Quad Cortex plugin support officially goes from "soon" to "eventually"

This is actually true and a registered trademark.

Neural DSP thought they can pull a complete "Uli" on Line 6... but lady karma is a real b*tch.

No, it can't be. Because that would mean that i no longer live in reality, but in a parody universe where everyone wears clown masks.

Seriously? "Native"?! It's like they don't give a shit anymore.
 
Last edited:
This is actually true and a registered trademark.

United States Patent and Trademark Office website:

Neural DSP thought they can pull a complete "Uli" on Line 6... but lady karma is a real b*tch.
I really hope they don't intend to use that as the name for the editor software. That would be way too on-the-nose Helix ripoff. I mean it's not like Helix Native is a good, descriptive name for a VST plugin in the first place.

I hope they've just trademarked it so someone else doesn't.
 
I would guess it tweaks a pre-made preamp, think of it as advanced automatic tone matching.

Very likely.

I believe this may be correct, as I recall hearing that the Kemper has something like 5 "reference" models, and during the capture process it works out which one is closest to the test tones and uses this as the basis for the profile.

Yeah. There's once been a video/photo (can't find it right now), showing the Kemper while being updated, and it apparently said something like "updating amp models" or so on the screen for a short while.
Also serves as a good explanation why some things simply can't be profiled - in case the "structure" is too far away from whatever baseline amps, things will simply not work well anymore.
And it'd also explain why the Kemper is so quick doing profiles, especially given it's over a decade old.
 
The thing is.... if that were true.... why haven't Kemper consistently been adding different amp model topologies to fill in the gaps.... or do they really think their product is done and dusted?? I have no idea, but what I can say is, it isn't accurate. I tend to think they're at the upper end of their computation limitations, which is why no amp profiling improvements have been seen since the original fixes to the aliasing and the low-end "accuracy" (which to my mind, just moved the goalposts and didn't actually fix anything)
 
The thing is.... if that were true.... why haven't Kemper consistently been adding different amp model topologies to fill in the gaps

Apparently they did - to a certain extent (that's possibly even why there was that "updating amps" screenie thing). And from all I remember, the "profiling compatibility" has been improved during the first years.

or do they really think their product is done and dusted?

Either that or they're now not able to push the technical boundaries any further. If Kempers profiling was indeed based on single amp models being match-EQ-ed (maybe with an additional "in between match point" or so...), that could only take you this far.
Seriously, their profiling is incredibly fast, the .kpr files are basically just a few lines of text (hence "playback instructions", if you will), so I can perfectly understand that there's not much room for improvements anymore, because of...
I tend to think they're at the upper end of their computation limitations,

...this.

To me, while it's not accurate in many situations, it's still an incredible masterpiece of technology (and I'm still thinking of finally buying one, at least one day, as I happen to dig the design and the easy on-unit operation a lot).
 
The thing is.... if that were true.... why haven't Kemper consistently been adding different amp model topologies to fill in the gaps.... or do they really think their product is done and dusted?? I have no idea, but what I can say is, it isn't accurate. I tend to think they're at the upper end of their computation limitations, which is why no amp profiling improvements have been seen since the original fixes to the aliasing and the low-end "accuracy" (which to my mind, just moved the goalposts and didn't actually fix anything)
It's also possible they fear breaking any existing profiles and have people suddenly complain how their favorite profile sounds worse now. You can look at Line6 or Fractal and people always complain after any major modeling update how they need to redo presets. Now apply that same thing to having to redo a lot of profiles and it's far more of a problem because a lot of users don't make those profiles themselves but rely on 3rd party providers.

One way around it is versioning of whatever they are doing under the hood, so profile A is made with amp model X version 1.0 and they add X version 2.0 in an update. Then profile B will always use amp X v2.0 while profile A will always be matched to 1.0. All you will notice is that profiles made today will somehow end up sounding better.

But eventually that can start to eat up a lot of storage space (which is quite limited on a unit 10+ years old) as those updates pile up. For example Fractal got rid of their firmware backward compatibility feature because they needed that space for more firmware features. Maybe Kemper is also running into this sort of problem where there's just not much firmware space for anything.
 
I'll say I just fired up ToneX in 4cm with the Stomp and went through a few HX drives and it kicks some serious booty. You need to find that cap that trips your trigger then never touch the software again :rofl
So I stared at this rig for about 45 seconds and realized the "cap" that trips my trigger (as the cool kids say...) is hardwired into my power amp: the Quilter Lead channel. :)

Crazy thought for the day: cash in ToneX + HX Stomp and finally see what FM3 is all about. @Gearzilla approved?

(You'll all be confounded to hear that what I really want is a second QC. But I can't justify $1850 just to avoid tearing out a bunch of cables every so often.)
 
Seriously? "Native"?! It's like they don't give a s**t anymore.
But let's not pretend Line 6 was forging new ground with the term "Native", either.

Logo-Native-Instruments.jpg
 
They should rather do a SC for 500 then. Mono I/O would be fine for me. Kinda like a Tonex on fire.
Might be best to avoid the phrase, "on fire".

"On steroids"? No that'll just lead to jokes about NDSP's newly installed gym. :rofl
 
To be fair, that’s a company name and while I can certainly understand the implication, there’s a bit of a difference between a company name and two products made by two different companies that do, seemingly, the exact same thing.
I don't entirely disagree, but product/ company names aside, "native" is a long-established term with an established meaning. It's almost like calling them out for calling their inputs "Input" or whatever. Surely we have bigger fish to fry...
 
ToneX and QC snaps are literally new versions of the thing that Christoph Kemper invented.

It's better for us that there are alternatives to Kemper. I'm all for it. And if there's anything legal or not to say about it, let them work that out. But this isn't a grey area like the beginning of modeling. If anyone else was even thinking about something like what Christoph Kemper came up with, they weren't talking about it, at all. Whatever, he may be relieved that after all this time other manufacturers are taking on the task of adopting his tech and responding to the public in ways that may not interest him.


Sometimes I think your undying loyalty to that device is just meant to spite the rest of us. And then I look in the mirror, after ten years with the KPA. Wait, are these digital forums what aged me?
The Kemper invention is adjusting a number of parameters of an underlying amp model until the error signal is a minimum. You feed the output of the underlying model and the output of the DUT (Device Under Test) to a difference block and generate an error signal. You then adjust various parameters to minimize the error. Presumably you would use techniques like gradient descent. The hard part is determining what to measure to generate the error and devising test signals to measure that.

Machine Learning is not an infringement of that patent. It trains a neural network which is, by definition, not parametric.

Now, whether a certain device actually uses AI/ML/NN is something only the developers of said device know. However some interesting data points are available:
- It takes about 20 minutes, for example, to train ToneX using a modern GPU.
- It takes over three hours to train it using a modern CPU.
- Training time for NAM is "a couple hours" using a modern GPU.

Now the device being discussed is somehow able to train a neural network in a minute or two using a low-power DSP with no vector, AI or multi-media instructions. If you were to use this DSP to train a ToneX or NAM model, assuming it even has enough addressable memory, I would estimate training times on a scale of days. Interestingly the test tones generated by said device are remarkably similar to those used by the Kemper.
 
Back
Top