mistuned harmonics and duplex perception

[HARTMANN%MSUPA.BITNET@xxxxxxxxxxxxx]

A recent note from Valter Ciocca says:

  "In support of Al's point that pitch can be affected by
  grouping, Chris Darwin and I found that the pitch of a complex
  could be changed by mistuning one of its low-numbered components
  only when the mistuned component was grouped with the
  remaining components of the complex."

This is a very nice result. Unfortunately, some other studies do not agree
with the conclusion. In 1984 I did mistuned harmonic experiments with
brief tones, using a 200 Hz complex tone with 7 harmonics, except that the
fourth harmonic was mistuned.

Experiment 1 found the amount of mistuning required to hear out the
mistuned harmonic as a separate entity. For a tone duration of 40 ms this
threshold mistuning was 22 Hz (2.75 % of the correct harmonic frequency of
800 Hz). For mistuning less than 22 Hz the complex was a single entity and
the mistuned harmonic did not pop out.

Experiment 2 measured the shift in the virtual pitch caused by mistuning of
the fourth harmonic. (The virtual pitch is in the vicinity of 200 Hz.)

Data, Experiment 2:

      frequency shift       shift of the         s.d. of the virtual
      of mistuned 4th       virtual pitch             pitch match
           (Hz)                 (cents)                (cents)
           0                       3                      8
           2                       4                      9
           6                       9                      7
          10                      11                      6
          20                       8                      7
          30                      18                      8
          40                      17                      7

Comparison of the two experiments shows that mistuning the fourth harmonic
by 30 or 40 Hz led to virtual pitch shifts larger than the s.d. and also
caused the mistuned fourth harmonic to pop out as a separate entity. This
mistuned harmonic is perceived in a way that is reminiscent of duplex
perception in speech, whereby a single acoustical element is perceived in two
different ways.

A similar conclusion was reached by Moore, Glasberg and Peters [JASA vol 77,
1853 (1985)] who found that with still larger amounts of mistuning the
mistuned harmonic made decreasing contributions to the virtual pitch shift
but not abruptly so.

>From the point of view of auditory organization one has to say that
although pitch may be the great organizer, not every experiment will
show that unequivocally. Maybe though, Ciocca and Darwin have done a
better experiment than we have. Need more information.

Bill Hartmann