DSP Robotics Support

Hi,
This is just a short demonstration of a concept that I used in some of synths, inspired by native instruments "session horns" kontakt library. The main idea behind this synth is randomizing the sound on each pressed midi note by sending each of the 4 synth's oscillators to it's own frequency range area. Indeed, when you decide to go random, there are no limits for what can be achieved, so this synth is only a starting point to much complex designs. Hope you'll like it.

Cheers

I think that’s a great concept for producing sound variations.

The problem I heard while auditioning was random clicks, caused by stream interruptions due to the selectors causing schematic re-compilation. There is a small delay due to the Ruby random number generator and array system sending in green to update the routing selectors, just after the sound has started.

It’s a fascinating idea in principle, but I think it needs Adam’s stream-based random number generator and a stream-based audio gating/routing system to work well without glitches. Go for it!

Cheers

Spogg

Spogg wrote:I think it needs Adam’s stream-based random number generator and a stream-based audio gating/routing system to work well without glitches

Thanks for review it, spogg. You spotted some points I've even didn't think about. How do I find those modules? Thought the slight delay came due to the small amount of attack from the envelope.

I really like this idea for giving a bit more "organic" variation in the sound. I agree with Spogg; it's a good demonstration of the kind of effect that can be achieved, but if it was possible that the variation was "locked in" per-voice, that would make it an incredibly useful technique.

Interesting!

kortezzzz wrote:

Spogg wrote:I think it needs Adam’s stream-based random number generator and a stream-based audio gating/routing system to work well without glitches

How do I find those modules?

I've attached Adam's stream-based random number generator. It creates a new number in the range you specify every time a new channel is opened (every new note played).
Also an illustration of what I mean by audio gating at sample rate.

Routing would probably have to be in the poly stream.

Cheers

Spogg

Thanks, Spogg (and Adam) - I hadn't seen those modules before; they look rather handy!

Thanks spogg!

I looked into these modules and they look quite useful for some tasks, but not in this case I think. What we need here is more related to arrays randomizing. If the values 0-3 are not generated without duplications, each to different path, the concept won't work. For gating I used a slight different method but in the blue section. The results are very similar. Also changed the trigger source from the midi mono output; frequency changes based trigger makes more sense, I think.

The original download link updated (in the first post)

This sounds less clicky for sure!

The issue I immediately noticed is that if you hold a note or chord, and then play others, the held notes change with every new note. For some situations that might be an interesting effect, but I don’t think it’s what you had in mind. Of course it’s fine for mono playing.

I didn’t give too much thought about it yesterday when I sent the random number thingy. I’ll try to crystallise my idea:

The random number created when a channel opens would have to be turned into a stream integer, say 0-3. That number would then open a single audio stream gate to pass into 1 of 4 blue filter channels. A situation could arise where you get some channels where the audio passed into the same filter, since the numbers generated would not be unique. But would this really matter? The benefit would be that the randomly selected blue filter would hold for the duration of the note, until after release time has elapsed in the ADSR.

To sum up, the random routing number would come at the output of each ADSR, and be directed into one of the combiners and thus into the chosen filter by a set of 4 audio gates. You then don’t need any MIDI for routing, since the software opens the channels for you, and creates a random route for each note.

I haven’t tried this, but as a thought experiment only, I think it might fly.

Cheers

Spogg

A single random number could be used in conjunction with simple look up tables, possibly - there are only factorial(filter_count) = 24 possible routing combinations. The table could be arranged as a 2D matrix with a row for each routing and a column for each filter - so only a 24 * 4 = 96 element table would be needed. It should be quite simple to work out a Ruby routine that could populate the table automatically for any given number of filters, then convert float array to a mem for stream lookup.