[jOrgan-user] Timber Discrepancy

[John Beach]

Last month (June 14), I wrote the following under a Subject header of “Trendline Synthesis”:

(What is described below refers to additive synthesis in an additive synthesizer, NOT Trendline Synthesis.)

“If, for example, one synthesizes additively by varying the amplitudes of component sine frequencies, beginning with 8'-C,
4'-C, 2-2/3'-G, 2'-C, 1-3/5'-E, 1-1'3'-G and 1'-C, the computer has the ability to transpose the resulting (wave file) timber for each of the frequencies of
pitches of the keyboard. The notion that the timber of 4'-C of a Principal stop is physically-modelled using the recipe of harmonic amplitudes for D#,
it seems to me, is one of "same difference." The proposition becomes that, "frequency varies timber." I think this is deceptive, since it is provable
that timber remains the same because of its constituent properties, just as frequency is fixed.


I have since proved that I am wrong (last sentence above). Using the exact, same “recipe” for a Clarinet (00 6141 432), at 16’ and 8,’ theoretically, there should be
no perceptible change in timber when moving from the last note Bb of 16’ pitch (Midi note #47), synthesized using 16’ as the fundamental, and
C (Midi note 48) synthesized using 8’ as the fundamental. The sine waves (A=880) of the synthesizer are the same for all footages up to two feet, and above 2 feet,
through 1/4’, they are all a sine wave of A=1760. The difference is, literally, one octave higher for all the synthesizer footages, respectively. Yet, the timber, synthesized
at 16’ pitch is so noticeably different from that synthesized at 8’ pitch that the two can not be used in the same stop. Given this reality, it is difficult to conceptualize
how “digital scaling” could be designed to achieve the same effect that the scaling dimensions of pipes achieve. The difference is like that of the Basson-Hautbois
which makes one wonder how it could be useful.

Audsley, repeatedly, insists that a stop must be carried throughout its compass with the pipes of the same type. Mixing is an adulteration. Synthesis proves this to
be true! The recipe for a Contra Clarinet 16’ (NOT what is described above) is different from an 8’ Clarinet (which is above), and the timber is, obviously, different.

John Beach

[Graham Wykes]

Hi John

Post by John Beach
“recipe” for a Clarinet (00 6141 432)

I’m assuming from the format that this “recipe” is a Hammond registration. If so, It does not represent true additive synthesis as the drawbars do not produce accurate harmonics. (e.g. the 3rd harmonic is around 1Hz flat) This means you can get very noticeable discrepancies. It’s part of the Hammond sound.

Here is a link to a recording of a synthesised (i.e. not sampled) organ. It uses additive synthesis with each stop defined in a binary file of its own. It means that it is possible to “roll your own” synthesis table to modify stops. You can also design your own organ definition if you want. Multiple definitions are also possible.

https://drive.google.com/file/d/0B44nK7q6Xns7cG9uRlh5Z0x5NzA/view?usp=sharing <https://drive.google.com/file/d/0B44nK7q6Xns7cG9uRlh5Z0x5NzA/view?usp=sharing>

By the way, the reverb is inbuilt and is the same as Zita-rev1. If you want to know more, the page for Aeolus id at http://kokkinizita.linuxaudio.org/linuxaudio/aeolus/ <http://kokkinizita.linuxaudio.org/linuxaudio/aeolus/>. AFAIK there is no windows version.

Cheers
Graham

[John Beach]

Graham, thanks for your analysis and comment and the link to the Aeolus organ file. It is an excellent sound and one of the reasons that I want to get a computer up and running Linux.

While the “recipe” I use is a Hammond organ registration, the additive synthesizer that I use is a jOrgan disposition which I made which uses a soundfont of organ stop footages
with properly tuned (nazards/fifths +2 cents, tierces –14 cents), and includes the ability to synthesize at all octave footages from 32’ to 1’ and contains upper partials of 4/5’, 2/3’,
1/2’, 2/5’, 1/3’ and 1/4’ which can be synthesized and sampled at the two lowest octaves (Midi notes 36 and 48.). It produces good results and the fact that the timber of the Clarinet, synthesized at 16' differs from the timber of the Clarinet sampled at 8’---using the same registration---is not an anomaly or error of the tuning of the synthesizer.
I had the idea that the synthesis would merely be an uninterrupted continuation of the exact, same timber since the the registration was the same for both, my reasoning, being based
on the belief that a pure sine wave is a pure sine wave and 32 Hz is simply one octave lower than 64 Hz, so the footages or harmonic partials in the synthesis should produce a tone
---of the same timber---simply one octave lower than that synthesized using 64 Hz as the fundamental. This is not the case and I thought it was a flaw or error in my soundfont
construction. I examined it again to make sure that any factor which might create a discrepancy in the 16’ synthesis was corrected, but there were no errors.

From what you said, I think you inferred that I was sampling an actual Hammond organ. That is not the case.

John B.