DSP Robotics Support

Is it possible to separate poly signal/channels into multiple poly signals after say, an interpolated read? like post read pre poly to mono, 1 poly stream>>> splits into respective poly monos >>> polytoblue....

I’m not certain what you want to achieve but the mono to poly module in the attached schematic might help.

I’ve made use of this for turning a monophonic LFO into poly, rather than using a poly LFO. That way the LFO is free-running but all opened poly channels (notes) get the same in-sync LFO modulation.

Cheers

Spogg

In short, no - splitting a poly stream into one stream per-voice isn't possible. A poly stream is effectively an array of mono4 streams, one for every four voices that are playing - but there's no way to extract any particular set of four voices from this array.

The only way to apply different processing to individual voices is to use the 'voice tag' and/or 'id' outputs of the 'Voices to Poly' primitive to alter the parameters of whatever poly processing you are doing. Depending what it is that you're trying to do, this might offer a work-around for you.

Agree with Trog, short answer is "no". But there is also a long answer "yes,but somewhat tricky". I have looked into this because FS' poly implementation does have some limitations. Here is a set of modules for 4, 8, 16 and 32 voices. However, once you go this road you have to do everything in blue: enveopes, fast retrigger etc. I am using this system in Randy, btw.
Have fun!

Holy hell martin, that's not what I asked for, but that's more of something than I could have imagined... I'm sure I'll find a way to make use of this somewhere in the future. Thanks !

martinvicanek wrote:But there is also a long answer "yes,but somewhat tricky"...

I hadn't really looked into how you were processing things when I saw it in Randy, but now that I've had a closer look, I must say that it's very ingenious - the cyclic multiplexer is a masterful hack indeed!

trogluddite wrote:but there's no way to extract any particular set of four voices from this array.

REDACTED! - one of those times when it's very nice to be proved wrong!

I've experimented with blue 'voice cards' myself in the past, but my solution was rather different - a custom voice allocator to replace the 'Voices to Poly' entirely. I had done it in the past with 'green', but found I could make it a lot more flexible in Ruby. I imagine that this is much less efficient than your version, and note-ons may not be sample-accurate, but it means that you can customise how notes are assigned to voices too. In particular, it makes unison voices easy to set up; a single note can play as many voices in parallel as you want, with the polyphony reduced in proportion to stay within the maximum number of voices.

As usual, I never actually got around to making a synth with it, aside from a 'proof-of-concept' prototype...