ACE, Diva, Hive, Repro - release candidates

[e-crooner]

Hey Urs (os someone U-he!), can you have a more detailed explanation about the new 'voice allocation algorithm'?

It's more easy to experience than to explain:

Set any of ACE/Diva/Bazille/Hive to 4 voices. Hold a 3-note chord with your left hand. Play a solo with your right. The 3 note chord will still sound, almost no matter what you do with your right hand (unless you hack like crazy, of course). Now try the same with any other synth. Any other synth we tried will sooner or later drop a note or two from the held chord.

Isn't that the way it should be? I mean, obviously you have a polyphonic patch, so, you will play sloppily with your solo hand and one little overlap of two notes with that hand is enough to reduce the chord of the other. Unless there are actually more than 4 voices behind the scenes.
Also, when you use a polyphonic patch anyway, why limit the number of voices to 4 when virtually any soft synth has at least 8 voices?

[Urs]

Hey Urs (os someone U-he!), can you have a more detailed explanation about the new 'voice allocation algorithm'?

It's more easy to experience than to explain:

Set any of ACE/Diva/Bazille/Hive to 4 voices. Hold a 3-note chord with your left hand. Play a solo with your right. The 3 note chord will still sound, almost no matter what you do with your right hand (unless you hack like crazy, of course). Now try the same with any other synth. Any other synth we tried will sooner or later drop a note or two from the held chord.

Isn't that the way it should be? I mean, obviously you have a polyphonic patch, so, you will play sloppily with your solo hand and one little overlap of two notes with that hand is enough to reduce the chord of the other. Unless there are actually more than 4 voices behind the scenes.
Also, when you use a polyphonic patch anyway, why limit the number of voices to 4 when virtually any soft synth has at least 8 voices?

Maybe this is they way it should be, but I have not seen any other synth work that way.

The 4-voice example is purely hypothetical of course, but it demonstrates that the voice stealing algorithm is "more musical" than pretty much any other I came across.

[e-crooner]

Or you allow the user to set a split point and the number of voices and polyphony mode separately for the two resulting ranges, basically a left- and a right-hand range
This way you could also include a switch for the pitch bend and mod wheels so that either range or both are affected.

Maybe, in poly mode, the synth should itself decide intelligently how many voices are being needed at any moment (probably based on the envelopes being active?) and turn all others off in order to save CPU.

[e-crooner]

Oops, replied instead of edited

[EvilDragon]

Yeah it's true what Urs is saying. I can't find a synth that responds to the example he mentions without dropping a voice or two on the held chord. So basically, notes that are held down are higher priority and aren't as easy to steal.

Maybe, in poly mode, the synth should itself decide intelligently how many voices are being needed at any moment (probably based on the envelopes being active?) and turn all others off in order to save CPU.

This is how u-he synths already work.

[e-crooner]

Of course one might say. I mean, 4-voice limit, 5 notes played, the synth has to cut somewhere and assumes that the notes played more recently are more important.

Another thing I sometimes wonder is what happens if you have set a certain voice limit and play that number of notes plus one, and all at exactly the same time, which one gets ignored?

[paradiddle]

Can't seem to download repro windows. Maybe it's being updated?

[Urs]

Of course one might say. I mean, 4-voice limit, 5 notes played, the synth has to cut somewhere and assumes that the notes played more recently are more important.

Well, there you go, that's what most synths probably assume, and it's a lot less musical than what we do now.

[EvilDragon]

But it's usually a good assumption that the latest notes played are the ones you want to hear, no?


Either way, what you did here does seem to work pretty well.

[Urs]

But it's usually a good assumption that the latest notes played are the ones you want to hear, no?

For people who play with 1 finger, yes

[e@rs]

^

[Howard]

Musing: The voices sustained via keys are more important than those sustained only via pedal.

[EvilDragon]

Yep, that's for sure.

[Mushy Mushy]

I asked this in another thread and Urs was kind enough to reply to me but I do need some further guidance.

Not sure if there's another, newer thread but anyway...

Is there a way to generate (or simulate) a random LFO in Repro1? I'd like to assign a random LFO to the filter cutoff. Possible?

Stretching my knowledge: The recent beta/RC of Repro 1.1.1 has Hive-style modifiers for the ModMatrix. If you use a fast looping AR Envelope (in JAWs) as source and LFO as source for Sample & Hold (right-click) you pretty much get a random S&H LFO. Use right-click and slew to taste.

- Blank patch
- Jaws switched on and release set very miminally (ie, 1 or 2)
- Perform section switch to Wavefolder AR in the main drop down.
- Perform right clicked and selected S&H Trigger -> LFO
- Perform destination Filter with depth all the way up

And the result...... nothin'

[Urs]

JAWs set to minimum is gonna be very quiet.

JAWs’ AR envelope also works while JAWs is on standby.

(that said, I haven’t tried it myself yet)