Tao Zhang
Lawrence L. Feth
Dept. of Speech and Hear. Sci., Ohio State Univ., Columbus, OH 43210
Ashok K. Krishnamurthy
Ohio State Univ., Columbus, OH 43210
Previous work in this lab has developed a short-term IWAIF model for auditory frequency discrimination of dynamic signals. In this presentation we will report further tests of the model. The STEP versus GLIDE discrimination task used previously [Feth et al., J. Acoust. Soc. Am. Suppl. 1 86, S222 (1989); Madden, J. Acoust. Soc. Am. 95, 454--462 (1994)] has been replaced by an FM detection task. Sinusoidal FM of the GLIDE tone can be substituted for the STEP signal [Feth et al., J. Acoust. Soc. Am. Suppl. 1 88, S48 (1990)]. Listeners are asked to determine which of two tones is sinusoidally modulated in frequency. In a limiting case, the GLIDE has a slope of zero, that is, it is a steady tone. The detection task then is similar to FM detection. To reduce the influence of absolute frequency discrimination on the listeners' performance, the experiment was run using roving frequency. For every interval of each 2Q, 2AFC trial, the starting frequency of the signal was selected from a uniform random distribution. Preliminary results indicate that performance in the task is similar to FM detection when the ``unmodulated'' signal is a GLIDE. However, the modulation index of the just detectable FM is approximately four times larger than the index for a steady reference tone. A comparison of IWAIF model predictions with listener performance will also be given. [Work supported by a grant from AFOSR.]