Some more buzz listening tests

I ran the impulse train test I described at the bottom of this thread, to
see if the differences between frequencies, or the VF is the more important
factor in creating periodicity buzz:

/tuning/topicId_95522.html#95556

Here are the results:

http://www.mikebattagliamusic.com/music/buzzvfvsdiff344.wav
http://www.mikebattagliamusic.com/music/buzzvfvsdiff172.wav
http://www.mikebattagliamusic.com/music/buzzvfvsdiff86.wav

The files play first an impulse train, which looks like this (view in fixed
width)

|___|___|___

which has all harmonics from 1 to infinity (or at least up to 22050 Hz). So
you can think of this as the ultimate JI chord, made of sines, all at equal
volume.

Then, they play then an impulse train in which every other impulse is
flipped upside down, like this:

|___ ___|___
|

This has only odd-numbered harmonics from 1 to infinity. Also, if you've
noticed, the period has now been cut in half. So the "difference tones"
remain the same between each frequency, but the period has been halved.

The file then goes back to the old impulse train waveform...

|___|___|___

And finally goes to a different version of the impulse train in which the
period remains the same, but an upside down impulse is inserted between each
impulse:

|_ _|_ _|_ _|_
| | |

The file then repeats itself.

The aim is to see between which two waveforms the periodicity buzz remains
constant: the waveforms that have the same period, or the waveforms that
have the same frequency differences? It seems to me that the buzz remains
constant between the first, second, and third waveforms (same as first), and
then the fourth one buzzes twice as fast. So it would appear to have more to
do with frequency differences than waveform period, which makes sense from
the gammatone filter plot from before.

-Mike

One last test: I also wanted to test the hypothesis that periodicity
buzz had something to do with the brain summing the signal and then
performing some sort of time-domain analysis. So for this test, I took
an impulse train and played it binaurally, but with the two ears
aligned such that the signal, if summed, should be transposed up an
octave. Here's a plot of what I'm talking about:

http://www.mikebattagliamusic.com/music/staggeredtrains.png

The top graph is the signal being played into the left ear, the middle
is the signal being played into the right ear, and the bottom is the
resultant signal that should come out if the two are summed. Note that
the bottom one is the same as the other ones, but transposed up an
octave.

I end up playing the left signal into the left ear, and the right
signal into the right ear, such that we have stereo decorrelation
between the two signals. If periodicity buzz is somehow rooted in a
temporal process taking place in the brain, and not in the cochlea,
then when the binaural signal is present, the buzz should sound twice
as fast as when either signal is played alone.

Here's the example:

http://www.mikebattagliamusic.com/music/staggeredtrains.wav

First you get the left ear alone (pay attention to the speed of the
buzz), then the right ear alone, then the binaural signal (buzz
doesn't sound like it speeds up to me), then the signal you WOULD have
gotten if a summation were involved (now the buzz speeds up).

In other words, for the binaural signal, the buzz between the two ears
is perfectly out of phase with one another, such that between each
particle of buzz in one ear lies a particle of buzz in the other ear.
The brain doesn't seem to be able to put this together to hear buzz
twice as fast, thus buzz must be taking place before any summation
occurs, if a summation occurs at all.

Renditions of both of these examples that involve chords like 5:6:7
and such will follow, for those skeptics out there who are still
holding out that the buzz in these examples isn't "true" periodicity
buzz, although they're going to be more difficult to work out.

-Mike

On Tue, Jan 18, 2011 at 3:43 AM, Mike Battaglia <battaglia01@...> wrote:
>
> I ran the impulse train test I described at the bottom of this thread, to see if the differences between frequencies, or the VF is the more important factor in creating periodicity buzz:

--- Mike Battaglia <battaglia01@...> wrote:

> Here are the results:
>
> http://www.mikebattagliamusic.com/music/buzzvfvsdiff344.wav
> http://www.mikebattagliamusic.com/music/buzzvfvsdiff172.wav
> http://www.mikebattagliamusic.com/music/buzzvfvsdiff86.wav

You haven't established that these files contain periodicity
buzz. It certainly doesn't sound like periodicity buzz.
Why not try testing your idea with signals closer to what
we usually call periodicity buzz?

-Carl

On Mon, Apr 25, 2011 at 3:33 PM, Carl Lumma <carl@...> wrote:
>
> > http://www.mikebattagliamusic.com/music/buzzvfvsdiff344.wav
> > http://www.mikebattagliamusic.com/music/buzzvfvsdiff172.wav
> > http://www.mikebattagliamusic.com/music/buzzvfvsdiff86.wav
>
> You haven't established that these files contain periodicity
> buzz. It certainly doesn't sound like periodicity buzz.
> Why not try testing your idea with signals closer to what
> we usually call periodicity buzz?

This is out of context now, but was originally in response to what you
wrote at the bottom of this:

/tuning/topicId_95522.html#95557

I wrote:
> > The theory that you'd hear the buzz frequency correspond to
> > the VF means that you'd hear a buzz-unit (a buzzon?) once per
> > period, and the theory that you'd hear the buzz frequency
> > correspond to the difference tone means that you'd hear
> > a buzzon once per impulse, whether upside down or not. So if
> > I generate an odd-harmonic buzz waveform, and the buzz sounds
> > twice as fast, then that means the difference tones are what's
> > being heard, not the VF.

Carl wrote:
> Ok, I'm down.

-Mike