NAM: Neural Amp Modeler

BenIfin in a NAM thread be like;

7hq5fp.gif
 
StratsoundFan said:
My experience is the exact opposite of yours. There is an abundance of low thud, depending on the amp, the same goes for the high end, it can be bright, but it's never harsh. NAM is as good as it gets.
Click to expand...

SwirlyMaple said:
Hey @BenIfin , I know you mean well and tend to be very diplomatic in your posts. With all respect, I think you somewhat misinterpreted @StratsoundFan 's tone, and honestly your reply comes across as a bit harsh. We're all just casually expressing our opinions here. Somebody saying a statement like "NAM is as good as it gets" is often just an abbreviated way of saying "To me, NAM is as good as it gets." He wasn't contradicting your opinion--just expressing his own, and I don't think there was any need to tell him how to do so.

I mean this only kindly because I'd hope that if I did something similar (and I know for sure I have!), I'd appreciate it if somebody politely let me know as a friend, too.
Click to expand...
Hey SW and SF !

Fair call - I'm not the best with "absolutist" statements when it comes to guitar gear, so went in a bit "hot" there. I didn't actually want or intend to be offensive, so to you SF, I apologize. Thanks SW for your post. It is appreciated :)

All the best to you both / all.
Ben

PS. The meme's have been especially good today !:)
 
BenIfin said:
Hey SW and SF !

Fair call - I'm not the best with "absolutist" statements when it comes to guitar gear, so went in a bit "hot" there. I didn't actually want or intend to be offensive, so to you SF, I apologize. Thanks SW for your post. It is appreciated :)

All the best to you both / all.
Ben

PS. The meme's have been especially good today !:)
Click to expand...
Group Hug Win GIF by MLB
 
NAM is as good as it gets, and it can be proven with science.

@Benlfin - this is probably the first time I've ever said this, but if you truly are hearing what you say you're hearing, then your ears must be a thousand times better than mine. I simply cannot hear any difference between the real amp and the NAM model.
 
Now we have people with good amps that know how to use NAM, I would like to request that we standardize the training tone level for +12dBu inputs.

-9.8dBFS = 1.00v for +12dBu Inputs
Output a sine tone at -9.8dBFS from your DAW and adjust the Reamp output to measure exactly 1.00v.
Make sure the training tone track is not attenuated or normalized after calibrating the Reamp box.

@Orvillain You captured at -17dBFS = 0.500mV right?
That is equivalent to -11dBFS = 1.00v, because double the voltage is +6dB.
Your capture is ready for +13.2 dBu inputs.
 
James Freeman said:
Now we have people with good amps that know how to use NAM, I would like to request that we standardize the training tone level for +12dBu inputs.

-9.8dBFS = 1.00v for +12dBu Inputs
Output a sine tone at -9.8dBFS from your DAW and adjust the Reamp output to measure exactly 1.00v.
Make sure the training tone track is not attenuated or normalized after calibrating the Reamp box.

@Orvillain You captured at -17dBFS = 0.500mV right?
That is equivalent to -11dBFS = 1.00v, because double the voltage is +6dB.
Your capture is ready for +13.2 dBu inputs.
Click to expand...
Yeah, so for my soundcard line outs it seems 1kHz sine-tone at -17dBFS is equivalent to 500mVAC. I'm a bit confused by all the level stuff to be honest. Happy to investigate it all and find the best practices.

I've matched my reamp box's level knob so that when I plug it into my Neve DI box which then goes into input 8 on my Discrete 8, I get the same -17dBFS signal, when using the same sine-tone.

I should note however that my setup has a few complications. I'm not just using one interface. Here's what it looks like:

- Presonus Quantum as main audio interface.
- Antelope Discrete 8 Pro SC word-clock'd and ADAT'd to the Quantum.
- Line output 1 on the Quantum goes to reamp box. The maximum output level for these line outs is +18dBU. But I wasn't getting a 500mVAC signal until I attenuated the sine in the DAW by -17dBFS.
- Reamp box goes to the amp (or any intervening EQ, OD, or boost pedals)
- Amp itself goes to Suhr Reactive Load.
- The Suhr comes into the 2nd input on the Discrete 8 Pro. This is set to line level.
- My Neve DI box goes into input 8 on the Discrete 8 Pro, and this has a maximum input level of +18dBU.
 
Last edited:
Listened to Orvillains files, slapped the second one into Logic, tried my best to line both NAM and amp up so I could invert one. No, they're not nulling but they're amazingly close. So close that I'd place a bet on nobody being able to tell in a real ABX blindtest.
I have done the same with plenty of comparable files, nothing has ever come that close. And well, this here even is an MP3 file.
Amazing, really.
 
I'm trying with my own training track, I don't like the results with mid gain tones.
Besides the first clicks, sweeps and noises, there is a high level of noise floor where the sped up guitar and bass parts in the v1_1_1.wav training file, I also think that speeding up the guitar part does more bad than good.
I'm using Google Colab with the Main repository, I don't want to install a few gigabytes just to train a few models.
 
James Freeman said:
I'm trying with my own training track, I don't like the results with mid gain tones.
Besides the first clicks, sweeps and noises, there is a high level of noise floor where the sped up guitar and bass parts in the v1_1_1.wav training file, I also think that speeding up the guitar part does more bad than good.
I'm using Google Colab with the Main repository, I don't want to install a few gigabytes just to train a few models.
Click to expand...
You could record and “borrow” TONEX’s training track. It covers a lot of ground pretty thoroughly :) Plus, we’d have the amusing irony of using TONEX files to train NAM :bag
 
Orvillain said:
Heheh. You could compile the Kemper, QC, ToneX, and NAM training tracks all into one long-ass thing.... wonder if it would do anything different??
Click to expand...
I added a short guitar DI to the NAM input file. I have only tested it once though it did seem to improve the feel of the capture when using that specific playing technique.
 
Orvillain said:
Heheh. You could compile the Kemper, QC, ToneX, and NAM training tracks all into one long-ass thing.... wonder if it would do anything different??
Click to expand...
Wait. What if you connected NAM’s output to TONEX’s input, and NAM’s input to TONEX’S output, and then let them train each other… Did we just create cold fusion? Or maybe this how we open Stargate?!

big bang supernova GIF
 
Orvillain said:
NAM is as good as it gets, and it can be proven with science.

@Benlfin - this is probably the first time I've ever said this, but if you truly are hearing what you say you're hearing, then your ears must be a thousand times better than mine. I simply cannot hear any difference between the real amp and the NAM model.
Click to expand...
Hey !

I *wish* I had the golden ears that some on other forums claim to have - I definitely do not.

When I listened to your clip I just let it play all the way through both clips non-stop - definitely could not tell which was the Amp and which was NAM - and letting them play through - non-stop - I also could not detect any meaningful differences. Pretty much why I said that live or recorded its irrelevant.

To me this also applies to the good hardware units - Fractal, L6, Boss GT, Kemper, QC, Tonex etc. the "good" stuff these days is all at such a high level of accuracy and quality that it has just become academic to me - b.t.w have I ever mentioned how much I despise Neural as a company ;)

It was *only* when I clicked back and forth repeatedly on your clip that I noticed the [small] differences I mentioned in my prior post .... cant really say much more or why - but to my ears, that's just what I did, and what I heard (?) Others doubtlessly heard it differently.

An interesting aside - about 2 months ago I bought a 2nd hand Nux Amp Academy [for a "play"] and for the equivalent of $US 130 - I loaded my own IR and its raw Amp tones are utterly stunning - if I posted clips of it against NAM / Tonex / Helix / Boss GT clips all using the same IR - no-one would ever be able to tell which is the $130 modeler - and yes, it also "feels" excellent.

Anyway - back to the issue - check-out this clip posted in the last day or so - Real Amp vs Tonex vs Proteus vs NAM - can anyone realistically and credibly claim or say that Tonex sh%ts on NAM, or NAM sh%ts on Tonex, or NAM Captures the Amp "better" than Tonex or Tonex Captures the Amp "better" than NAM (?) I don't / cant. Maybe if I had the original .WAV files and could be bothered, and went back and forth quickly between clips etc. I might be able to pick some small differences between Tonex and NAM .... but what's the flying-f%ck point of that ?

Lastly - as I have said before, I have absolutely zero issue with NAM as a potential product - zero.

Its the whole "keys to the kingdom" vibe that grates me.

To me, none of the above is controversial - yet it seems that the response to anyone who tries to point out some of the above, is that NAM is new, and will only get better - well, I say to that, of course it will fu%king get better - no-one is denying that.

The "grate'y" bit for me - and this is the "kicker" - is that those same people also choose to deliberately ignore the fact that IK Tonex [for example] which is also A.I built - will also fu%king get better - maybe one will get "better-er" than the other - who the f%ck knows.

Anyway - that's my 2c explanation / reasoning.

Ben
 
BenIfin said:
Hey !

I *wish* I had the golden ears that some on other forums claim to have - I definitely do not.

When I listened to your clip I just let it play all the way through both clips non-stop - definitely could not tell which was the Amp and which was NAM - and letting them play through - non-stop - I also could not detect any meaningful differences. Pretty much why I said that live or recorded its irrelevant.

To me this also applies to the good hardware units - Fractal, L6, Boss GT, Kemper, QC, Tonex etc. the "good" stuff these days is all at such a high level of accuracy and quality that it has just become academic to me - b.t.w have I ever mentioned how much I despise Neural as a company ;)

It was *only* when I clicked back and forth repeatedly on your clip that I noticed the [small] differences I mentioned in my prior post .... cant really say much more or why - but to my ears, that's just what I did, and what I heard (?) Others doubtlessly heard it differently.

An interesting aside - about 2 months ago I bought a 2nd hand Nux Amp Academy [for a "play"] and for the equivalent of $US 130 - I loaded my own IR and its raw Amp tones are utterly stunning - if I posted clips of it against NAM / Tonex / Helix / Boss GT clips all using the same IR - no-one would ever be able to tell which is the $130 modeler - and yes, it also "feels" excellent.

Anyway - back to the issue - check-out this clip posted in the last day or so - Real Amp vs Tonex vs Proteus vs NAM - can anyone realistically and credibly claim or say that Tonex sh%ts on NAM, or NAM sh%ts on Tonex, or NAM Captures the Amp "better" than Tonex or Tonex Captures the Amp "better" than NAM (?) I don't / cant. Maybe if I had the original .WAV files and could be bothered, and went back and forth quickly between clips etc. I might be able to pick some small differences between Tonex and NAM .... but what's the flying-f%ck point of that ?

Lastly - as I have said before, I have absolutely zero issue with NAM as a potential product - zero.

Its the whole "keys to the kingdom" vibe that grates me.

To me, none of the above is controversial - yet it seems that the response to anyone who tries to point out some of the above, is that NAM is new, and will only get better - well, I say to that, of course it will fu%king get better - no-one is denying that.

The "grate'y" bit for me - and this is the "kicker" - is that those same people also choose to deliberately ignore the fact that IK Tonex [for example] which is also A.I built - will also fu%king get better - maybe one will get "better-er" than the other - who the f%ck knows.

Anyway - that's my 2c explanation / reasoning.

Ben
Click to expand...
TONEX is great! I like it, I own it, and I think a lot of the other people in this thread do, too. But pardon me for pointing out… this is the

“NAM: Neural Amp Modeler”​

thread. It’s right up there at the top, where the titles go ^ ;) :whistle :rofl
 
You must log in or register to reply here.
Back
Top