[sdiy] Moogey jitter

[Ian Fritz]

At 08:44 PM 4/15/06, Kenneth Elhardt wrote:

>The link below is a magical moogy sawtooth wave spliced into a cold,
>perfect, digitally stable, un-antialiased sawtooth generated in CoodEdit.
>It switches back and forth several times.  Remember, this is the most
>extreme difference in the world theoretically.  But there is virtually no
>audible difference.  Note, I have already gone through the work of measuring
>jitter in a moog waveform and posted the results on AH along time ago.  I
>did it down to about 1/32 of a sample of resolution.  Since the modern
>Technosaurus oscillators had three times the amount of jitter than the Moog
>did, that must mean the Technosaurus sound more Moogy than the Moog.  I
>don't suppose Keven mentioned that.  Measuring jitter means nothing unless
>one compares it to other analogs.
>
>http://home.att.net/~elhardt5/Sawtooths.wav

Kenneth --

I have had careful look at your example.  Thanks for providing it.

When I first listened to the file I could hear a small difference between 
the two instruments.  However, there is about a .5 dB amplitude difference 
between the two.  I leveled this difference with an editor and now I can't 
hear any difference between the two signals.

The amplitude variations seen in the Moog waveform are due to small 
transients at the switching point of the waveform.  These come and go at a 
rate of ~5 Hz.  I can't hear any modulation at that frequency, and I 
believe the spikes are inaudible artifacts. But people will always say your 
data has am noise on it anyway, I'm sure.

A spectral analysis of the two waveforms shows that the true saw rolls off 
faster above 10 kHz than the Moog does.  Maybe sharp young ears can hear this.

I used Sound Forge to resample and interpolate the waves at 4*44.1 
kHz.  With this done there are transients at all the switching points and 
they are all nearly the same.  So there is very little amplitude 
variation.  I measured the Moog period at five different points with a 
resolution of .01 ms.  The periods range from 10.28 ms to 10.30 ms, a 
variation of 0.2%.  If the jitter is a fixed amount of time, then it would 
be 2% at 1 kHz, which might be audible.

I'm afraid that this is looking like a golden-ear-audiophile kind of 
argument.  The only thing I would want to look closer at would be data at 
higher frequency, say 2 kHz.

   Ian