Keyboard tracking feature for delay modules

[davep]

How about a button on one of the delay modules to allow it to accurately track the keyboard, for DIY physical modeling patches?!?

I know we have the comb filter and the string osc for PM stuff, but you can do a lot more types of PM sounds using delay modules with a feedback loop, like all of the cool NM Classic PM flute patches from Chet Singer (like the classic "CS mono flute").

But without a keyboard tracking feature like this, you have to figure out some incredibly convoluted way to process the keyboard signal so you can use it to control a delay module in a way that will generate the proper pitch.

I would think this would be pretty simple to implement, no?

[blue hell]

This might not be as easy as it seems.

Not only the delay itself determines the frequncy of the instrument but other delays in the circuit are important as well, especially for higher frequencies a few samples delay somewhere might matter.

But you are right, looking at Chet's current PM patches the tuning is, well, not very pleasant to look at.

Jan.

[Tim Kleinert]

Regarding the tuning issues induced by delays caused by added modules in a feedback loop, wouldn't it perhaps be better to see some dedicated physical modelling modules in the first case?

Perhaps something along the line of the Yamaha VL1-concept (which divides between drivers and resonators) would be nice. AFAIK, Clavia already considered this and IIRC, I saw something like a beta version reed model on a Clavia Musikmesse coverage somewhere, but I heard they were reluctant to implement it due to the prospect of potential lawsuits caused by patent infringements.

But Korgs new OASYS and Alesis' new FUSION synths are brim full of those typical Stanford-type physical modelling algorithms, so maybe some patents have expired by now.

A modal oscillator with assorted drivers (à la Tassman) would be nice too.

Sorry for the OT, but you just got me thinking again.

[Chet]

I agree with the idea of some dedicated modules. While it's great that you can roll your own stuff using the current modules, patching around the tuning issue is just too trying for most people's patience.

It would be great to see some true 'string' and 'pipe' modules. That's where the problems are: keeping the resonators in tune while changing the cutoff frequency of the loop lowpass filters. It can be done: Clavia could implement an interpolating two-dimensional table, with one dimension being the pitch, and the other being the LPF cutoff frequency. That way, you could have 'pitch' and 'damping' knobs, both with modulation inputs, and the resonator would always be in tune. In the best implementation, you'd have a choice between a 1pole or 2pole filter.

My guess is that the patents governing resonators like this should expire soon, if they haven't already. I think these resonators were among the first things patented.

One final thought: monophonic wind patches sound pretty synthetic when played through a single resonator. A dual-pitch resonator that crossfades from the previous to the current pitch is pretty important for simulating note-to-note transitions.

[Robby]

First of all, there already is a delay module that accurately tracks
the keyboard. It's called the string oscillator. It really is just a delay
with a hard-wired feedback loop (called decay) and a low-pass filter
on it (called damping).

You can turn off that loop by turning decay all the way down,
and then patch your own instead. I posted a patch like this
in the acoustic patches area.

Even with a delay module that accurately tracks the keyboard,
you have tuning issues as long as you allow users patch their own
feedback loops. Everything on the feedback loop adds some processing
delay, and this delay is constant. It's insignificant for lower
pitches, but it makes more & more of a difference at higher pitches,
so the thing tends to go flatter & flatter at higher pitches.

Because the string oscillator's feedback loop is hard-wired it can
"know" how much time the loop requires and compensate for it.
Hard-wired loops kind of defeat the modular-ness, though.

Possibly Clavia could add a delay module with "inserts" on the feedback
line that calculate the amount of processing time required by the
modules on that line and then add pitch compensation for whatever is
there.

This would be a full general solution to the problem.

Still, I've been able to get 5 octaves in tune to +/- 5 cents with just the
keyboard level scaling module. What acoustic instrument are you
likely to model that has a range greater than even 4 octaves and stays
in tune to an accuracy greater than +/- 5 cents? Unless you're modeling
grand pianos or pipe organs, it just doesn't happen. Even the guitar only
has a range of about 3.5 octaves.

Also, since the issue is that pitch is exponential while the delay added is
linear (constant), an exponential function with the right parameters should
give you completely accurate pitch compensation - in theory.