Here’s an LFO question. I was hoping to use the LFO to randomise a small set of samples (as part of a larger sample pool), but it seems this isn’t so easy.
As a test, I set up a set of three melodic samples with noise samples either side in the pool, like so:
1: noise 1
2: bell 1
3: bell 2
4: bell 3
5: noise 2
The noise samples were distinguishable (one was filtered). The initial sample slot was set to 3.
I set up a random LFO on HOLD, assigned it to sample slot and set the depth to 1, then placed 16 triggers.
I was hoping that a depth of 1 would restrict the effect of the LFO to slots 2-4 (a slot higher and a slot lower than the initial setting of 3) but when the pattern ran, I occasionally heard samples 1 and 5 coming through.
My best guess here is that the LFO is rounding to the nearest number rather than rounding down, so if I’ve managed to get the depth knob at 1.51, I’m going to get some instances of 2 in my results. Of course it’s very hard to set a precise LFO value at that level.
This seemed to be confirmed when I set the LFO depth to 0 and still heard some variation in the sample - the only way I could get the sample to remain on slot 3 was to use function and the depth knob to lock the value at absolute zero.
Any thoughts on insights here? I’m going back to this tonight to see if very careful tweaking can get the result I’m after, i.e. try to stop the depth knob as soon as it hits 1. I thought I’d done that the first time, but maybe I wasn’t careful enough.
I’m also wondering if setting the LFO somewhere between 0 and 1 would solve the puzzle. I guess the question here is how the LFO value is quantised for controls like sample slot that work in discrete steps. I’ll be going back for more investigation, but if this has already been cracked I’d love to know.
Yeah, there’s the animation of the knob turning even if the LFO value remains the same, which would suggest that there are more steps than 128.
I would check the amount of turning in the animated lfo ammount knob between 1 and 2…
Yes, I think the LFO knob, as with other knobs, has much higher resolution than the values it displays (possibly they’re 12-bit, for 4096 values). But given that the display locks the amount to 128 steps (in either direction), I was hoping the displayed value would map directly to parameters with discrete values. There are certainly plenty of instances where you’d want to make use of the finer resolution, but it’d be convenient if discrete parameters took a rounded down value (assuming my guess is correct).
I’m thinking this will be easier to test with the square wave LFO, so I’ll give that a try later and see what happens. If I’m on the right track, I’ll be able to get the sample slot to jump up by two spaces before the LFO depth knob actually reads 2…
you can do exactly as you want with fine adjustment, the depth knob is 14bit (i.e. 128 x 128 if you like, each visible integer step contains a full 7bit range of adjustment)
.
an important #ARtip aspect to keep in mind is that the modulations are centred on the ‘bottom’ of the slot, not the ‘middle’
.
what this means in practice is that if the modulation depth was equivalent to an amount equal to (but just short of) a slot higher upwards at its peak, a square lfo will only alternate between the selected slot and the lower one !
.
intuitively you’d expect(hope) it to be symmetric about the selected slot, it’s not !
.
see squiggles … (it may have made more sense to annotate the pluses and minuses the other way round as per AR)
.
you could ‘fake’ symmetry by doubling the number of slots used, using pairs of identical samples and selecting the upper(lower num id) of the two central ones
Ah, interesting - many thanks! That may well explain things. I was working on the centre-based assumption that I’d need a pool with an odd number of samples (giving me a middle point to pivot around), but it looks like an even number is the solution. Not especially intuitive, but no problem once you know. Thanks again.
yeah, you could have
.
A
B
C__
C
B
A
.
so the range is ‘symmetrical’ or
.
A
A
B
B
C__
C
B
B
A
A
.
if you wanted it to not have equal probability of A,B,C but to be more intuitive - with a random LFO
.
set up fine tuning of depth with a square LFO makes it easy to dial in range
.
or do it with ‘sample chains’ on one slot and vary start/end together by linking to velocity and then mod velocity)
