My next processor will either be an FM3 or Gt1000.
Neural DSP Users?
- Thread starter Masta' C
- Start date
My next processor will either be an FM3 or Gt1000.
On the "AI" thing - it's absolutely marketing but they do it because in 99% of people's heads machine learning (which is what they are doing) and "AI" are the same thing - and it's not really their fault. Companies like Google, Apple, Tesla, and many others have all dubbed machine learning as "Ai" because AI sells basically.
So yes Neural isn't building generalized artificial intelligence, but they have built and trained a very specific and sophisticated ML algorithm to do their capture. While what they are doing exactly will obviously never be talked about, they do talk about it hearing "like a human" and so I suspect they have invested a lot of time into trying to train a model that is aware of (and hence captures) the psychoacoustic anomalies of the human auditory system. Which is really quite interesting. If you've done music and mixing for any length of time you know there are just as many auditory "illusions" as there are visual ones - our brains and ears play tricks on us in certain ways. That's what led to the loudness wars for a while - because "louder" sounded "better" - only it didn't really. It's the basis for a lot of compression technology because frequency masking is a thing and the "loudest" thing at any given moment masks a quieter thing in the same frequency ranges so we can preserve the loud thing, discard info for the quieter thing and our dumb human hearing can't really tell the difference. We are more sensitive to 5-7k (baby cry) because.. we just are, cause babies crying, and so on - we perceive certain frequency balances more or less depending overall volume, etc, etc
So, I believe neural have spent time trying to model THAT. Not to take a flawless, 20-20k perfectly flat "sample" and determine the precise deltas between source and response. I think their capture tries to "hear" the response as a human ear would, with all of its dumb little quirks and illusions. Without that, yes you can capture a perfect "tonal" profile - you can reverse engineer an eq curve and determine how a very specific waveform you fed in gets clipped, you can do Fourier analysis to determine harmonics and overtones... but sometimes there's just a weird... swell, or a "bubble" or a grunt, or thump, or any other esoteric term we try to use to describe the "feel" of an amp. but we KNOW it's there we HEAR it - like a human. And they happen over time, things bloom or diminish over time, so you need to sample and "learn" the response of what you're capturing a certain way to get that. I think Neural - whatever they are doing - has THAT, a little something extra. Something that doesn't just get very close on tone, but also gets more (not all) of the "feel" because it's "hearing" things like people do. It's not a rigid algorithm, it's adaptive, it's "learning" (so yes, misnomered as 'intelligent') and they can keep training it on new things, making it better, and then update the models on all the QCs. It's the adaptivity, I think, that sets their POV on how to capture apart from the rest.
So yes... "AI" is marketing. but it's not fair to discount their approach to the technology of capture as garbage or just more of the same because they use a marketing term they basically MUST use because others before them have abused it so much that it's the only thing Joe Public will understand and respond to when referring to machine learning and adaptive algorithms.
And this isn't meant to sway you, just wanted to clear up the marketing nonsense POV - yes it's hype, but it's also unique and has a very real basis in some good technology.
Disclosure - I work in technology for a silicon valley company and actively build, train, and use ML/AI models for various parts of my day job. And yes, much of what I said is speculation because Neural isn't going to give up the special sauce, but I at least have a basis in understanding how and what I think they are trying to do and I suspect I'm in the ballpark.
So yes Neural isn't building generalized artificial intelligence, but they have built and trained a very specific and sophisticated ML algorithm to do their capture. While what they are doing exactly will obviously never be talked about, they do talk about it hearing "like a human" and so I suspect they have invested a lot of time into trying to train a model that is aware of (and hence captures) the psychoacoustic anomalies of the human auditory system. Which is really quite interesting. If you've done music and mixing for any length of time you know there are just as many auditory "illusions" as there are visual ones - our brains and ears play tricks on us in certain ways. That's what led to the loudness wars for a while - because "louder" sounded "better" - only it didn't really. It's the basis for a lot of compression technology because frequency masking is a thing and the "loudest" thing at any given moment masks a quieter thing in the same frequency ranges so we can preserve the loud thing, discard info for the quieter thing and our dumb human hearing can't really tell the difference. We are more sensitive to 5-7k (baby cry) because.. we just are, cause babies crying, and so on - we perceive certain frequency balances more or less depending overall volume, etc, etc
So, I believe neural have spent time trying to model THAT. Not to take a flawless, 20-20k perfectly flat "sample" and determine the precise deltas between source and response. I think their capture tries to "hear" the response as a human ear would, with all of its dumb little quirks and illusions. Without that, yes you can capture a perfect "tonal" profile - you can reverse engineer an eq curve and determine how a very specific waveform you fed in gets clipped, you can do Fourier analysis to determine harmonics and overtones... but sometimes there's just a weird... swell, or a "bubble" or a grunt, or thump, or any other esoteric term we try to use to describe the "feel" of an amp. but we KNOW it's there we HEAR it - like a human. And they happen over time, things bloom or diminish over time, so you need to sample and "learn" the response of what you're capturing a certain way to get that. I think Neural - whatever they are doing - has THAT, a little something extra. Something that doesn't just get very close on tone, but also gets more (not all) of the "feel" because it's "hearing" things like people do. It's not a rigid algorithm, it's adaptive, it's "learning" (so yes, misnomered as 'intelligent') and they can keep training it on new things, making it better, and then update the models on all the QCs. It's the adaptivity, I think, that sets their POV on how to capture apart from the rest.
So yes... "AI" is marketing. but it's not fair to discount their approach to the technology of capture as garbage or just more of the same because they use a marketing term they basically MUST use because others before them have abused it so much that it's the only thing Joe Public will understand and respond to when referring to machine learning and adaptive algorithms.
And this isn't meant to sway you, just wanted to clear up the marketing nonsense POV - yes it's hype, but it's also unique and has a very real basis in some good technology.
Disclosure - I work in technology for a silicon valley company and actively build, train, and use ML/AI models for various parts of my day job. And yes, much of what I said is speculation because Neural isn't going to give up the special sauce, but I at least have a basis in understanding how and what I think they are trying to do and I suspect I'm in the ballpark.
I am currently getting my Master's in Computer Science (undergrad CS too), am actually taking a KR&R course atm; haven't taken ML yet. That said, I can't think of any reason/benefit to apply ML to performing a capture, which is something that can be done procedurally (Kemper).
Sometimes when you have a hammer, everything looks like a nail. One of their developers may be trained in ML, but that doesn't mean there is any tangible benefit from using it in captures, beyond marketing.
Unless they explain the algorithm (they won't), my gut reaction is that its ALL marketing. There may be elements of Machine Learning in their code, but the problem of profiling an amp doesn't suggest machine learning. Maybe, it will do something unique, like let the user twist all the eq and gain knobs during the profiling process to generate an intelligent profile that encompasses EQ band/sweep, multiple guitars, etc so the profile actually works like the real amp. That would appeal to me.
These auto tuning and profiling solutions have been used in car audio for 20+ years. They may be taking a different (not necessarily more effective) approach to avoid patent/copyright infringement. IDK the legal situation around Kemper, but if Kemper holds patents or copyrights on using auto tune for guitar rigs, if Neural all of a sudden has the same technology it could cause issues (not lawyer), they may be trying to skirt any legal issues (again, not lawyer.)
On TGP today there was a lengthy post by someone who recognized the same thing I did; that all of the demos sound the same. Earlier in this thread I mentioned it sounds like they don't have many models yet, but it may just be the Neural "sound". I will have to download their plugin demos at some point. They sound good but I'm not about to spend $200 on a plugin with a few artist sounds, so better to not even try them. Whole business model looks like took a page from Apple. I'm a PC/Linux/Android guy.
Video embargo lifted today - so I think we'll all see a lot more info dropping soon and a lot less speculation.
I disagree to some extent that I definitely see the use case of an ML approach - that doesn't instantly make me think it's "best", only that I understand the thought process of addressing it as an ML challenge because if you're modeling the complexities of a biological organism (us) and their hearing, then there are opportunities to optimize that in certain ways via a trained model vs a procedural algo. They STILL need to make a good model, but there's nothing inherently "wrong" about tackling the problem or capture in that way.
I disagree to some extent that I definitely see the use case of an ML approach - that doesn't instantly make me think it's "best", only that I understand the thought process of addressing it as an ML challenge because if you're modeling the complexities of a biological organism (us) and their hearing, then there are opportunities to optimize that in certain ways via a trained model vs a procedural algo. They STILL need to make a good model, but there's nothing inherently "wrong" about tackling the problem or capture in that way.
Masta' C
Well-known member
00:00 - Intro
00:35 - Factory Presets
04:47 - Quad Cortex Overview
08:00 - Control Layout & Build Quality
09:33 - Preset/Scene/Stomp Modes
12:04 - Demoing Scene Mode
12:43 - Quad Cortex Dynamic Response
13:33 - It Does More Than Metal!
14:55 - Demoing Stomp Mode
16:07 - Creating a Preset from Scratch
16:17 - The Directory/Librarian
16:55 - Factory Neural Captures
19:04 - Adding FX Blocks to the signal chain
21:07 - Adding Delay and Reverb
22:00 - Adding a Drive Block
23:11 - Assigning Footswitches in Stomp Mode
24:05 - Creating Scenes in Scene Mode
25:16 - Utilising More Than One Row
27:07 - Saving Your Preset , TAGS & Setlists
27:52 - Quad Cortex I/O Capabilities
29:33 - Patching Effects Back Into The Inputs
34:03 - Neural Capturing a Wampler Dual Fusion Pedal
37:23 - Comparing The Capture To The Real Pedal
39:19 - Marshall Silver Jubilee Capture Demo
41:51 - Outro
Masta' C
Well-known member
00:00 - Intro & Unboxing
04:55 - Strat Ambience Preset
06:18 - Major Strat Vibes
11:40 - Clear Lead
13:24 - Jerry Can't Tell
13:50 - Nolly Andy Timmons Preset
14:28 - The Best Preset Ever!
16:23 - Pitched Delays
17:24 - Necromangust
19:13 - Baritone Ambience
20:33 - Grimacemonger
22:12 - Heavy Baritones
23:38 - Summary
I'm not hearing it. Sounds like a Helix. Its in that "realm", so I'm not complaining; I know good music can be made with it, but it would be hard to say its a "slam dunk" from the samples posted.
None of these guys are going to make a negative review; they are in the game of getting free gear and don't want to spook any companies. I like all these reviewers, and I respect their playing, but I also know their careers depend upon not pissing anyone off and getting more subscribers.
I think it will honestly take a year for it to shake out. When Fractal owners buy this and make the comparison, I think we will get a better picture. To my ears, its in the Helix family of tone.
None of these guys are going to make a negative review; they are in the game of getting free gear and don't want to spook any companies. I like all these reviewers, and I respect their playing, but I also know their careers depend upon not pissing anyone off and getting more subscribers.
I think it will honestly take a year for it to shake out. When Fractal owners buy this and make the comparison, I think we will get a better picture. To my ears, its in the Helix family of tone.
Masta' C
Well-known member
That's totally fair. Figured some of the vids might hit on the features you were wondering about. I haven't had time to watch them yet.
Thoughts today...
*It looks like it has a patch/scene switching scheme like the Helix.
*Doesn't have instant patch switching or spillover.
*The UI looks like the Helix.
*Someone on TGP made a good point that this costs 1600, while an FX3 with a footswitch costs 3K. If this has similar processing power to the FX3, it is price competitive on "horsepower". This has the potential to run really expensive reverbs for less than an FX3. The FM3 apparently doesnt have enough processsing to run the expensive verbs with much else.
*Its roughly the size of a GT1000, which is sweet spot form factor. Small enough to stuff in a bag and take with you, yet has a full compliment of switches.
*Seems thin on features and FX.
*No one has tested USB interface performance.
It makes a good case against both the Fractal and the Kemper. It has a nicer interface than the Fractal and wins in price comparison against the FX3. It has a MUCH nicer interface than the Kemper and better FX.
It still costs twice what I can get a Boss GT1000 for and has far fewer features. It doesn't sound "better" to me from demos I've heard, and I dont care about profiling, so I'm still not interested in it. But I understand where they are coming from. They are taking a bite out of Kemper and Fractal sales.
Interesting question, what will Fractal and Kemper do next? Fractal has a 1K box which will continue to win customers on price. In a few years they will need a serious UI redesign. They could build a successor to the AX8, but it would look an awful lot like the QC, minus the pretty interface. I suspect they will pass on that.
I suspect Kemper will ride off into the sunset.
Masta' C
Well-known member
Dang, that's harsh! I think Kemper will come up with something once they start to feel the squeeze. Besides, many musicians are brand-loyal and "Kemper" is the OG profiler. Watch...that green paint and monochromatic LCD display will become as "classic" and desirable as a 50's Tele!
:nana:
I think that kinda nails it - it kinda hits a few sweet spots on price/features/etc. It may not be the best at any one thing right now (again we're comparing a 1.0 of a totally new product - which isn't fair to compare even against a 1.0 of the FX3 or similar for example as those are still building upon the versions that came before them) but I do believe it will get better, the hardware itself seems very capable and I imagine things like gapless switching (which depending on your needs scene mode might work just fine for, or clever use of a split between multiple rigs running in a single preset as I've suggested a few times on the neural discord) will improve. Everything I've seen says it WILL do FX tails but I just don't think we've seen it demo'd yet.
And I think it is a bit thin, especially on FX at launch, but again, I don't think, and I'm not sure why anyone else would, that they're' "done" that what it ships with is all you're going to get - far from it. Now if you absolutely NEED a specific reverb from moment one - then yeah, wait (or buy a pedal for it and put it in the FX loop). But I think the people realizing it's really a nexus of a few things, better than X at Y and better than A at B (but perhaps not better than EVERYTHING at EVERYTHING) while being priced kinda in the middle of the pack is exactly the niche I think Neural was trying to carve out here. So far I think they're pretty close to hitting it. More videos coming and hopefully a clearer picture will emerge as soon as "regular" people get their hands on it. I definitely plan to test it as an interface (8in/8out) and see how it holds up.
I think there's drama and religion around any new piece of gear in music, everyone has a side and their beliefs and holy wars will be fought. It's half the fun, but it can also be awful. I think Neural has a solid platform (hence why I bought in) and I truly believe they're trying to make the "best" product - maybe that means the best THEY can make, maybe at some point the best in the industry (for a brief time anyway till the next thing gets released), but it's not a "bad" product, and I genuinely think the price is fair. When you consider a good high-end tube head can go for 2-4k and if you really want THAT sound it's worth it, but this is less than that and will at least get you close (and maybe very close at least in a blind test) but also get you much more capability beyond that one sound, I think it's fair.
All fair points, I hope you enjoy it.
Reading the TGP thread, the only thing that stood out was the videos of the aliasing performance, apparently the QC isn't great at it, which afaict, means there is unaccounted background noise; a side effect of trying to model analog signal in digital realm. Apparently Neural plugins have had some issues, but this may be fixed in software updates, increased sampling rate, etc.
I don't fully understand it (haven't studied it specifically), but that is a technical area that separates various MFX processors.
When I tried the Helix Native, I was very disappointed in the character of the distortion, there was something unmusical about it, something wrong with the top end that I didn't perceive until I had played it myself. That might have been aliasing?? Then there are all the reports about squirrels, which tells me modeling with perfect fidelity is not a trivial problem. Maybe on something like the Helix, they are CPU bound and have to make compromises with sample rate, which 97% of the users won't hear. But something like the QC should have enough power to process at a much higher sample rate, or at least have that capability in the future.
That said, I really can't judge the QC character until I have played it myself. Initial listening to youtube demos were telling my ears there was some similarity with Helix, I am hearing some kind of hashing, some kind of blur, that *might* be the same thing I disliked about the Helix, which might be aliasing. Or it could be youtube compression algorithm and have nothing to do with the QC.
IOW, I think its possible for a modeler to make mistakes or compromises on the math, that might make it sound inauthentic to some users. I don't think the QC is a slam dunk. Its not clear that it is better sounding, and the feature list is currently thin. In fact it might take years of tweaking and development to get it equivalent to some of the others. Verdict is still out. If you buy it now, you are entering into an extended waiting game of "when will they add ______?" It doesnt have a computer interface/patch manager atm, which tells me this is a hard pass, at least until they have that.
Last edited:
I've been following the TGP discussion on the QC, but don't have an account over there. The conversation has taken a negative turn; it seems that Neural may intend to gate access to profiles. There are assurances that it will remain open (in that users can free trade profiles through their cloud), but it seems they are setting it up as a store front, so people/companies can sell profiles.
This reminds me of something I read a couple years ago about the Kemper, that there were legal ramifications of "copying" an analog device and selling a digital copy of it. At a minimum there would be trademark and/or potential copyright issues.
How I see this playing out is that Neural has anticipated the legal side of the profiling business, and will have a store front already set up when courts decide that profiles of specific amps are violating trademarks and/or copyrights. They will have the ability to license and sell digital "copies" of physical properties. When this happens the message to users will be, "so sorry, this is totally out of our control... but we cant allow this profile to stay on our cloud service because it violates copyrights/trademarks." Of course they will have liscencing agreements worked out with the amp builders, and "will be there for you", when the shit hits the fan. Which I suspect, they know it will.
IMO, the whole thing stands to be very costly, depending on how certain legal battles play out. Its OK for someone to build a model that "sounds like" a Fender or a Marshall, and give it a clever name, but when its an "actual" profile of the "actual" amp, giving it a clever name won't be enough to avoid copyrights.
The interesting thing I wanted to point out, is if the profiling world takes this legal turn, Neural will be positioned to benefit from it. In fact, they are probably pushing for this outcome.
Still not a fan of the whole profiling schtick. All it shows is that a computer can match EQ and choose gain staging so that the profile matches what it was listening to through the mic. You can do all this by ear with a good modeler and in the end it is more versatile that a folder full of profiles.
This reminds me of something I read a couple years ago about the Kemper, that there were legal ramifications of "copying" an analog device and selling a digital copy of it. At a minimum there would be trademark and/or potential copyright issues.
How I see this playing out is that Neural has anticipated the legal side of the profiling business, and will have a store front already set up when courts decide that profiles of specific amps are violating trademarks and/or copyrights. They will have the ability to license and sell digital "copies" of physical properties. When this happens the message to users will be, "so sorry, this is totally out of our control... but we cant allow this profile to stay on our cloud service because it violates copyrights/trademarks." Of course they will have liscencing agreements worked out with the amp builders, and "will be there for you", when the shit hits the fan. Which I suspect, they know it will.
IMO, the whole thing stands to be very costly, depending on how certain legal battles play out. Its OK for someone to build a model that "sounds like" a Fender or a Marshall, and give it a clever name, but when its an "actual" profile of the "actual" amp, giving it a clever name won't be enough to avoid copyrights.
The interesting thing I wanted to point out, is if the profiling world takes this legal turn, Neural will be positioned to benefit from it. In fact, they are probably pushing for this outcome.
Still not a fan of the whole profiling schtick. All it shows is that a computer can match EQ and choose gain staging so that the profile matches what it was listening to through the mic. You can do all this by ear with a good modeler and in the end it is more versatile that a folder full of profiles.
Masta' C
Well-known member
Completely disagree.
First, modeling technology has been trying to do just this since the beginning, including Fender's own modeling technology. Advanced iterations of modeling software have literally tried to recreate entire (patented) circuits digitally, right down to the board-level components and have gotten away with it.
Additionally, all recorded music is essentially a capture of the specific character and sound of real life products. Yet Fender doesn't have the right to my music simply because I used their amp to create my own sound "profile" and record it digitally.
There's plenty of precedence to keep lawsuits at bay, such as the longstanding Kemper technology, but also plenty of other reasons ($$$) why licensing arrangements could be a valuable tool for Neural in the future.
Last edited:
Its not a matter of disagreeing. It hasn't been tested in court. What I read is that it probably will be, and then the whole landscape will change.
If I am selling a trinket, and you make a mold of that trinket and recast it, is that legal? What if it was made with brass, and you recast it in tin. Even through it was made with different material, it would still be a copy.
I'm not just playing devils advocate, there was a discussion about this regarding Kemper.
With the QC, because they have control over the cloud distribution, it would actually benefit them, and badly hurt Kemper if a legal decision decided profiles violated copyright. Keep your eye on this over the next couple years. It will play out in the courts. Until then we are just guessing the outcome, but its clear to me that Neural would benefit.
Masta' C
Well-known member
I see Neural benefiting either way from the model they are proposing, just possibly for a different reason than you imply.
Your assumptions about infringement haven't played out in court, either. Kemper, Line6, and other modelers have been at the same game for nearly 2 decades now.
Also, your "making a mold and recasting it in a new material" analogy isn't accurate. They aren't buying a Fender amp and creating a physical clone just with a different label on the front or a different color of tolex on the cab (cough::Bugera::cough). They are modeling a non-physical, digital audio profile after a sonic footprint produced by a physical product.
I suppose a better way to think about it is this...if I take a photo with my cell phone of a Fender amp, is it infringement? It's obviously a "snapshot" of the real product, but it's fully digital and nothing like the product itself, which only exists in physical form. Thus, Fender can't come along and claim that my picture of their product is equivalent to the product itself, because it's an entirely different medium.
The other analogy I already used is music...Fender doesn't own all music made with their amps simply because the audio effectively "captures" what their amp sounds like, even when this audio gets shared/sold for profit.
The fact that Fender, Marshall, Peavey, etc. have also dabbled with modeling of competitors products in their own products strongly weakens their case.
I guess what I'm saying is, you can assume all you want, but there is far more precedence to suggest there's little that amp manufacturers can do to stop this direction of modeling/capture than to fight it. Not saying lawsuits won't arise, because this emerging tech is definitely a threat to the big companies, but I think Fender will end up approaching Neural to willingly offer a "licensed" suite of amp profiles before they ever win in a legal battle.
Your assumptions about infringement haven't played out in court, either. Kemper, Line6, and other modelers have been at the same game for nearly 2 decades now.
Also, your "making a mold and recasting it in a new material" analogy isn't accurate. They aren't buying a Fender amp and creating a physical clone just with a different label on the front or a different color of tolex on the cab (cough::Bugera::cough). They are modeling a non-physical, digital audio profile after a sonic footprint produced by a physical product.
I suppose a better way to think about it is this...if I take a photo with my cell phone of a Fender amp, is it infringement? It's obviously a "snapshot" of the real product, but it's fully digital and nothing like the product itself, which only exists in physical form. Thus, Fender can't come along and claim that my picture of their product is equivalent to the product itself, because it's an entirely different medium.
The other analogy I already used is music...Fender doesn't own all music made with their amps simply because the audio effectively "captures" what their amp sounds like, even when this audio gets shared/sold for profit.
The fact that Fender, Marshall, Peavey, etc. have also dabbled with modeling of competitors products in their own products strongly weakens their case.
I guess what I'm saying is, you can assume all you want, but there is far more precedence to suggest there's little that amp manufacturers can do to stop this direction of modeling/capture than to fight it. Not saying lawsuits won't arise, because this emerging tech is definitely a threat to the big companies, but I think Fender will end up approaching Neural to willingly offer a "licensed" suite of amp profiles before they ever win in a legal battle.
I liked my analogy, but this is an argument for lawyers.
You're right there may be liscenced packs before direct challenges. But that will bolster later challenges. They will assert they have been selling licensed packs of ____ amp for ____ years and then ask, how can it be legal for other people to give them away? The only people who wont be on board that line of thinking will be Kemper. It will be Kemper vs everyone else. I hope you can see how this will play out.
The discussion on TGP is a dumpster fire - "Cloud Gate" - and everything else is people seriously going off the rails making massive speculations about things for no reason and more than half the people doing it seem to be the ones who are "out" who have decided NOT to buy, etc etc, and yet NEVER stop shitposting in the thread - It's not an airport, you don't need to announce your departure, just GO, thanks. But they never go. Whatever else, I've gained a lot of respect for Doug for dealing with the disaster over there and remaining pretty calm and neutral.
