speech Information Transmission Analysis (SINFA) [Wang, Behav. Res. Methods Instrum. 8, 471–472 (1976)]. A categorization experiment can be run by specifying many stimuli that span a continuum, allowing only two responses. The program requires a PC or AT compatible computer, a Koala Pad connected to a game port, and a Data Translation DT2801‐A D/A output board. [Work supported by NIH.]