It depends on the precision of the step sequencer and the rates (even easier with a custom shape).jmg8 wrote: Wed Jan 08, 2020 8:21 pmThat wouldn't work exactly.
The request is for a single cycles to have their own speeds. In a step sequencer the length of each step would be fixed, so for example the first step might last for 3.5 cycles but the second step could be 2 cycles. I believe the request is to have the first cycle at a certain speed and then the second cycle at a different speed etc.
Of course a step sequencer with 10 steps and 1 Hz rate would have 0.1 seconds per step.
But if you use for example a 10-step sequencer with 1 Hz and you set the steps like "1, 2, 2, 3, 3, 3, 4, 4, 4, 4" you would have the value 1 for 0.1 second, the value 2 for 0.2s, the value 3 for 0.3s and the value 4 for 0.4s. The value 1 would then set the rate of the main LFO to 0.1 Hz (with the help of some transformation), the value 2 would set it to 0.2 Hz etc. and in the end you would have run the main LFO with 4 different rates: 0.1 Hz, 0.2 Hz, 0.3 Hz, 0.4 Hz = 1 Hz. Due to this calculation also the second LFO is done after 1 Hz and both are in sync.
Getting values like "3.5 cycles" should be easily possible by using a larger number of steps (or custom shapes). 20 steps for example could be steps of 1 (range 1-20) or steps of 0.5 (range 1-10).
It all depends on the transformation possibilities and some math to calculate the steps and the rate of the second LFO. "1, 2, 2, 3, 3, 3" would be possible with a 6-step sequencer and a 1/6 Hz rate for example.. I guess..
If the step sequencer does not provide enough freedom you could use a custom shape but this depends on how accurate the points can be set. I also think this solution should work for voice and normal modulators. If you not only want different rates but also different shapes you could use the second LFO to modulate for example the shape value of the main LFO with a different transformation to get the desired result or use multiple "main LFOs" and use the second LFO for enabling and disabling them.
Still it is probably easier to just use a custom shape with different shapes connected one after the other, perhaps this can be easily done by exporting existing shapes to csv, merging the csvs and importing the merged csv. Of course to get different rates for each shape you need to scale them independently in the merged custom shape.
Adding randomization to both approaches seems pretty difficult because you need to randomize the value of the second LFO together with the length of its sub-cycles.
I am actually not sure anymore if this even makes any sense.. in theory it does I think but of course it has a huge amount of math involved. Don't even think about modulating anything in this concept..
