Random LFOs

It would be great if one of the wave choices for the LFO was a random wave; or perhaps something like the "humanize" parameter on the Microwave XT. Currently all LFOs are periodic . . . I'd like to see some randomization possibilities!

 

Yes, random LFO would be nice, although perhaps just a smoothing option on the Sample/Hold mode would do the trick. I'd love to have a random LFO with a whole set of adjustments covering max/min amplitude, period, slope, etc. However, read on...

You might try choosing Noise 1 or some such, setting the Sec period to very low (like 0.0101), and enabling Sample/Hold. Note that the combination of the LFP playback rate (the upper Sec field) and the S/H sampling rate (the lower Sec field) determines where on the selected curve the samples will be made. (Sorry I couldn't say that more clearly...) Play with the relative values using a Sine LFO, and you'll see how the S/H values can come closer in pitch or farther apart, independent of the speed of the output.

For example, with almost any waveform, you can set a very slow wave rate like 130, and a very fast S/H rate like 0.05, and it will sound a lot like a random waveform, complete with apparent smoothing. I've attached an MP3 of an LFO modulating an oscillator's pitch using this approach. Sections sound smoothed, and sections sound stepped, which would also happen with a truly random waveform.

You can also just use a random Envelope. For testing, create a patch with Osc A set to Single and Sine.

Then go to the Envelope page and make a +New envelope for Oscil A Main Pitch.

Create an Envelope waveform using the Transform dropdown: Load Envelope from Library, using one of the Random selections.

Then click Transform again and use the Scale dialog to adjust the envelope's properties. You can scale the overall range of the envelope, and offset it higher and lower, change the slopes of all the nodes at once, and stretch or compress the time scale, all of which give you a ton of control.

Admittedly, it's not really random, but there's a lot of power to tweak the pseudo-ness of the pattern into something that can sound seriously random while never hitting any really inconvenient values.

In the zip I've also included an ultrasimple AB4 patch (ksd) using both the LFO and the Envelope approach. It defaults to just the LFO, but if you disable the LFO and turn off BYPASS on the Oscil A Main Pitch waveform, you can experiment with both approaches.

ac

 

Hm, I'm probably misunderstanding you, but... with slow LFO rate and fast S/H rate wouldn't the result be much like the original waveform without S/H, just slightly steppy? The extreme would be sampling at audio rate, no?

Anyway, for this technique I recommend to choose rates that are related by prime numbers or some such. This way, there will be almost no repetition of a pattern. And yes, smoothing (glide) for the sample values would help a great deal, I've suggested this a while ago - actually I'd be happy with just that.

Another option is using just VERY slow LFO rates (note that it goes up all the way to 999.999sec) along with truely chaotic waveforms, so consecutive values really "jump" around (heavy fractalizing is a good way to achieve those). Aby's LFOs will interpolate between samples, so the transition is smooth and not steppy. Tbh, I don't understand why this is not available as an option for S/H. Also it should be noted that this does NOT apply to envelope LFO, these will play in steps.

And lastly, don't forget A4's new ability of crossmodulation for LFOs and wave morph. LFO morph, depth, rate and S/H rate are even available as destinations for audio mod and velocity. So there's a whole new universe of (controlled) chaos to discover. ;-)

A bit OT, but the waveshaper with certain waveforms along with appropriate settings (experiment!) can produce some great chaos as well. With the new flexible module routing and audio mod this could be just what the doctor ordered.

werner

 

Werner, yes, you're right -- the limit of high S/H rate vs low LFO rate would be similar to just playing the wave. But if you play with the relative speeds, you can get samples coming out at a desired rate, with values close enough together to simulate smoothing.

Of course it depends entirely on all 3 frequencies -- the LFO wave itself (it's technically a shape, but any periodicity will affect the outcome), the LFO rate, and the S/H rate.

I think of it as "walking the waveform" -- one interesting effect is to set the relative speeds so the LFO samples are almost single-stepping the wave table.

An LFO smoothing control would indeed give us a lot, but if we're going that far, a couple of additional parameters would add a lot more...

OTOH, Envelopes are so flexible, and so much more useful, that I'm happy to use them instead. But it would be best to have both. It's interesting that even with envelopes, LFO's are still relevant.

All my comments on this topic, btw, are based on the idea of controlled pseudo-randomness, not anything truly random, or even truly unpredictable. I like being able to prevent the randomness from hitting inconvenient values or exceeding certain limits, so Envelopes are a more ultimate solution for me.

Still, I'm always eager for more ways to create completely algorithmic output.

ac