NAM: Neural Amp Modeler

[MirrorProfiles]

Iirc the output impedance on the Suhr is pretty high for a "line" output. The effect will depend on the interface line level input impedance, though. 10K might be fine, 2.2K maybe less so.

Regular RL specs say “below 6k”, RLIR says “600 Ω”.

I guess cable length and input impedances may play into it too.

 

[paisleywookiee]

I’m just going to use a speaker level DI between the head and cab. Don’t really need a reactive load. Yeah, it’s gonna be noisy for a few minutes, but who cares?

 

[AlbertA]

I measured my Suhr RL - With the line output DI level knob al the way down, about 2.2KΩ.

Knob all the way up, at 100Hz about 5KΩ, at 1KHz , 9.5KΩ.

So following 10x rule of thumb - choose an input that has an impedance > 100KΩ

 

[Chocol8]

I’m just going to use a speaker level DI between the head and cab. Don’t really need a reactive load. Yeah, it’s gonna be noisy for a few minutes, but who cares?

I recommend that anyway because the actual load of the cab gets captured. Not as big of a deal for heads, but for combos I generally want the real cab impedance curve. YMMV as it is up to preference.

 

[Chocol8]

I’m just going to use a speaker level DI between the head and cab. Don’t really need a reactive load. Yeah, it’s gonna be noisy for a few minutes, but who cares?

BTW, if you already have the speaker level DI, I would definitely get the Lehle P split over the IK capture box. You would only use half of the IK box and I think the Lehle is going to have less impact on the signal.

 

[paisleywookiee]

With NAM, how hot do you want your recorded reamp to be?

 

[MirrorProfiles]

With NAM, how hot do you want your recorded reamp to be?

For amps, I usually normalise so the peaks are at -1dBFS.

It doesn’t really matter for amps, but for pedals and preamps, if you want to preserve output levels for gainstaging then you’d measure your headroom and embed that info in the metadata and then leave the captured levels alone

 

[Dom Mcsweeney]

With NAM, how hot do you want your recorded reamp to be?

my preference is no louder than -6db peaks.
it basically dictates how loud you like your amp model to be when you load it into whatever plugin or device

most the time you don't want it too loud in any situation or else it is pretty much on the edge of clipping and I am not a fan of that
ymmv

 

[paisleywookiee]

For amps, I usually normalise so the peaks are at -1dBFS.

It doesn’t really matter for amps, but for pedals and preamps, if you want to preserve output levels for gainstaging then you’d measure your headroom and embed that info in the metadata and then leave the captured levels alone

my preference is no louder than -6db peaks.
it basically dictates how loud you like your amp model to be when you load it into whatever plugin or device

most the time you don't want it too loud in any situation or else it is pretty much on the edge of clipping and I am not a fan of that
ymmv

Alrighty then, well that gives me a ballpark to work with. I accidentally did one reamp with the hardware compressor on my interface turned on. That really blew my hair back!

 

[bikescene]

When I aim for -6dB peaks on gainy amps, the Raw output setting gets close to the volume when the Normalize slider is on. For me, the goal is to match the Normalized volume.

 

[paisleywookiee]

When I aim for -6dB peaks on gainy amps, the Raw output setting gets close to the volume when the Normalize slider is on. For me, the goal is to match the Normalized volume.

Well, hopefully this weekend is fruitful. Going to aim for at least a few cool captures. I found my old Radial Reamp box, and a shitty but functional speaker level DI, so I’m good to go.

 

[paisleywookiee]

Here's me, btw.

paisleywookiee's NAM profiles & IR's · TONE3000

Explore paisleywookiee's latest Neural Amp Modeler (NAM) profiles and impulse responses (IR's) for guitar tones, bass tones, recording and more! · I'll be occasionally adding FAS captures for myself, so I can use them to record demos whilst out and about with my laptop. However, I'll also be...

www.tone3000.com

 

[Elric]

Is there any kind of standardization on NAM test signals/levels/etc?

 

[2dor]

Is there any kind of standardization on NAM test signals/levels/etc?

There is a guideline for measuring the signal levels of your reamp:

How to determine the calibration levels of your recording setup — neural-amp-modeler 0.12 documentation

and then leverage this data along with your audio interface's specifications to get the same experience the author of the profiles was aiming for:

 

[Progressive_disclosure]

Just did a new batch of captures of my ENGL Fireball 25, this time using the internal (resistive) load.

One capture also has the clean channel with a Wampler Euphoria in front. Any tips to get even better captures are welcome!

Link: https://www.tone3000.com/tones/engl-fireball-25-internal-load-37275

 

[Progressive_disclosure]

Is there a way to add the calibration metadata to the captures on Tone3000 directly? Maybe I’m missing something?

 

[Dom Mcsweeney]

Is there a way to add the calibration metadata to the captures on Tone3000 directly? Maybe I’m missing something?

yes there is if you are training the model on tone3000

click on 'create tone' and then in trainer settings > click "advanced" and you should see this page. Look at the bottom of the page and you should see calibration at the bottom.

If you are using the local trainer, or google colab trainer and then uploading your models to Tone3000 after, you can use the advanced settings in the local or colab trainer instead.

 

[Progressive_disclosure]

yes there is if you are training the model on tone3000

click on 'create tone' and then in trainer settings > click "advanced" and you should see this page. Look at the bottom of the page and you should see calibration at the bottom.

If you are using the local trainer, or google colab trainer and then uploading your models to Tone3000 after, you can use the advanced settings in the local or colab trainer instead.

View attachment 51700

Ah, there it is, thanks! I’ll try to figure out the values for when I do captures with a real speaker!

 

[2dor]

The new NAM training software released a few days ago brought in some great speed enhancements for Apple Silicon machines.
Training locally on a 14" M2 Max, I get 2-3 seconds / epoch which is the same as on my nVidia RTX 4070 desktop machine. Crazy stuff.

 

[Sascha Franck]

The new NAM training software released a few days ago brought in some great speed enhancements for Apple Silicon machines.
Training locally on a 14" M2 Max, I get 2-3 seconds / epoch which is the same as on my nVidia RTX 4070 desktop machine. Crazy stuff.

So how many epochs do you usually go for?