Subject: Re: High frequency sensitivity From: Dick Duda <DUDA(at)AI.SRI.COM> Date: Wed, 29 Mar 1995 11:03:34 -0500> Does anyone know any studies that point to frequencies > > 20kHz affecting perception (maybe of transients rather > than continuous tones). > Does anyone have an opinion? The debates over bandwidth needed for high-fidelity sound reproduction are long standing and unlikely to be resolved to everyone's satisfaction in the near or even the distant future. However, I will note that I have found the commercial 44.1-kHz sampling rate to be marginal in studying binaural hearing and spatial localization. A number of studies have shown that normal listeners can localize sources in azimuth to about 2 degrees, and can reliably detect changes in position (the minimum audible angle) to about 1 degree (for example, see A. W. Mills, "Auditory Localization", in J. V. Tobias, Foundations of Modern Auditory Theory, Vol. II, pp. 303-348 (Academic Press, NY, 1972)). To the extent that this performance is dependent on detecting interaural time differences, it implies the ability of the auditory system to detect shanges in arrival time as small as 9 usec. At 44.1 kHz, the sampling interval is about 22 usec, making it hard to create digital audio test signals with smaller interaural time shifts to test the limits. Of course, the ability to detect a 10-usec difference in arrival time does not imply that the auditory system has a 50-kHz Nyquist frequency, and I do not believe that this would have any appreciable effect on home entertainment systems, even for listeners with "golden-ears" (which excludes me). However, I often wish that the sampling rate were closer to 100 kHz for my own work. Dick Duda -------