In 1693, astronomer Christiaan Huygens, standing at the foot of a staircase at the castle at Chantilly de la Gour in France, noticed that sound from a nearby fountain produced a certain pitch. He correctly observed that the pitch was that of an open organ pipe whose length corresponded to the depth of one stair tread and concluded that the perceived pitch was caused by periodic reflections of the sound against the steps of the staircase. What Huygens discovered is that, when one or more delayed repetitions combine with the original sound, the resulting sound has a pitch which corresponds to the inverse of the time delay. This phenomenon is often called repetition pitch. The original sound can be noise, music, speech, single pulses, combinations of pulses, or just about any sound. Because the frequency response characteristic of such a time delay system has a comb like structure, the process is often referred to as comb filtering. The pitch effect is audible for time delays ranging from 1 to 10 ms.
Now you hear a four octave diatonic scale played with pulse pairs that are samples of a poisson process.
Now you hear a four octave diatonic scale played with bursts of echoed or comb filtered white noise
One way to demonstrate repetition pitch in a convincing way is to play scales or
melodies using pairs of pulses with appropriate time delays between members of a pair.
In the first demonstration, a 5-octave diatonic scale is presented using pairs of identical pulses. In each pair the second pulse repeats the first, with a certain time delay t, which changes from 15 ms to 0.48 ms as we proceed up the scale.
Next a 4-octave diatonic scale is presented using trains of pulse pairs that occur at random times. Time intervals between leading pulses have a Poisson distribution. Delay times t between pulses of each pair vary from 15 to 0.95 ms. Although the pulses now appear at random times, a pitch corresponding to 1/t is heard, as in the first demonstration.
The final demonstration is a 4-octave diatonic scale played with bursts of white noise that are echoed 15 ms to 0.95 ms later.