Discrimination of FM glide direction and slope
Lawrence L. Feth, Mary E. Neill, Chien‐Yeh Hsu · The Journal of the Acoustical Society of America · 1990
Transitions of the second formant in speech sounds have been called the single most important carrier of linguistic information in the speech signal [Liberman et al., Psychol. Rev. 74, 431–461 (1967)]. Both the direction and the slope of the formant transition may be essential in cuing place of articulation, for example. Attempts to study listener performance in laboratory tasks that use frequency-modulated tones to “simulate” such formant transitions are often confounded by pitch differences between the signals to be discriminated. A procedure analogous to the “roving level” procedure used in profile analysis and intensity perception studies to counter the influence of this pitch artifact has been devised. In a two-cue, two-alternative forced-choice (2Q, 2AFC) paradigm, listeners are asked to determine which FM glide differed either in direction or in rate of transition (slope). For each listening interval, the starting frequency of the glide is randomly chosen from a uniform rectangular distribution centered on fc. Because each glide in the trial begins on a random frequency, the listener cannot use pitch differences to determine glide direction or slope. The current version of the program operates three simultaneous adaptive-tracking rules, with signals generated in real time using an Ariel DSP-16 board. Preliminary results for fc from 250 through 4000 Hz, and glide durations from 25 through 100 ms, will be presented. [Work supported by a grant from AFOSR.]