So with all them V2 capture/profile updates…

Ed DeGenaro

Ed DeGenaro

Shredder
Messages
1,991
So Tonex, NDSP, upcoming Kemper etc…

I can’t shake the feeling that the timing of them all coming up with the new improved profiling that needs a computer/online acccess has more in common than not.

If I was a conspiracy kinda guy I’d say they all incorporated NAM and selling it as their new improved tech.

Thoughts?
 
While NAM may have been the spark to improve their own machine learning, they don't need to use NAM under the hood at all. NAM itself isn't that complex code, nothing they couldn't replicate themselves using whatever neural network models each company uses.

More likely reasons for this simultaneous timing is that each company has been working on the same thing possibly even using the same machine learning networks, Amazon Web Services cloud training and such.

Kemper, NDSP and Tonex are all competing to have people use their specific closed ecosystem to sell you hardware and software products. For Fractal NAM makes more sense because it's more like "Ugh, you want captures, here's NAM support!" type thing.
 
Ed DeGenaro said:
they all incorporated NAM and selling it as their new improved tech.

Thoughts?
Click to expand...

Definitely not. I think it is a combination of competition driving everyone to improve, and catch up to NAM in sound quality, plus advancements in the tools available.

We don't know much about Proxy yet, but Tonex, NDSP, and Kemper couldn't be incorporating NAM and running on the hardware they do as well as they do. They are all way more efficient run time, so while they may or may not have learned some tricks about training from the NAM community, the underlying technology is still very different.
 
Is the underlying technology all that different though? Aren’t they all running test signals through a rig, capturing the audio, loading the results to cloud computation, and then getting inputs back to feed the hardware modeler? They may not be running NAM code but seems they are essentially doing the same thing.
 
Jarick said:
Is the underlying technology all that different though? Aren’t they all running test signals through a rig, capturing the audio, loading the results to cloud computation, and then getting inputs back to feed the hardware modeler? They may not be running NAM code but seems they are essentially doing the same thing.
Click to expand...

Can’t really say if the underlying code is the same, because only NAM is open source.
 
Jarick said:
Is the underlying technology all that different though? Aren’t they all running test signals through a rig, capturing the audio, loading the results to cloud computation, and then getting inputs back to feed the hardware modeler? They may not be running NAM code but seems they are essentially doing the same thing.
Click to expand...

Tonex doesn’t use cloud training. The others do, but that is only on the training side. The big difference is what the computation, cloud or otherwise, produces and then how the resulting output file is then used to create the amp sim.

We don’t know much about Kemper v2 but Kemper v1 had a very different approach than any of the others and it runs on ancient DSP hardware. Tonex v2 runs on a Tonex One or Tonex Plug which suggests IK has a much more efficient playback engine as well and I am pretty sure NDSP has a more efficient playback system than NAM also. I think they all are and have been ahead of NAM on that side. Kemper and NDSP also have been way ahead of NAM (and Tonex) with ease of capture process.

While there is a lot of overlap in the concepts, I think there is significant differences in their underlying technology, the capture files, and the playback systems.
 
Chocol8 said:
Tonex doesn’t use cloud training. The others do, but that is only on the training side. The big difference is what the computation, cloud or otherwise, produces and then how the resulting output file is then used to create the amp sim.

We don’t know much about Kemper v2 but Kemper v1 had a very different approach than any of the others and it runs on ancient DSP hardware. Tonex v2 runs on a Tonex One or Tonex Plug which suggests IK has a much more efficient playback engine as well and I am pretty sure NDSP has a more efficient playback system than NAM also. I think they all are and have been ahead of NAM on that side. Kemper and NDSP also have been way ahead of NAM (and Tonex) with ease of capture process.

While there is a lot of overlap in the concepts, I think there is significant differences in their underlying technology, the capture files, and the playback systems.
Click to expand...

Yeah as cool as NAM is, it seems to be a bit of a pig when it comes to playback.
 
Ed DeGenaro said:
So Tonex, NDSP, upcoming Kemper etc…

I can’t shake the feeling that the timing of them all coming up with the new improved profiling that needs a computer/online acccess has more in common than not.

If I was a conspiracy kinda guy I’d say they all incorporated NAM and selling it as their new improved tech.

Thoughts?
Click to expand...

I was speculating about this in the Neural thread.

If NAM is producing the most accurate results… and it’s open source… id bet they’ve all taken more than a passing glance at the code/process.

I’d bet some portion of them are either retrofitting it or at least using it as the launching point for their service. (And I doubt any of them would be overly inclined to openly admit it, as to waive the the fairy dust over their process)

But I guess you could argue that’s not a bad thing if everyone is offering a high quality capture service. It will ultimately just become a commodity, not a differentiator.
 
Ed DeGenaro said:
So Tonex, NDSP, upcoming Kemper etc…

I can’t shake the feeling that the timing of them all coming up with the new improved profiling that needs a computer/online acccess has more in common than not.

If I was a conspiracy kinda guy I’d say they all incorporated NAM and selling it as their new improved tech.

Thoughts?
Click to expand...
Kemper NAM? That's a hard no. All they can do is make minor tweaks to their algorithm at this point, due to the processor they are using. Implementing an inference engine would be the remotest of possibilities.
 
Chocol8 said:
Tonex v2 runs on a Tonex One or Tonex Plug which suggests IK has a much more efficient playback engine
Click to expand...

I think it’s mind-boggling that I can run fantastic Tonex v2 captures on a simple Tonex One, while Fractal has not been able to make the Axe-FX III run NAM captures (confirmed by Cliff). I mean, Tonex v2 and NAM are pretty similar in my experience.

As much as I love component modeling and Fractal's approach to almost everything, I would love to see NAM player functionality added across the board on Fractal products (including the AM4).
 
Richard_G said:
I mean, Tonex v2 and NAM are pretty similar in my experience.
Click to expand...

Similar in results but VERY different in how resource intensive they are. That's been the real genius behind Tonex, the ability to run on affordable hardware. Even when it was noticeably not as good as NAM in sound quality, the hardware drove the popularity. Now it is so close, my interest in NAM has been highly reduced.
 
2dor said:
I'm just here to say that I love that we now have good profiling tech from most major players & I'm happy that NAM came along to put a bit of fire under some a$$e$.

Everybody wins!
Click to expand...
Exactly what I was thinking this week.
Fair play to kemper for pioneering it, but we're at the point now where capturing will be in Kemper, Fractal, Line6, NDSP, ToneX, Darkglass, Dimehead, Knock Off NAM units, other random profiling solutions.

It's pretty much on its way to being commoditised which is awesome.
Kemper, QC and ToneX v1 always had its quirks in capture quality. ToneX and QC are in the ballpark now where any improvement would be a nice to have, the quality is as good as you'd expect now. NAM amp fidelity was always there and any quality improvement is 1-2% moves. Kemper..... who knows, see what v2 is like when it launches.

TMP is the only modeller I can think of with no capturing on the horizon (AFAIA).

Crazy how its come from such a unique concept to pretty much being everywhere and in such high quality. I want to see more strides in modelling but that could be an endless quest. It's great to have high tier capturing alongside it while we all wait.
 
Chocol8 said:
Similar in results but VERY different in how resource intensive they are. That's been the real genius behind Tonex, the ability to run on affordable hardware. Even when it was noticeably not as good as NAM in sound quality, the hardware drove the popularity. Now it is so close, my interest in NAM has been highly reduced.
Click to expand...

The only thing I wish IK would do is to add cloud based processing. On my Macbook Air, training just takes so long it's driving me mad.
 
This is why I always thought it was not good that NAM became the standard capturing solution for any modeler.

NAM was huge. It´s still the more accurate static capturing tech based on null tests (whatever that means, you know). It´s also been a shake in the capturing world, and companies reacted by trying to develop their on tech. And that has proved to be good for the progress of capturing. The fact that companies want to have the best system is good for development and improvement.

ToneX improved. NDSP improved vastly (up to the point that now they broke the "time based capturing" barrier with their compression capturing capabilities). Kemper is supposed to improve too (still to be seen). For me, that´s why NAM shouldn´t be THE SOLUTION for all modelers. I prefer companies to struggle developing improvments. And that´s why I think Fractal decision of just adopting NAM is not goping to be enough to compete with what´s coming in next years (capturing being able to manage time based effects, fuzzes, tonestacks, tweaks and whatnot).
 
I don't think you can credit NAM too much. All of the platforms received major feedback on their profiling accuracy way before NAM came on the scene.

Most of these companies know exactly where their tech is flawed. They just won't talk about it publicly.
 
Orvillain said:
I don't think you can credit NAM too much. All of the platforms received major feedback on their profiling accuracy way before NAM came on the scene.

Most of these companies know exactly where their tech is flawed. They just won't talk about it publicly.
Click to expand...
I'm not saying it's entirely responsible for where we are today but it was a catalyst to the developments which came after its launch.
 
2dor said:
I'm not saying it's entirely responsible for where we are today but it was a catalyst to the developments which came after its launch.
Click to expand...
THAT'S EXACTLY WHAT YOU SAID! I'M GOING TO CHOOSE TO TAKE YOU SOOOOO LITERALLY THAT MY BACK BREAKS WHEN I BEND OVER TO TRY AND KISS MY OWN ASS! I'M ON A FUCKING TOUR BUSSSSSS!! TOURRRRRR BUSSSSSSSSSS!!

... but yeah, I get you.
 
You must log in or register to reply here.
Back
Top