Audibility of harmonics in 'periodic white noise'

In a previous article (Duifhuis, 1970) results' concerning the audibility of harmonics in a periodic pulse have been presented. Each of the lower harmonics could be perceived separately, whereas the high harmonics were heard together as one complex signal. High harmonics, however, appeared to be perceptible as a part-tone (see P.12) they were spectrally suppressed. This phenomenon was attributed to time analysis, the resolving power of which increases with decreasing frequency resolution, and conversely. The present experiment is meant to give additional evidence in support of the previous findings. Therefore, we carried out a similar experiment using stimuli having the same frequency amplitude spectrum as the periodic pulse, but with a different time pattern. For these stimuli we chose a special type of pseudo-random binary noise, built up out of maximum length sequences (see, e.g., Golomb, 1967; Bilsen, 1967). The period T of the noise was variable. For the sequence length we chose 27-1 127. This value determines the envelope of the frequency amplitude spectrum of the periodic noise. With the chosen value this spectrum is flat up to the 50th harmonic. The difference with a periodic pulse is found in the phase spectrum (and in the time pattern). The phases of the harmonics of periodic noise are pseudo-randomly distributed whereas the harmonics of a periodic pulse have equal phases. To the extent that the ear performs an analysis of the frequency power spectrum, there should be no difference between the audibility of harmonics in either periodic noise or in periodic pulses. In case of time analysis, however, considerable discrepancies are likely to be found.