Voice Component Modeling with the Prophet Rev2

OakBloodThree · « **Reply #20 on:** March 19, 2019, 08:34:55 AM »

Quote from: maxter on March 14, 2019, 04:59:21 AM

Yes, intonation offsets will be great! I have to try this. I think this plays a major part in the VCO sound, and I'm a big fan of "stretching" the tuning on the upper and lower registers, like on a piano.

Stretching can also be [fairly] easily accomplished with an alternate tuning. Tunings do not have to have a 2/1 octave. For example, using 19 equal divisions of the tritave (3/1) gives you a perfect twelfth and a slightly stretched octave.

(I have actually been experimenting a bit with regular "squeezed" tunings instead, to make thirds and sevenths "sweeter" at the expense of octaves and fifths.)

creativespiral · « **Reply #21 on:** March 19, 2019, 09:59:50 AM »

Quote from: panic on March 19, 2019, 08:04:48 AM

, it just doesn’t go as far as creativespiral had hoped in mimicking certain behaviors as voice count or predictability of results.

Overall, the voice behavior is very predictable with this modeling.. ie: if you setup a "virtual voice count" of 5-voices in the gated sequencer, and then play a series of three-note chords, you will get the correct offsets of each voice/note. (1,2,3) (4,5,1) (2,3,4) (5,1,2) (3,4,5) (1,2,3) etc... just as you would with a cyclical voice scheme on a real five voice synth.

The only areas where it doesn't act exactly like real X-voice synth is if you play more than that amount of notes at once, you don't get the voice stealing... which actually isn't necessarily a bad thing. And as mentioned, some synths like the Jupiter-8 use a voice reset/backtracking scheme.

But overall I'm very happy with the results... It's a pretty accurate way to model 4-voice, 5-voice, 6-voice or 8-voice vco synths from the past.

Also, I'm having great results with patch design of other acoustic/analog instruments - Modeling trumpets, flutes, violins, cellos, string ensembles, guitars, pianos, organs, etc. For these type of real world instruments/ensembles, having unique (and stable) tuning offsets per voice is very realistic. I'm convinced this is the best approach... everything sounds more organic, and chords have a warmth/richness that can't be achieved with other methods like lfos and slop.

creativespiral · « **Reply #22 on:** March 19, 2019, 10:19:28 AM »

Quote from: maxter on March 15, 2019, 08:25:22 AM

Well, had time to do just a little more testing. It does indeed work the way creativespiral intended. If you switch mode and then back, the sequencers get off-set somehow.

Yeah, I've noticed there is an intermittent bug in the gated sequencer / firmware where if you are doing a bunch of step editing / switching between poly and gated seq / changing reset positions around, then some step resets will sometimes get "stuck on"... I've got to do some more experimentation to lock down the exact repeatable behavior and send a bug report on to SCI... it seems to be very intermittent, but I have run into it a couple times out of fifty or so patches I've setup so far using this method.

If you power cycle it does fix itself. And once you have patches saved with the gated seq steps correctly, they do work as intended and have repeatable virtual-voice allocation behavior every time you load them.

Proton · « **Reply #23 on:** March 19, 2019, 10:32:48 AM »

I´ve tested the patches you´ve thankfully sent to me and implemented this concept in some of my patches. you are right: the result is an organic and rich voice behaviour - thanks again - fantastic work !

creativespiral · « **Reply #24 on:** March 19, 2019, 10:40:59 AM »

Quote from: bldrman on March 16, 2019, 06:41:08 AM

Really nice work! But what is the difference here using this method instead of using one or two LFO's with destination OSC SLOP? Set LFO1 to Triangle and LFO to Random both having OSC SLOP as DESTINATION.

If you're referring to the VCO Harmonic Jitter modulation experiment:
I did test out replicating the VCO frequency jitter with Osc Slop Modulation, but was not able to get the same results as measured on the three VCO synths I tested. The reasons are described in more detail down at the bottom of this thread (https://forum.sequential.com/index.php/topic,3412.0.html)

If you're referring to the Voice Modeling:
The short answer is that randomness and pronounced detuning motion is less realistic than having voice offsets that are more stable. You may still want some randomness and tuning motion, but that should be secondary to setting up the basic voice offsets first. In the web page I setup, there's a bunch more details and info: http://www.VoiceComponentModeling.com/vcm.aspx

panic · « **Reply #25 on:** March 19, 2019, 03:22:17 PM »

I could open the clip from philroy before, but now that i have seen it, it is clear that you were indeed right on this, and that age is eating my brain away more rappidly than i d hoped. Thanks for sharing, very useful insight for me!

nanoengguy · « **Reply #26 on:** March 19, 2019, 06:51:17 PM »

Quote from: maxter on March 14, 2019, 04:59:21 AM

I'm a big fan of "stretching" the tuning on the upper and lower registers, like on a piano.

Hi Folks,

I can't really wrap my head around how one would program this, and I'm enthralled by the concept.

Any ideas?

Thanks!

creativespiral · « **Reply #27 on:** March 19, 2019, 07:13:44 PM »

Quote from: nanoengguy on March 19, 2019, 06:51:17 PM

Quote from: maxter on March 14, 2019, 04:59:21 AM
I'm a big fan of "stretching" the tuning on the upper and lower registers, like on a piano.

Hi Folks,

I can't really wrap my head around how one would program this, and I'm enthralled by the concept.

Any ideas?

Thanks!

There's two different approaches to changing tuning:
1. If you're going for accurate "Stretched Piano Tuning", your best bet may be to load an Alternate Tuning definition table to the Rev, which sets specific tuning templates to the keys. Page 71/72 of the Rev2 manual discusses these alternate tunings. Also, I saw that OakBloodThree just uploaded an updated syx here (https://forum.sequential.com/index.php/topic,3460.0.html), and you might wanna follow that thread.

2. You can use the Mod Matrix to route Note Num to Osc Frequency... Note Number amount of 128 means that every consecutive key press increases the output by 1. So if you're routing to Osc Freq, then every consecutive key press will add 1 to Osc Freq (aka +50 cents)... that's on top of Osc Key Tracking and Osc default frequency. Also, you need to add / account for the first 36 midi notes/keys before you actually get to the first key on the physical Rev2 board. (in default octave) So, yeah, using Note Num modulation, you're probably gonna want very low numbers for the amount. This is the method I'm using for "intonation tuning offsets" with voice modeling... to mimic behaviors I've researched on VCO synths. Routing even an amount of '1' with Note Number is too much in most cases,.. I'm using gated sequencer (along with two mod slots) to scale down the note number modulation to less than 1. If you're interested in an example patch, send me your email and I'll send you over some examples of this method... although, as mentioned, the alt tuning definitions might be more what you want if you're shooting for classic stretched piano tuning.

creativespiral · « **Reply #28 on:** March 31, 2019, 12:51:58 AM »

Got my first video up that shows off some voice modeling with the Prophet Rev2:
Creation of a Giant Brass Pad Sound like the Vangelis Blade Runner Theme.

https://youtu.be/jB9HG3k3vvQ

I'll be doing some more videos on voice modeling and Rev2 patch creation soon.. I'm working my way through a custom bank of 128 voice modeled patches.

Proton · « **Reply #29 on:** March 31, 2019, 02:00:12 AM »

perfect !

Frocktar · « **Reply #30 on:** March 31, 2019, 08:06:01 PM »

Really great work on this mate.

Thanks for the tip!!!

creativespiral · « **Reply #31 on:** March 31, 2019, 08:53:39 PM »

Quote from: creativespiral on March 19, 2019, 10:19:28 AM

Quote from: maxter on March 15, 2019, 08:25:22 AM
Well, had time to do just a little more testing. It does indeed work the way creativespiral intended. If you switch mode and then back, the sequencers get off-set somehow.

Yeah, I've noticed there is an intermittent bug in the gated sequencer / firmware where if you are doing a bunch of step editing / switching between poly and gated seq / changing reset positions around, then some step resets will sometimes get "stuck on"... I've got to do some more experimentation to lock down the exact repeatable behavior and send a bug report on to SCI... it seems to be very intermittent, but I have run into it a couple times out of fifty or so patches I've setup so far using this method.

If you power cycle it does fix itself. And once you have patches saved with the gated seq steps correctly, they do work as intended and have repeatable virtual-voice allocation behavior every time you load them.

Just to update this item... I did track down two minor bugs with the Gated Sequencer / Key Step mode... they have both been reported last week to SCI and confirmed... hopefully they'll be fixed soon, in next OS update. So if you're trying out voice modeling right now, be aware:

1. When stacking two layers, with both using the Key Stepped Gated Sequencer, the sequencers don't currently keep track of the correct step.. they skip alternating steps. You still get the the voice modeled sound, but if you were trying to model an 8-voice virtual synth, it will actually just use four of the steps you defined. (In this patch I created for the Blade Runner sound, that is the case... It still sounds good with the skipped steps, but once the bug is fixed it will sound even more realistic... IE: more like an 8-Voice CS-80. ) So, not a deal breaker, but something you might notice if you're playing around with stacked, voice modeled layers in OS 1.1.5.9. (this isn't an issue in single layer programs)

2. There is another weird bug where if you set a Gated Seq Reset, then turn it back to a normal step and set a Reset on another step in the track, the reset positions may not be correct... So when editing, try and be careful to set your resets on the right step the first time... Alternatively, if you're moving reset positions around, just save the patch regularly. For some reason, the act of saving the patch corrects the bug (must be something in a sort of temporary buffer). So this one is just an editing inconvenience... once patches are saved, they act as intended.

Anyways, just thought I'd mention those so if ya'all are experimenting with this technique, you don't get frustrated with weird results. Hopefully will be all fixed up soon... and stacked, voice-modeled patches will sound even more organic.

minor7th · « **Reply #32 on:** April 02, 2019, 11:41:50 AM »

Brilliant and very inspirational! I reprogrammed one of my soft analouge pads replacing osc slop and LFO modulations with gated sequence values and boy does it sound much better! Thank you for sharing your ideas

jurfin · « **Reply #33 on:** April 22, 2019, 09:05:52 AM »

Thank you so much for your hard work on this, I made some patches with your techniques and they are absolutely awesome. After doing some testing and comparisons, I have to agree that your method is superior to using slop to recreate subtle differences between voices.

I really appreciate your thoroughness and attention to detail. I’m also incredibly impressed with the creativity behind your approach. It’s such a reasonable solution, I’m annoyed I didn’t think of it first haha!

Would you be opposed to me sharing your site with the tips and tricks threads I’ve made both on this forum and on Gearslutz? It would be a great addition to each of them if you’re alright with me spreading the good news.

Thank you again man, this is really high quality work.

Quote from: creativespiral on March 13, 2019, 09:32:25 AM

I found a way to better capture the character of classic VCO analog poly synths with the Prophet Rev2.

OneLittleFonzie · « **Reply #34 on:** April 22, 2019, 10:51:49 AM »

Quote from: creativespiral on March 31, 2019, 12:51:58 AM

Got my first video up that shows off some voice modeling with the Prophet Rev2:
Creation of a Giant Brass Pad Sound like the Vangelis Blade Runner Theme.

I'll be doing some more videos on voice modeling and Rev2 patch creation soon.. I'm working my way through a custom bank of 128 voice modeled patches.

That was really impressing. Not only for Vangelis style big sounds (which I'm not a huge fane of) bit also for other sounds that could use a bit more life. I wouldn't mind getting the patch. But I could make a base patch and use it as template for new patches. I'm not really into saving patches but this could be a nice new basic patch.

Thank you!!

creativespiral · « **Reply #35 on:** April 22, 2019, 11:15:09 AM »

Hey Jurfin - thanks! yes, feel free to spread the word about these methods / share links.

I got a chance last week to visit a local L.A. synth repair/rental shop and take recordings of several classic synths - to investigate voice variance and harmonic jitter.

I took recordings from a Prophet 5, Prophet 10, MemoryMoog, Korg PolySix, Yamaha CS-80, Oberheim OBX, Oberheim OBXA, Oberheim SEM, plus a Roland Jupiter 4 and Jupiter 8. I'm gonna do some in-depth analysis, and provide some more details on my findings soon.

psionic11 · « **Reply #36 on:** April 22, 2019, 12:39:48 PM »

During your analysis, I imagine your main focus will be on pitch variations. Vintage synths also varied in envelope response -- exponential vs linear vs logarithmic stages for attack and decay. Do you think it worthwhile to explore these as well?

creativespiral · « **Reply #37 on:** April 23, 2019, 10:45:32 AM »

PM'd you with some example patches, OneLittleFonzie.

Quote from: psionic11 on April 22, 2019, 12:39:48 PM

During your analysis, I imagine your main focus will be on pitch variations. Vintage synths also varied in envelope response -- exponential vs linear vs logarithmic stages for attack and decay. Do you think it worthwhile to explore these as well?

Yeah, I'm mostly focused on voice-by-voice pitch variance and harmonic jitter analysis. I agree that envelope variance is definitely present in classic synths... Envelope timing (Attack, Decay) is one of the behaviors I've been modeling into most of my voice-modeled patches (like in Blade Runner example video). It's especially noticeable in patches with pronounced cutoff sweeps... I made a "Tom Sawyer"-like patch that uses two stacked layers with envelope variance... it sounds massive and very organic. Envelope shape is another topic (ie: concave, linear, convex) that is also worth investigating... but beyond what I'm working on for now. Maybe next round.

Pym · « **Reply #38 on:** May 02, 2019, 02:32:27 PM »

Just FYI, I have explored this concept quite extensively from many perspectives and I already have most of the code figured out for future products but it isn’t easy to port back due to design choices. I’ve always been fascinated by chaos theory and how to both bound and seed randomness in a way that emulates the part of analog we all love

I may even have some tricks up my sleeve you haven’t thought of yet... stay tuned

creativespiral · « **Reply #39 on:** May 02, 2019, 03:39:51 PM »

Quote from: Pym on May 02, 2019, 02:32:27 PM

Just FYI, I have explored this concept quite extensively from many perspectives and I already have most of the code figured out for future products but it isn’t easy to port back due to design choices. I’ve always been fascinated by chaos theory and how to both bound and seed randomness in a way that emulates the part of analog we all love

I may even have some tricks up my sleeve you haven’t thought of yet... stay tuned

Awesome, Chris! I'm excited to see your sleeves! This method of defining per-voice variance with a sort of "lookup table" definitely adds some classic analog realism over random slop and more artificial detuning motion. The key is having fairly stable tuning per voice / per oscillator, but with unique offsets for each. And tackling other parameters in the VCO, VCF and VCA sections with per-voice offsets can really create unique analog character.

I went out a few weeks ago and captured samples of a bunch of synths and I have data tables of voice-by-voice tuning offsets for MemoryMoog, OBX, OBXA, OB SEM, CS80, Korg PolySix, Jupiter 4, Jupiter 8, Prophet 5 and Prophet 10... if you're interested. I will say, the predominant characteristic is that all of them have these tuning offsets, and most have significant intonation variance up and down the keyboard.. I've been playing with note number modulater to replicate it.

The handling of voice-allocation would be a good area to develop further, as there are a few distinct variants (round robin, reset to voice one, backtracking voice allocation, per key voice allocation, etc..) Currently the round robin approach is the only one available through the Rev2 modulation.

It's a testament to the design of the Rev2 that it works so well even with current implementation... but yeah, if we just had "another instance" of the gated sequencer dedicated to being a lookup table for this type of per-voice modulation offsets, I think it will be a really interesting feature on a future board. If it had a few more lanes of data (6 or 8 lanes), and ability to capture the different voice allocation schemes, that would be awesome!

Also, I was thinking it would be gravy if the lookup tables could be saved independent of presets... so you could just have prebuilt voice templates for a variety of different classic synths, along with variants of each where they have less-or-more voice character. Then you could load a patch, or an INIT patch, and then navigate through a list of voice modeled presets (Voice Modeling OFF | MemoryMoog Prime | MemoryMoog Hot and Humid | Prophet 5 | etc...... )

At the bottom of the VoiceComponentModeling.com web article, in Section 5, I outlined a bunch of "future wish list" items.
http://www.voicecomponentmodeling.com/vcm.aspx

Excited to see what comes next!

Cheers! - Jason