I got an idea yesterday and did a quick test to confirm that it worked. I’m probably not the first to think of this, but since I haven’t seen it mentioned here before I thought it would be helpful for at least some of you for me to write a detailed explanation.
So, the idea was to program an ever-changing beat by using silent slices and the random lfo to control the probability of a sample firing or not.
A very basic setup would be two slices. The sampe on slice 1 and silence as slice 2. With a random lfo modulating the start parameter, a depth of 0 will play the sample every time (obviously). Depth set to 1, however, will play the sample two out of three times. Increasing the Depth would reduce the odds down towards 50%.
For lower probability, a silent slice 1, the sample as slice 2, silence as slice 3 and slice 2 as default will work better. That way Depth as 1 would fire the sample one out of three times instead. And increasing Depth will quickly drop the probability down towards 1/128.
To have greater control, one can have several slices with the sample. For example:
1: silence
2: sample
3: sample (default)
4: sample
5: silence
Depth 2 would fire the sample 60% of the time, Depth 3 43% etc.
It gets more interesting, though. By using two random lfos, each contributing half of the depth, you’ll get a bell curved probability. In the example above, one lfo with Depth 2 would mean slice 3 has 20% odds of being fired. Using two lfos with Depth 1 each gives slice 3 50% odds instead. The slices at the ends can still be selected, but the odds for that happening are much lower than with a single lfo.
This bell curved probability seems to me to be very useful. For instance, for my test I made several slices with my snare. The “middle ones” had accurate slice points, but the others had gradually more sloppy start points. I still had silent slices at the ends. So by placing trigs all over the place and adjust the lfo depths I could make the snare always play at some key steps, and with slight timing variations. But offbeat “weird” steps would play a snare just once in a while, and when it did, it was more likely to have a sloppy timing.
So, anyways, that was my idea. I’m not sure I’m explaining myself well. But it really is very easy and quick to do. I’m a Octatrack n00b, but was still able to make the following test in less than 15 minutes. It is a single 16 step pattern with 5 samples firing randomly. The “melody” on top is not really relevant to this topic; it is just a random midi arpeggiator with some random lfos assigned to pitch and arpeggiator speed etc.