Prophet 5 Vintage knob behavior

[Sorcerio]

I'm loving my new Prophet 5 Rev4 desktop module! However I'm wondering a little about the functionality of the Vintage knob.

When I trigger a single chord repeatedly from my Akai MPC with the Prophet 5's Vintage knob turned up it does not seem to provide the random, "warm, organic, and alive" quality that I was hoping for as is stated in the owner's manual. It seems to be taking a set of randomized values and repeating those exact same values for each repetition of the same chord triggered.

I made a little recording of an example of what I'm talking about. In this recording, the Vintage knob is turned all the way clockwise to "1". A chord will be played and repeated a few times to demonstrate the effect I speak of. Then I play a different lower register chord one time to "reset the randomization snapshot" (my term) and return to the original chord and repeat it again a few times.

Here's the sound clip: https://www.sndup.net/9d9n When you click the "Download audio!" link at the bottom of the page it should just open in your browser as mp3 audio.

I'm wondering if this is working as intended or if there is another mode of operation for the Vintage knob I can switch to that would impart randomness to each and every repetition of a note or chord played.

[Dsil303]

When you put the synth in round robin mode the synth will "go around" each voice, instead of using the same voice again.

Each set of randomised values differs per voice, but indeed on the same voice it does not change the values after a note on\off

So in RR mode you will hear it the way you thought it would be.

I had exactly the same thing

[Sorcerio]

Ahh thank you! How do you put it in Round Robin mode? Is that documented in an addendum somewhere? I couldn't find it in the main manual.
Nevermind, I just found it on the changelog page. Thanks again!

[Dsil303]

You're very welcome 👍

[waxdoctor]

When you put the synth in round robin mode the synth will "go around" each voice, instead of using the same voice again.

Each set of randomised values differs per voice, but indeed on the same voice it does not change the values after a note on\off

So in RR mode you will hear it the way you thought it would be.

I had exactly the same thing

I tried it out some time ago (on my P5) to see how much the sounds per voice differ from each other.
It sounds like a repeating pattern to me. I also thought it was always recalculated randomly (per key trigger).
But now I think it's just certain programmed and repeating values, and the vintage knob just regulates these values in intensity or attenuation.

(You can try it by hitting just one note repeatedly (in RR) and a simple patch)

[Qwave]

The Vintage knob does emulate the per voice variations very well. The randomized offsets are per voice and not just randomly reassigned to each new key lik on some other brands. I prefer it do be per voice for the most authentic vintage feel.

[waxdoctor]

The Vintage knob does emulate the per voice variations very well. The randomized offsets are per voice and not just randomly reassigned to each new key lik on some other brands. I prefer it do be per voice for the most authentic vintage feel.

Okay, then mine is broken, because there are no random offsets in the "vintage" values.
There are always 5 fixed, but different, vintage offset values, which are applied stronger or weaker depending on the setting of the vintage knob setting.
and yes, with key licks or chords, you don't notice it so much because these offset values are distributed/assigned differently(but still within a 5-voice counter range, before it repeats, I want to believe), so it also depends on whether even or odd numbers of notes are played.
Again, maybe my P5 is broken, or maybe there are different actions on different OSs. (I am on OS  2.0.4. / 1.1.3)

I hope my English is understandable as it is not my mother tongue and I also do not want to be a smart-ass here, or kill the fun, it is just my observation.

[Qwave]

The Vintage knob does emulate the per voice variations very well. The randomized offsets are per voice and not just randomly reassigned to each new key lik on some other brands. I prefer it do be per voice for the most authentic vintage feel.

Okay, then mine is broken, because there are no random offsets in the "vintage" values.
There are always 5 fixed, but different, vintage offset values, which are applied stronger or weaker depending on the setting of the vintage knob setting.
and yes, with key licks or chords, you don't notice it so much because these offset values are distributed/assigned differently(but still within a 5-voice counter range, before it repeats, I want to believe), so it also depends on whether even or odd numbers of notes are played.
Again, maybe my P5 is broken, or maybe there are different actions on different OSs. (I am on OS  2.0.4. / 1.1.3)

I hope my English is understandable as it is not my mother tongue and I also do not want to be a smart-ass here, or kill the fun, it is just my observation.

I think, you got me wrong. The offsets are the vintage changes. So one filter is slightly tuned higher or lower. Or the decay is a bit off compared to another. These offsets are random. The amount of offset is changed by the vintage knob. But just as I wrote and you confirmed, these offsets are not randomised each time you play a new note, but assigned to one voice. so you get the same detuning or other variation, if the hardware voice is reused in the note assignment.

[LPF83]

Anyone please feel free to correct me here, especially if you have hands-on experience calibrating vintage synths.  My understanding is that on an older synth,  envelope and tuning anomalies / variations from original manufacturing spec would likely occur per-voice and would mostly be reproducible each time you played that voice (as already noted, on the Rev 4 whether or not that's per key or not depends on voice allocation mode).  Minor variations, if that were the case, would then tend to occur with temperature fluctuations, but once a synth is at a stable temperature, would be relatively consistent within a specific unit. 
Again, don't hesitate to keep me honest here, those who know..  this is just my current understanding and not being sold as fact.

[Quatschmacher]

They are stable, repeatable per-voice offsets. This is by design and reflects measurements taken from vintage synths, where voice-to-voice variations would occur due to component tolerances. So for example in a vintage unit an oscillator’s tuning might be off by say 5 cents; the vintage knob will set the “depth” of that offset between in tune (at 4) and 5 cents off (at 1).

The only thing I dislike about the vintage knob on the P5/10 is that it applies offsets to the amp envelop sustain level so at certain settings you get audible sound after the decay stage has completed even if the envelope sustain is set to zero. I think they may have removed that when they added vintage knob to P6/OB-6; they certainly made some refinements to the implementation. I wish they’d tweak the algorithm in P5/10 to fix that too as zero should always be zero, otherwise it renders certain settings useless.

[waxdoctor]

They are stable, repeatable per-voice offsets. This is by design and reflects measurements taken from vintage synths, where voice-to-voice variations would occur due to component tolerances. So for example in a vintage unit an oscillator’s tuning might be off by say 5 cents; the vintage knob will set the “depth” of that offset between in tune (at 4) and 5 cents off (at 1).

The only thing I dislike about the vintage knob on the P5/10 is that it applies offsets to the amp envelop sustain level so at certain settings you get audible sound after the decay stage has completed even if the envelope sustain is set to zero. I think they may have removed that when they added vintage knob to P6/OB-6; they certainly made some refinements to the implementation. I wish they’d tweak the algorithm in P5/10 to fix that too as zero should always be zero, otherwise it renders certain settings useless.

something like this is what I've tried to explain - yours sounds much more professionally founded...
so real randomization in the code/algorithm is off the hook then, and it is more a "pseudo randomization", done by the "even/odd voice counting logic" (combined with P5 or rr settings) which I've tried to explain as well, haha... thanks

[waxdoctor]

Anyone please feel free to correct me here, especially if you have hands-on experience calibrating vintage synths.  My understanding is that on an older synth,  envelope and tuning anomalies / variations from original manufacturing spec would likely occur per-voice and would mostly be reproducible each time you played that voice (as already noted, on the Rev 4 whether or not that's per key or not depends on voice allocation mode).  Minor variations, if that were the case, would then tend to occur with temperature fluctuations, but once a synth is at a stable temperature, would be relatively consistent within a specific unit. 
Again, don't hesitate to keep me honest here, those who know..  this is just my current understanding and not being sold as fact.

I have a similar opinion - but I have neither practical experience nor am I an expert.

[Quatschmacher]

They are stable, repeatable per-voice offsets. This is by design and reflects measurements taken from vintage synths, where voice-to-voice variations would occur due to component tolerances. So for example in a vintage unit an oscillator’s tuning might be off by say 5 cents; the vintage knob will set the “depth” of that offset between in tune (at 4) and 5 cents off (at 1).

The only thing I dislike about the vintage knob on the P5/10 is that it applies offsets to the amp envelop sustain level so at certain settings you get audible sound after the decay stage has completed even if the envelope sustain is set to zero. I think they may have removed that when they added vintage knob to P6/OB-6; they certainly made some refinements to the implementation. I wish they’d tweak the algorithm in P5/10 to fix that too as zero should always be zero, otherwise it renders certain settings useless.

something like this is what I've tried to explain - yours sounds much more professionally founded...
so real randomization in the code/algorithm is off the hook then, and it is more a "pseudo randomization", done by the "even/odd voice counting logic" (combined with P5 or rr settings) which I've tried to explain as well, haha... thanks

It’s not random at all; the offsets are created by a lookup table. At a given vintage knob setting, the offsets applied to the parameters of a given voice will always be the same. (Contrast this with the original “slop” function of the P6 where the offset value constantly moved as it was basically being moved by a really slow LFO.)

There’s a very good explanation of this here https://www.voicecomponentmodeling.com/

Jason’s stellar and painstaking work has been incorporated into several synths.

[waxdoctor]

They are stable, repeatable per-voice offsets. This is by design and reflects measurements taken from vintage synths, where voice-to-voice variations would occur due to component tolerances. So for example in a vintage unit an oscillator’s tuning might be off by say 5 cents; the vintage knob will set the “depth” of that offset between in tune (at 4) and 5 cents off (at 1).

The only thing I dislike about the vintage knob on the P5/10 is that it applies offsets to the amp envelop sustain level so at certain settings you get audible sound after the decay stage has completed even if the envelope sustain is set to zero. I think they may have removed that when they added vintage knob to P6/OB-6; they certainly made some refinements to the implementation. I wish they’d tweak the algorithm in P5/10 to fix that too as zero should always be zero, otherwise it renders certain settings useless.

something like this is what I've tried to explain - yours sounds much more professionally founded...
so real randomization in the code/algorithm is off the hook then, and it is more a "pseudo randomization", done by the "even/odd voice counting logic" (combined with P5 or rr settings) which I've tried to explain as well, haha... thanks

It’s not random at all; the offsets are created by a lookup table. At a given vintage knob setting, the offsets applied to the parameters of a given voice will always be the same. (Contrast this with the original “slop” function of the P6 where the offset value constantly moved as it was basically being moved by a really slow LFO.)

There’s a very good explanation of this here https://www.voicecomponentmodeling.com/

Jason’s stellar and painstaking work has been incorporated into several synths.

Great, thanks for that!

[strytten]

I really enjoyed this thread - thanks to all who contributed. I also enjoyed the link to the article on voice component modeling. Great stuff! I recently ordered a Yorick LFE and I'm looking forward to experimenting with some of these concepts by applying some slow frequency LFO settings on my Prophet 10 rev4 and some other synths that this unit supports.