Fractal SynthFx ?

[creativespiral]

Just resurfaced from a deep dive in GearSlutz forums regarding sonic differences between old VCO and modern VCO based synths, and someone brought up Roland's ACB/Aira as potentially being a better contender to mimicking an old synth sound than some modern VCO based hardware that use surface mount and modern electronics components that don't match the old electronic components or circuit design exactly. (ACB is Roland's component based software modeling.... similar to what Fractal does for Guitar Effects... modeling each resistor, capacitor, diode, transistor, etc along the signal path)

I was wondering if Fractal would consider going in this direction? Seven years ago, I would have never thought a modeling unit would replace my guitar rig and tube amps, but since becoming an AxeFX user, I'm a believer, and I've sold all but one of my amps and many of my effects.

There is a large market, and high end polyphonic synth gear sells in the $2k to $4k range (or $6k to $8k if you want a Moog One). Many synth players buy multiple units to replicate the sound of old Moogs, Prophets, Jupiters, OBs, Junos, DX7s, and other classics.

I'm imagining a High End Rack Unit with powerful dsp/fpga and internal electronics like the AxeFx, and Fractal's attention to detail, to replicate everything from vintage MemoryMoogs and Jupiters up to modern day synths. As an alternative to FC foot controllers, maybe offer some tightly integrated keyboard controllers with knob per function type mapping, or just allow users to interface with any number of other MIDI synths / controllers already on the market... . Also a nice editor / plugin interface like AxeEdit.

Fractal has developed advanced component based modeling for amp and effect electronics... I would think there's probably already a good amount of overlap and code blocks that could be ported. Of course, the specific variations of oscillator circuitry, synth filters and envelopes could be a lot of work to design... but I would think its right in the wheelhouse of Fractal to accomplish this type of synth modeling, and do it better than anyone else out there... Fractal has disrupted business as usual in the guitar amp / effects field... could they do it for synthesizer market?

I wonder, are there any specific constraints that would make this harder to accomplish with analog synth modeling?... Beside the fact that there's only one Cliff and only so many hours in a day? Is emulating VCOs, VCFs, VCAs any more difficult than what has been done with amp modeling in the Axe FX?

Often the complaints about digital vs analog synthesis comes down to issues with digitally modeling oscillators vs VCOs... but by using a high end dsp/fpga architecture, could this be nullified?... ie: wave aliasing, limitations of modeling non-linear functions, etc..

 

[Muad'zin]

Just resurfaced from a deep dive in GearSlutz forums regarding sonic differences between old VCO and modern VCO based synths, and someone brought up Roland's ACB/Aira as potentially being a better contender to mimicking an old synth sound than some modern VCO based hardware that use surface mount and modern electronics components that don't match the old electronic components or circuit design exactly. (ACB is Roland's component based software modeling.... similar to what Fractal does for Guitar Effects... modeling each resistor, capacitor, diode, transistor, etc along the signal path)

The only thing wrong with modern based hardware that uses SMD and other modern electronics is that its qualitatively too good, as the old electronic components of the past were all over the place in quality and deviation. A 5% deviation in a 100K pot means it can be between 95K and 105K. But a 10% deviation puts it between 90K and 110K. Already 20K extra chance of sonic deviation. And I reckon the old stuff of the 60's and 70's would deviate even more. Add to that degradation of the capacitors, especially electrolytics and you can basically have 2 exact same units that sound different. This was why with fuzz face pedals guitarists were actually auditioning various pedals to see which one sounded best. The deviations in electronics of the Big Muff alone have inspired volumes of writing. And with JFET overdrive pedals you actually had to sort through large groups of JFET transistors in order to find the right transistors that would share the proper bias values in order to make the pedal work right. Or use trimpots to bias the transistors properly. With SMD JFET transistors that's far less of an issue as they tend to be far more consistent in their values.

 

[lqdsnddist]

Market for high end synths is quite small, and pretty crowded, with most folks using vst software, and those that want hardware tending to prefer real analog.

I used to own gear like the Virus, but have since switched to The likes of UHE Diva, Reaktor etc, or hardware like the Behringer Nuetron and Moog DFAM.

The time and effort that would go into coding a high end virtual synth would be massive, and I don’t know if Cliff even has much knowledge in that area, nor a collection of hardware to compare too etc.

Then even if he did it how much demand would there be for a $2500 rack mount virtual synth ? You’d still need an interface for it as 5 knobs wouldn’t cut it, so you’d have to produce more hardware too

Just sounds like a losing business idea in my book

 

[creativespiral]

The only thing wrong with modern based hardware that uses SMD and other modern electronics is that its qualitatively too good, as the old electronic components of the past were all over the place in quality and deviation. A 5% deviation in a 100K pot means it can be between 95K and 105K. But a 10% deviation puts it between 90K and 110K. Already 20K extra chance of sonic deviation. And I reckon the old stuff of the 60's and 70's would deviate even more. Add to that degradation of the capacitors, especially electrolytics and you can basically have 2 exact same units that sound different. This was why with fuzz face pedals guitarists were actually auditioning various pedals to see which one sounded best. The deviations in electronics of the Big Muff alone have inspired volumes of writing. And with JFET overdrive pedals you actually had to sort through large groups of JFET transistors in order to find the right transistors that would share the proper bias values in order to make the pedal work right. Or use trimpots to bias the transistors properly. With SMD JFET transistors that's far less of an issue as they tend to be far more consistent in their values.

Yeah, this is exactly what I was researching... and what makes "component based software modeling" so much more interesting than traditional means. If you're really designing the software like a virtual pc board, and have ability to adjust each pot/resistor/cap value, you can really dig deep and potentially recreate a wide swath of classic synthesizer models. If you've software modeled a MemoryMoog down to the component level, then you can fine tune it and "virtually age" components, and play with tolerances, to get the sound of a 30-40 year old VCO analog synth. And then a moment later switch over to another component based model of a different synth, ie: Prophet or Juno or Andromeda or something significantly different. This is essentially what Fractal has done in the guitar amps / effects field... and their attention to detail has seriously disrupted the field.

 

[creativespiral]

Market for high end synths is quite small, and pretty crowded, with most folks using vst software, and those that want hardware tending to prefer real analog.

The time and effort that would go into coding a high end virtual synth would be massive, and I don’t know if Cliff even has much knowledge in that area, nor a collection of hardware to compare too etc.

Then even if he did it how much demand would there be for a $2500 rack mount virtual synth ? You’d still need an interface for it as 5 knobs wouldn’t cut it, so you’d have to produce more hardware too

Just sounds like a losing business idea in my book

Yeah, I'm not sure if Cliff has an "interest" in synth programming/modeling... but I bet he would be good at it. This type of project seems to be right in the wheelhouse of what they do. Maybe they would need to hire an additional programmer or two... but I bet a lot of the IP and code blocks from AxeFx could be ported/used in development of such a product.

A $2500 rack mount hardware synth engine could be a huge hit... if they could get the same sort of realism that they have done for guitar amps/effects. Imagine having a rack box that you could hook up any one of your MIDI controller keyboards to, and instantly have access to the sounds of a 1983 MemoryMoog, MiniMoog Voyager, Prophet 5, Roland Jupiter, OB-XA, Andromeda, and 100 other synths ranging from earliest monos to modern day digital polysynths. You would have a black box that potentially replaces hundreds of thousands of dollars of gear. Plus high end Fractal effects included... Tri-chorus, Delays, Reverbs, etc...

Besides Roland, with ACB, there's not many other companies out there pursuing this currently... high end, touring grade, immediate hardware based access to tons synth models, both analog and digital. I bet Fractal could hit it out of the park.

 

[lqdsnddist]

Yeah, this is exactly what I was researching... and what makes "component based software modeling" so much more interesting than traditional means. If you're really designing the software like a virtual pc board, and have ability to adjust each pot/resistor/cap value, you can really dig deep and potentially recreate a wide swath of classic synthesizer models. If you've software modeled a MemoryMoog down to the component level, then you can fine tune it and "virtually age" components, and play with tolerances, to get the sound of a 30-40 year old VCO analog synth. And then a moment later switch over to another component based model of a different synth, ie: Prophet or Juno or Andromeda or something significantly different. This is essentially what Fractal has done in the guitar amps / effects field... and their attention to detail has seriously disrupted the field.

Lots of software emulations already do this, check out UHE for example, or the Legend MiniMoog, component level modeling, drift, age etc all taken into account

UHE Diva also lets you take things like the offer style filter for a Moog and combine it with Different oscillators, envelopes etc, kind of mix and matching different synth circuits.

 

[creativespiral]

Lots of software emulations already do this, check out UHE for example, or the Legend MiniMoog, component level modeling, drift, age etc all taken into account

UHE Diva also lets you take things like the offer style filter for a Moog and combine it with Different oscillators, envelopes etc, kind of mix and matching different synth circuits.

No doubt, there's a lot of good software VSTs out there, and some may use some component modeling... but besides Roland ACB, I don't see other component based hardware offerings. Similar to the guitar field, you've got other competition in software form with Bias, Guitar Rig, Revalver, Amplitube etc... But that hasn't stopped Fractal from doing it better, and offering it in rack hardware form with instant access / no computer required... something that pro players and touring musicians can count on night after night.

If the usage of high end DSP/FPGA architecture, along with the attention to detail that Fractal has shown can be applied to synthesizers, my feeling is that they could be as disruptive to that market as they are to guitar field.

 

[lqdsnddist]

I think the difference is the guitar amp modeling market is much larger than the synth market, and part of that comes down to the nature of guitar amps.

If I want a Plexi it’s not so much the cost or availability, it’s the fact it’s too darn loud for home or many venues. A modeler lets me have cranked amp tones at any volume so it makes practical sense.

If I want a Moog synth, there is nothing holding my back from playing it at home etc, and there are also lots of real analog hardware to choose from like a Moog Sub Phatty for $700 or for $300 a Behringer Model D which sounds amazing for the money

A high end emulation of analog just doesn’t make a lot of sense in my book. In fact, i had a Access Virus Ti, virtual analog which costs me like $2000 something and it sounded good, but why was I spending that kind of money to emulate synths when I could buy the real thing for for way less ?

I don’t do much production any more but have a lot of friends who still do, they have all gotten into modular and semi modular for the most part. Used to have all the Arps and Juno’s etc but that stuff gets a little limited. There aren’t a lot of sounds out of a Juno 106 that haven’t been done to death. Instead people are getting modular which has some unique sounds, and investing in that. I just can’t see a huge market demand for high end virtual modeling.

It’s like a niche of a niche market, and given the resources to enter it, and the fact FAS probably wouldn’t sell 1/10th the number of III’s they do, it seems like it would just take resources away from doing soemthing profitable like making a new Ax8 which will sell out for 6 months

 

[creativespiral]

I think the difference is the guitar amp modeling market is much larger than the synth market, and part of that comes down to the nature of guitar amps.

If I want a Plexi it’s not so much the cost or availability, it’s the fact it’s too darn loud for home or many venues. A modeler lets me have cranked amp tones at any volume so it makes practical sense.

If I want a Moog synth, there is nothing holding my back from playing it at home etc, and there are also lots of real analog hardware to choose from like a Moog Sub Phatty for $700 or for $300 a Behringer Model D which sounds amazing for the money

A high end emulation of analog just doesn’t make a lot of sense in my book. In fact, i had a Access Virus Ti, virtual analog which costs me like $2000 something and it sounded good, but why was I spending that kind of money to emulate synths when I could buy the real thing for for way less ?

I don’t do much production any more but have a lot of friends who still do, they have all gotten into modular and semi modular for the most part. Used to have all the Arps and Juno’s etc but that stuff gets a little limited. There aren’t a lot of sounds out of a Juno 106 that haven’t been done to death. Instead people are getting modular which has some unique sounds, and investing in that. I just can’t see a huge market demand for high end virtual modeling.

It’s like a niche of a niche market, and given the resources to enter it, and the fact FAS probably wouldn’t sell 1/10th the number of III’s they do, it seems like it would just take resources away from doing soemthing profitable like making a new Ax8 which will sell out for 6 months

Maybe the synth market is not as large as the guitar market, but I think you're seriously downplaying its size, and the amount of activity... every year, there are hundreds of new offerings in the synth field, ranging from inexpensive modules and boutique units, up to major industry players releasing expensive poly-synths... there wouldn't be so much activity if the market didn't support it. Synth players are just as much crack-heads as guitar players when it comes to GAS... especially those that get involved in modular synthesis.. those guys are f'n crazy!

And its not just about making a choice of one specific synth vs a high end modeling unit. It's access to dozens or potentially hundreds of classic synths. If you want to buy a Roland Jupiter-8 in good working condition, be prepared to spend $15k... and that's just for one synth. (one of the most sought after). There are many others that are ridiculously high priced and rare (MemoryMoogs, OBs, Yamaha CS80, etc). Plus there are hundreds of other analog synths through the past four decades that all use similar components with different circuit layouts. Once you've accurately modeled a few VCO circuits, those code blocks would be reused in dozens of different models of synths... ie: Curtis CEM3340, SSM2030, Yamaha IG00153, etc...

For touring musicians / club players, it could be a portable, high quality unit that delivers all the sounds they require in a rock solid, dependable rack system. Just add your favorite MIDI controller keyboard(s) and you're good to go... If you could run two synth engines simultaneously (like two amp blocks in the AxeFX) you could even have two separate MIDI keyboard inputs and have one routed to a PolySynth and one routed to a MonoLead type model. It would be a unit that could be tweaked / fine tuned to behave like old Polysynths with unique quirks... and then have that same sound saved and on tap at any time. No fighting with constant synth maintenance or dealing with temperature/humidity issues changing the sound of your instrument from night to night and venue to venue. Also, no dealing with hauling a fragile 220lb Yamaha CS-80, or other old synth gear which can be very delicate.

For studios / producers / engineers, you'd have access to hundreds of thousands of dollars of virtual gear to experiment with and use in recordings. Plus, by porting over effects like Tri-Chorus, Delays, Reverbs, etc, you replace the whole synth chain.

 

[lqdsnddist]

Your ignoring though that there are so many cool new affordable true analogs on the market though that do a great job of “scratching that itch” for a lot of users. They have so many analog mono and poly synths that there isn’t nearly the demand for older models, short of collectors. Why mess with a mc202 or sh101 when you can get a Brute or mini for a few hundred bucks ? Especially if your gigging it....

Also, outside the studio, how many people are doing music that requires proper emulations of old synths live ?

Certainly a lot of synths in pop music, but does that market/fan really care of the synth guy in the background is using a Jupiter string pad or he’s using a modern work station ? I mean heck, does anyone know the name of most of these backing musicians ? Talented of course, but who’s Lady Gaga’s keyboard player ? Does amyone care what he’s playing ?

Essentially how many old school analog synth groups are there these days ? I know the stuff is still used in the studio, but how many people want to spend a few grand on a high end virtual and how is it going to be that significantly different than things like an Access Virus Rack, or Rolands ACM stuff etc ?


Beyond that, does Cliff even play synths ? He’s been a guitar player for decades and has a massive collection of amps, so it’s natural for him to know what a good emulation sounds like, but does Cliff know what an SEM filter should sound like? Does he want to drop $100k buying and repairing old synths to model ? Huge undertaking to put it mildly.

 

[creativespiral]

I'm not ignoring that there's lots of great new analog synths hitting the market... The new Model D, Neutron, SE-02, IK Uno, Minibrute, Monologue, BS2, etc... all great inexpensive kits, but they're all very limited in scope, and may (or may not) faithfully reproduce classic sounds. Further, there's a bunch of great new Analog Poly Synths released in the past few years... Prophet 6, OB6, Rev2, Quantum, Prologue, DM12, Peak, MatrixBrute, JDXA, and of course the new Moog One. But that doesn't make the idea of a Fractal Synth Black Box any less intriguing... in fact, I think it just validates that there is a strong market.

The thing is that even mortgaging your house and purchasing a Moog One ($8k) is going to give you a specific territory of sound. It may capture the classic MemoryMoog, Voyager, MiniMoog sounds well... although, because of modern advancements and perfection of electrical components, it may even lack some of the character that people love about those classics. And even though its $8000, that doesn't mean you're gonna faithfully create the sounds of an Oberheim or Prophet or CS80 territory. You're still gonna want an OB SEM type of synth if you wanna faithfully recreate some classic Rush, Police, or Prince sounds that used an Oberheim... just due to differences in circuitry design and components used for the VCOs, VCFs and VCAs.

It's not that any person NEEDS faithful access to all of the classic synth sounds... but if a unit like a Fractal SynthFX could offer it, I think it would be a huge hit for touring, club musicians, in the studio for recording, and for sound designers / film composers. The same could be said for AxeFX ... I don't NEED access to 260 Amps and 100s of Cabs and Effects... I don't NEED a dozen different variations of Friedman amps, but I love that I have access to them and that I can get authentic tones with such ease. The guitar sounds I play with now, I may have never found without the AxeFX.

I have no idea if Cliff has any interest in synthesizers, or the modeling of... but I bet Fractal would already be half way there in terms of product development - by applying their existing code blocks and IP for component based electronics modeling... also, they already have infrastructure setup for producing high end dsp/fpga audio systems. The entire effects engine for AxeFx could be ported and used for synths. Of course, there would certainly be thousands of hours of additional modeling of synth specific circuitry, and A/B testing and matching.... no doubt, it would be a significant investment of time and resources to be sure... probably a year or two development time. But, they could perhaps hire an additional programmer or two who are synth nuts, and I bet they could pull off something that would be amazing and positively disrupt the synth marketplace.

 

[shatteredsquare]

Behringer just came out with theirs, Yamaha has a new one that does a bunch of modelling, Yamaha did the THR amps already which are spot on accurate, and they are probably working the next gen of component modeled amps right now. FAS could do synths but it would necessitate hiring a lot more elves, probably more than creator would want to deal with. With a DSP platform like exists now, there is no limit to what it could do, all it needs now is a couple oscillator blocks and a few more free LFOs and it could already serve as a modular synth playground

 

[lqdsnddist]

I think what is making the market strong, or at least somewhat resurgent in terms of hardware, is that all these new bits of kit are great sounding and affordable, so its offering analog to a generation or two who never experienced it the first time around. I think the actual frame of reference for many of those older synths has been long forgotten by a lot of users, and never was for a good number of musicians. I'm a middle aged guy who'd been doing music my whole life, I know and love the likes of Rush, but I just don't think there are that many people out there who want to reproduce 70's and 80's synth tones, at least not to the tune of $2500. Plus a lot of these newer synths have what was previously missing on so much rack mount and VSTi gear... a dedicated interface. I've lost count of the knob of controller KB's I've owned, knob boxes etc, and its always a pain, even when they say stuff auto maps, because not everything does.

The whole thing I love about my Behringer Model D is the interface. My Legend and Monark VSTi's sounded equally good, but I was clicking the mouse all the time. With the model D I've got a one knob per function interface, and it just happens to have a great sounding analog synth engine to boot lol.

Fair enough, if your in a Rush, ELO, etc type cover band, and you want authentic tones, and you've got the budget, I could see people buying one if its far and away better than any other virtual analog on the market, but that is still a pretty small market.

I doubt few if any bedroom producers under the age of 40 are really going out and buying a synth because they want the most spot on "Lucky Man" tone possible. I worked a GC for a while, and covered the synth/dj/pro audio dept. and 99% of the people coming in where into EDM and Hip Hop, its what was selling the most, and what ultimately drives the market.

In the end though, its up to Cliff. Its his company. If he wants to table further III updates, AX8 II etc and spend the next 18 months making the best synth on the market, more power to him. Same if he wants to hire a whole new staff for the project etc. Might make him millions, might nearly bankrupt the company.. who's to say, not really either of us lol.

In the end, he seems a pretty smart businessman, so I think he'll take the company in whatever direction he deems best. Be interested to see what the product lineup is like in 2020.....

 

[chris]

this just shows how good the Axe-Fx concept is.

i can't see this ever happening though.

 

[creativespiral]

With a DSP platform like exists now, there is no limit to what it could do, all it needs now is a couple oscillator blocks and a few more free LFOs and it could already serve as a modular synth playground

Yeah, you are right... Fractal has already dipped into the synth arena with the AxeFx, although with limited functionality, focused on guitar tracking... The key for making a flagship rack synth product like this would be to really nail the modeling of those VCO, VCF and VCA circuits that are the heart of every classic analog synth... like Fractal has done with amp tube modeling.

Once you've modeled the Curtis CEM33XX series oscillators, filters and amp chips, you've got the backbone for a huge selection of classic synths from Moog, Sequential/DSI, Oberheim, Roland, and others. Like in the Guitar Amps and Effects field, most manufacturers have used the same electronic components from a few select manufacturers, just arranged in different layouts and with different modulation options.

https://en.wikipedia.org/wiki/CEM_and_SSM_chips_in_synthesizers

I know if Fractal undertook such a project, they would make something amazing... and the AxeFx guitar line would most likely benefit from research in the synth area too. It would require taking a real deep look at VCO and VCF chip variances and the effects of temperature and aging on chips vs their nominal performance characteristics. So much of the interesting, sought after sound of analog synths comes from little imperfections between chips that were meant to be matched in performance. VCOs exhibit unique harmonics in analog waveshaping and interesting detuning performance in the frequency clocks.

 

[creativespiral]

this just shows how good the Axe-Fx concept is.

i can't see this ever happening though.

Yeah, the focus on modeling down to individual electronic components/ICs, cutting edge dsp/fpga architecture, and Fractal's dedication to ongoing innovation have been an eye opener. Truly the best product I've ever purchased!

Going for synth modeling would no doubt be a huge investment in time and resources, and maybe beyond what Fractal is capable of with current personnel. But as I mentioned before, Fractal is probably half way there already with the technology and IP they have developed for guitar modeling.

There are some great synth modelers "in the box" (Omnisphere, Arturia V, NI Synths, UVI, Avenger and others)... some of them do a pretty good job of modeling classic analog synths... but they don't have that same sort of component modeling approach and attention to detail that Fractal does... also, they are developed to be used on a variety of PCs/Macs, so they have to work with constraints in algorithms.. instead of being able to develop a perfect dsp/fpga type platform optimized for high end analog modeling.

The current state of affairs for synth modeling is similar to how the guitar vst and amp modeling offerings were before Fractal came along. There's still some "spice" missing to capture the analog mystique. Fractal is all about that spice!