Discrimination of sounds in hearing
Harry N. Levitt · Spiral (Imperial College London) · 1964
This thesis describes an investigation of the relative ability of a listener to separate out one of several simultaneous sounds.Experiments were performed in which a listener was required to state the apparent lateral position of a sound in the presence or absence of a second, interfering sound.If no change in the listener's performance was observed whether one or two sounds were presented, then the desired sound was said to be totally separable from the interfering sound.If a decrement in performance was observed then the relative loss in the measured ability of the listener to lateralize the desired sound was used as a measure of the degree of separability.Headphone listening with interaural time delay as the physical correlate of lateral position was used, Preliminary experiments showed the technique to be one of great potential accuracy, but that certain problems associated with a drift in the interaural timedelay threshold needed to be resolved, The experiments were mainly with separate, but statistically identical, random noise sources.The results were interpreted in terms of a simple crosscorrelation model.Sequential estimation techniques were considered in order to compensate for gradual drifts in the listener's performance during a test.A recently devel-2" oped strategy was found to be especially suited to this task and it was used in an investigation on the relative separability of one of two physically similar, voiced messages; this being its first known practical application.The investigation showed that the degree of separability varied in a manner similar to that for the relative articulation for the corresponding listening task.The use of a sequential strategy in placing a sound image in a desired position was also demontrated.3.