ASA 128th Meeting - Austin, Texas - 1994 Nov 28 .. Dec 02

4aMUb4. Effects of processing audio signals through equivalent color matching filters.

Andrew Blackford

Elec. Eng., Univ. of Oklahoma, Norman, OK 73019

B. Espinoza-Varas

Univ. of Oklahoma Health Sci. Ctr., Oklahoma City, OK 73190

In applications that combine musical sounds with visual colors, it is relevant to define rules that may relate color dimensions to the pitch and timbre of sounds. In a first attempt toward defining such rules, this investigation examined effects of processing audio signals through equivalent color matching filters. Using Butterworth bandpass filters, a three filter bank (f1, f2, f3) was designed to approximate the response of filters used in the CIE 1931 Standard Observer [G. Wyszecki and W. S. Stile, Color Science (1982)]. The bandwidths of the color-matching filters were transposed to the audible frequency range (0.02--20.0 kHz); the filter passbands were 0.05--0.6 kHz and 9.0--11.0 kHz for f1, 1.0--4.0 kHz for f2, and 9.0--11.0 kHz for f3; the passband gains were 1.15, 0.35, 1.0, and 1.75, respectively. Synthesized audio signal were passed through the filter bank and the power, P, at the filter outputs was computed. The P values were used to specify the X and Y coordinates of CIE chromaticity diagrams as follows: X=Pf1/(Pf1+Pf2+Pf3); Y=Pf2/(Pf1+Pf2+Pf3). Chromaticity diagrams were obtained for: (a) 0.07--8.0 kHz sinusoids; (b) vowels with different fundamentals; and (c) complex sounds consisting of 2-21 equal-energy sinusoids. [Work supported by OCAST.]