On Tue, 2012-08-21 at 11:02 +1200, Jeff McClintock wrote:
>> > I think the most reasonable standard for an absolute 1/oct
>> > frequency unit is 0.0 = 440Hz?
>>
>> My modular plugins use a reference of 440Hz. Also parameters are ranged
>> between 0.0 - 10.0 but can exceed that if need be. (in a modular synth,
>> everything needs to interoperate).
>
> So for frequency 5.0 is 440Hz (Middle-A). i.e. the middle of the range -
> 5.0, is the standard 'middle' key.
> While I initially thought negative values was really weird, on second
> thought having it centred about 0.0 is nicer, especially considering the
> relative uses
> Great idea though. Octaves are far more universal than western semitones,
> yet trivial to convert between.
Aren't there a couple of misconceptions here or at least some potentialfor
incompatibility?
The signal itself is a digital representation of a voltage modifier. Thevoltage
has no semantic other than its value, it does not represent freqor gain or
cutoff or anything - the semantics of what happens only occursat the 'sink/dst'
of the signal.
Consider the following points:
a. 1/Oct is not a reference value, it is a definition for the effects of
amodifier signal. The value of 0.0 should definitely not refer to A440.When a
mod signal is applied to an osc then it modifies the settingof the osc. You set
the osc to A440 and apply 0v then it has no affect.If you apply 1v it should
output A880, that is 1/octave. If you want thesignal to be able to reduce the
frequency then you transpose the osc downby a couple of octaves and then set
the base mod signal at 2+ve to getback to a natural tuning, you can now reduce
the frequence by 1/Octave too.
Now let's have a look at those negative values with another example:
b. It should not support negative values. Let's forget about the 1/Octfor a
minute. This defines how a mod signal is applied: if it is appliedto an osc or
filter coff it represents 1/Octave. It can also be applied toan amplifier
though. If you accept negative values you will get phaseinversion and a
relative signal gain, that is probably not the intentionof the modifier though:
if you do not restrict the signal to +ve valuesthen the source of the mod
signal may need to understand more aboutthe semantics of the sink - in the case
that you have 0.0 to represent A440 then does it also have to represent 0dB (or
-96dB) when itis applied to an amplifier for example. This is excessive since
in anarbitrary system where rewiring is possible then sources can be
audiosignals or can be mod signals so now they need to understand where they
are patched to be able to deliver a correct reference. That is atbest difficult
and at worst unmanageable.
The above is partially opinion but is based on analogue signal pathsfrom the
old mono/mega synths. Agreed it might be time to move on buteither way, a
modifier signal of 1/Oct is based on the 1v/Oct and inthat model the modifier
signal has no reference regarding frequency,gain or anything else, that
reference is a function of what it is beingused to modified.
Just to be complete, I have no objection to such signals having someimplied
semantics. There will be some apps that do not have suchrestrictions and this
will result in inconsistencies - this is a good thing though, probably,
modifier signals and arbitrary routing wasalways used as a testbed to generate
new sounds so bring them on.
Regards, nick
"I had to enter a password that needed eight characters. I used "SnowWhite and
the Seven Dwarves",' Vine, 2011 Edinburgh Fringe (I think).
_______________________________________________
Linux-audio-dev mailing list
[email protected]
http://lists.linuxaudio.org/listinfo/linux-audio-dev
