Patchení s Nikol - Karplus-Strong algorithm 

I would just add, that you cannot put the LP filter in the feedback if you want good tracking. No matter how good the delay module is if the LP filter in feedback adds additional delay, which offsets the 1v/oct almost unpredictably. (ok, it is predictable, but it is a very complex issue. You cannot easily compensate for the filter's additional delay, because 1) The filter causes different delays to different frequencies (unless you use additional all-pass stages to equalise group delay response or bessel and gaussian are quite behaved alone...but do such VCFs exist in modular? ). So the delay time is not a number but a graph... and that's before we even touched the filters cutoff freq knob. 2) The whole group delay graph change when we turn the cutoff freq knob of the filter. In digital, you usually compensate for it by writing down all the delay times (that is delay time for each input frequency and each cutoff frequency setting) into a 2D table and then subtracting that from the main delay time. That's how it's done in yamaha VL-series synths. In analog you cannot do that easily, so you either not put time-screwing elements into feedback, or you just accept thatt tracking will suck and will have to tune by ear...

+Angakok Thoth correct me if I'm wrong, but wouldn't this be really easy to fix by having the filter track along with the (preferably 1v/oct) delay? As long as the phase shift is the same relative to the delay time it shouldn't be a problem.

@@ngkktht774 You are correct this is a big issue with this kind of synthesis. However the solution is easier than you suggest. Elaborate tuning tables for each note is a brute force solution, it works but its not very elegant. The problem is caused because a filter adds a fixed extra delay to each note (linear) - but frequency itself is an exponential scale. So in practice the filter is adding a different offset for each note. However you can compensate for that by running the pitch CV through an attenuator which can subtract a fixed linear amount from the CV voltages so you can compensate for the linear offset caused by the filter. Its true you have to retune things if you change the filter setting but actually its not that difficult to do - just involved moving the attenuator knob a bit. That's the system I use in my own physical models it works in both the digital and analogue worlds and is far simpler to implement than elaborate tuning tables. I can get accurate tuning across three to four octaves even in the analogue world, more in digital.

still?