Even further OT, I hear that Microsoft has a patent on algorithmic
composition!?! Don't know whether that is true, or where to begin
looking it up, but has anyone else heard this?
On Mar 2, 2007, at 3:52 PM, Richard Dobson wrote:
Jeff Evans wrote:
? Seems to me that what Kurzweil is talking about is merely that
they offer a chance to specify the Hertz of a pitch instead of
using a named pitch predefined by equal temperament. This would
allow one to affect the beat rate produced by harmonic
combinations, tuning for example a major third so that it was
beatless, as in "just" intonation.
It could also be used, true, to create an artificial tremolo by
slightly detuning two waves of the same nominal pitch, but the
process is not automatic as it is in the patent. The patent covers
an automatic method, simple as it is, of making an artificial
tremolo constant over a wide pitch range.
But that doesn't mean I believe the the patent office actually
looks at these applications.
What's to automate? Detuning by interval is a common procedure (as
used for example in harmonizer efffects) - the interval might be as
much as an octave, a major third, or some fraction of a semitone.
The second way is a fixed frequency offset, as in the Kurzweil
example - just as common. It is simply a matter of providing the
parameter to the user. Both will likely be provided together, so
you get conventional keyboard tracking ~plus~ the constant-Hz shift
- as clearly described in the Kurzweil manual, and in the patent
description and "circuit" diagrams.
There is even a third way to frequency-shift detune, which is to
apply the fixed offset harmonic by harmonic, which ostensibly can
only be done when the synthesis is by additive oscillator bank.
This makes literally all beats the same - but of course the timbre
detuned in this way becomes increasingly inharmonic (883/443 != 2)
as the difference increases. It is a standard additive method to
achieve structured inharmonic timbres. The sort of thing that might
be left as a student exercise in classes teaching Csound, Max/MSP,
etc. We even have pvoc (+ peak detection) processes dating back
decades, that apply this linear frequency shift to a sampled sound.
All three methods are therefore long-established and standard
practice. Any patent can only be valid if it finds some cheaper,
faster or otherwise novel way to do it. To do this it first has to
establish a ~genuine~ problem, where solutions either do not
exist, or are expensive or intractable for one reason or another.
There is no evidence of any such advantage even claimed in the
patent (and, um, patently, the "problem" hardly exists as such),
which simply thinks it has discoved the idea of frequency shift.
Unless of course they are trying to patent addition?
Richard Dobson
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/coreaudio-api/chad%
40chadawagner.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/coreaudio-api/email@hidden
