J. N. Ananthraraman
Ashok K. Krishnamurthy
Dept. of Elec. Eng., Ohio State Univ., Columbus, OH 43210
Lawrence L. Feth
Ohio State Univ., Columbus, OH 43210
The EWAIF model was based on the listener's ability to discriminate between narrow bandwidth, two-component complex tones. These tones shared a common envelope as described by Voelcker [Proc. IEEE 54 (1966)]. Later, the EWAIF model was applied to multi-component signals such as those used in profile analysis. Those signals were neither narrow bandwidth nor common envelope signals; however, the EWAIF model predicted a constant difference in pitch for just-detectable increments to the center component of otherwise flat-spectrum signals. Recently, Versfeld and Houtsma [Q. J. Exptl. Psychol. 43 (1991)] have applied to EWAIF model to a number of periodic signals, some of which are broadband. They report the failure of the model to predict the discriminability of these signals. Versfeld and Houtsma did not ensure that their test signals were common envelope pairs. Thus, the addition of envelope differences to their signals may have given their listeners additional cues upon which they could base discrimination. Signal pairs have been defined that are similar to those used by Versfeld and Houtsma, but that have common envelope properties. Recently the EWAIF model have been reformulated into what is better called an intensity-weighted average of instantaneous frequency (IWAIF) model. In this presentation, the relationship between EWAIF and IWAIF model predictions for discrimination of relatively simple, periodic signals are demonstrated and then the IWAIF model is applied to the discrimination of broadband complex signals. [Work supported by AFOSR.]