NAM: Neural Amp Modeler

[James Freeman]

The sooner people apply this, the sooner we'll have some consistency between captures.
The numbers cannot be more simple: -10dBFS sine tone in your DAW = 1vAC RMS on the 1/4" output of you Reamp box.
That's for 12.22 dBu inputs, which is within ±1dB of the most common audio interfaces.

 

[Gearzilla]

After all you nerds get things straightened out let me know when it’s ready.

Thank you

 

[Sascha Franck]

What your point?
Go to Amazon.com and buy a f*****g multimeter.

I wouldn't have issues without a multimeter because I've got what you could call a proper set of ears.

 

[MirrorProfiles]

posed the question to the facebook NAMimals. I think (as expected), its a thing that the majority either don't care about or don't understand. There are some that agree about the positives, and are open to it, and some that seem to be opposed (but only because they are misunderstanding why it needs to be done in the first place).

It really needs some kind of youtube video demonstrating why its important, just too many people saying things like "yeah but what about reamp boxes all being different?" or "it already has input and output controls on the plugin, can't you just use those?" or "yeah, but then all the pickup levels will be averaged out".

 

[James Freeman]

I think it's best to lead by example.
Professional/quality captures should be standardize to the most common audio interfaces high-z instrument inputs where people be using NAM the most.
Once people learn that these quality captures match the real amp when used with Focusrite/Audient/UAD interfaces with gain at minimum, everyone will start doing the same.

 

[MirrorProfiles]

I think it's best to lead by example.
Professional/quality captures should be standardize to the most common audio interfaces high-z instrument inputs where people be using NAM the most.
Once people learn that these quality captures match the real amp when used with Focusrite/Audient/UAD interfaces with gain at minimum, everyone will start doing the same.

Yeah definitely think if a few of the heavy hitters establish some kind of a standard, then it would be mad to do something different.

I haven't fully dived into using NAM yet - mostly because of the lack of AAX support, but also the fact that it only supports 48kHz. Its also too risky for me to rely on plugins that are less stable, and likely to change in behaviour as time goes on.

I really love the tech and the direction its going in, but I'd actually rather it was incorporated into a paid product with fixed criteria, proper support, guaranteed stability etc. They are all qualities that paid products should have, and to do properly, really requires dedicated work.

FWIW, besides the odd freebie, there's no way I could make all my captures for free. It's just way too much time and effort to do all that to a good enough quality that I'd be happy. For some people, its not a huge ask to put some time aside to do and I totally understand that others are happy to do all kinds of stuff for free. I would too maybe if it was 10/15 years ago.

 

[James Freeman]

High-Z Instrument Inputs of various audio interfaces (at minimum gain) form their manuals:

Focusrite Scarlett: +12.5dBu
Audient iD: +12dBu
UAD Apollo Twin X: +12.2dBu
SSL2: +15dBu
Helix: +11.5dBu
Arturia MiniFuse: +11.5dBu
Motu M4/6: +16dBu


If the capture was made for that ~12dBu range, it should sound accurate withing ±0.5dB of gain difference for most audio interfaces.
Audio interfaces that deviate from that number too much can be easily compensated, for example SSL2 should be boosted by +3dB to match a 12dBu capture.
You see, if we start with a standardized number, we can easily match any audio interface just by reading the manual.

Think of it like so, when you plug your guitar into your audio interface with NAM input at 0.0dB, the capture should have the same amount of gain like the actual amp that was captured or very close to it, of course you can tweak to taste from there.

@MirrorProfiles, you can post that on FB.

 

[the-trooper]

reading the manual

Unfortunately, you're asking way too much from people.

 

[MirrorProfiles]

High-Z Instrument Inputs of various audio interfaces (at minimum gain) form their manuals:

Focusrite Scarlett: +12.5dBu
Audient iD: +12dBu
UAD Apollo Twin X: +12.2dBu
SSL2: +15dBu
Helix: +11.5dBu
Arturia MiniFuse: +11.5dBu
Motu M4/6: +16dBu


If the capture was made for that ~12dBu range, it should sound accurate withing ±0.5dB of gain difference for most audio interfaces.
Audio interfaces that deviate from that number too much can be easily compensated, for example SSL2 should be boosted by +3dB to match a 12dBu capture.
You see, if we start with a standardized number, we can easily match any audio interface just by reading the manual.

Think of it like so, when you plug your guitar into your audio interface with NAM input at 0.0dB, the capture should have the same amount of gain like the actual amp that was captured or very close to it, of course you can tweak to taste from there.

@MirrorProfiles, you can post that on FB.

Have shared that info, as well as a few bits you've posted elsewhere. Feel like this info will probably get lost eventually. Ill try and knock up a .pdf and I can speak to Steve and Dom (the mod of that group) about hopefully making it a sticky.

It'll really need the main creators to adopt it too, and hopefully bore everyone to tears with the importance of it. But I can work on that too....

 

[usa423]

Some NAM models seem to be a decent amount of gain with the input at zero. I tried to make one last night, and to me it only sounds "right" if I turn the input gain all the way up inside NAM. What did I do wrong? Should I have normalized the reamped track before export?

 

[northern_fox]

Have shared that info, as well as a few bits you've posted elsewhere. Feel like this info will probably get lost eventually. Ill try and knock up a .pdf and I can speak to Steve and Dom (the mod of that group) about hopefully making it a sticky.

It'll really need the main creators to adopt it too, and hopefully bore everyone to tears with the importance of it. But I can work on that too....

I'm the other mod there and also working on an FAQ right now since there are so many new people, so it could also be a good spot to add it. If you also want to send it my way I can pass it along to Steve

Some NAM models seem to be a decent amount of gain with the input at zero. I tried to make one last night, and to me it only sounds "right" if I turn the input gain all the way up inside NAM. What did I do wrong? Should I have normalized the reamped track before export?

They key is to make sure you're reamping with the right level of gain coming from the interface to the amp via a reamp box (hence the above discussion). The best way is to actually measure that what you're sending into the amp is the right level via multimeter or level measurement. You could also do this more crudely the way tonex suggests by ear by plugging into the amp and then the interface->amp to make sure the gain is correct coming through the interface chain.

My process has been to measure the voltage of my reamp box to and then verify by ear that it sounds right. And ironically, this does not seem to get tonex to be correct - there's something weird going on there. But with NAM everything I capture is exactly where it should be with respect to gain levels.

Normalizing the track before export will only really change the final capture volume, but I normalize all of mine to -3 dB true peak for consistency. The normalization I mentioned earlier is just for plugin playback volume between captures, and has nothing to do with actually capturing

 

[usa423]

I'm the other mod there and also working on an FAQ right now since there are so many new people, so it could also be a good spot to add it. If you also want to send it my way I can pass it along to Steve



They key is to make sure you're reamping with the right level of gain coming from the interface to the amp via a reamp box (hence the above discussion). The best way is to actually measure that what you're sending into the amp is the right level via multimeter or level measurement. You could also do this more crudely the way tonex suggests by ear by plugging into the amp and then the interface->amp to make sure the gain is correct coming through the interface chain.

My process has been to measure the voltage of my reamp box to and then verify by ear that it sounds right. And ironically, this does not seem to get tonex to be correct - there's something weird going on there. But with NAM everything I capture is exactly where it should be with respect to gain levels.

Normalizing the track before export will only really change the final capture volume, but I normalize all of mine to -3 dB true peak for consistency. The normalization I mentioned earlier is just for plugin playback volume between captures, and has nothing to do with actually capturing

Ok so essentially the issue is the volume "level" out of one side of the back (mono) of a Scarlett 2i2 into the amp -6 db input? Hmm that should have been line level (out the Scarlett) and I would have thought greater than an instrument level input. But maybe I just needed more volume from the Scarlett?

 

[MirrorProfiles]

I'm the other mod there and also working on an FAQ right now since there are so many new people, so it could also be a good spot to add it. If you also want to send it my way I can pass it along to Steve

Ah that would be awesome, thanks.

Its really grown like crazy the last few weeks, you guys are doing a great job at controlling things there. Hopefully the new FAQ will really benefit new users and also help some of the more repeated questions being asked.

 

[James Freeman]

, this does not seem to get tonex to be correct - there's something weird going on there.

Maybe ToneX is doing auto gain correction in reference to some low level beep in the training tone, in that case, nothing we can do.
--

@usa423 You need a reamp box with analog output level adjustment.

 

[SwirlyMaple]

Another great video from Leo. He nulled the real amp against itself to show that a tube amp doesn't achieve a perfect null with the same DI, and then trained NAM up to 10k epochs to see how the ESR decreases with more epochs (it eventually hits an asymptote). He also compared that against the real amp's self-null, and calculated how long it would take to train all knob permutations if you segmented each knob to a limited set of discrete positions.

 

[2112]

Another great video from Leo. He nulled the real amp against itself to show that a tube amp doesn't achieve a perfect null with the same DI, and then trained NAM up to 10k epochs to see how the ESR decreases with more epochs (it eventually hits an asymptote). He also compared that against the real amp's self-null, and calculated how long it would take to train all knob permutations if you segmented each knob to a limited set of discrete positions.

Leo is awesome, great video!

 

[James Freeman]

So do you all think Kemper and other expensive capturing modelers will go-in-the-way-of-the-dodo?
Do you think capture loaders will be table stakes as IRs have become?

There is no stopping it, the cat is out of the bag and completely open source under MIT license.
We have yet to see a NAM profile loader pedal, but we will, sooner than we expect.

 

[Orvillain]

As soon as we get multi-capture captures (IE: proper bass/mid/treble/presence/depth/additional switches of the actual amp) then I think capture loders will be table stakes.

And yes it's a lot of work.

But how the flip does anyone think their drum software is being made? Multiple weeks in a studio hitting drums like a dummy, and then editing it all up, that's how!!

 

[SwirlyMaple]

So do you all think Kemper and other expensive capturing modelers will go-in-the-way-of-the-dodo?
Do you think capture loaders will be table stakes as IRs have become?

There is no stopping it, the cat is out of the bag and completely open source under MIT license.
We have yet to see a NAM profile loader pedal, but we will, sooner than we expect.

Just my opinion, but I do think Kemper’s sales of new units will begin to fall off drastically, if they haven’t already. It’s a solid device with years of adoption in professional settings, but there just isn’t much reason to buy one now that more accurate profiling tech is available (and for less $).

I think proprietary capturing will wane in popularity, too. This tech is only going to get more common, and other brands will offer capturing devices of similar capability. At that point, the devices with the largest capture ecosystem will be the best selling. TONEX got a great head start, and they might be able to hold on to that niche for a while. But if NAM becomes an open-source standard that other products utilize, it will probably take over. The more open an ecosystem, the more people who will use it.

 

[the-trooper]

So do you all think Kemper and other expensive capturing