NAM: Neural Amp Modeler

I used my Egnater Tourmaster V30 4x12 cab as the load, and it definitely sounds better than a loadbox.
+1
There are two very clear benefits of using a real cab as a load and tapping the signal in parallel.
1. You get an authentic frequency and time/resonance response.
2. You don't go full retard on the Master volume and keep it at realistic volumes.
 
1690706629786.png

Give that a bash. Calibrated for 18dBu.
 
All-In-One dBu Calculator:
Code:
https://drive.google.com/file/d/1yg3FPZPMTpNwaVd1IKfuy2zKMygpk_Bl

First post updated.
 
what makes it worse is that those same people seem to also have little to no idea how to dial in an amp :(

Man there are stacks of absolutely sh$tful shockers out there, both for Tonex Land and NAM Land.

Ben
Have you provided any amp captures to the NAM or Tonex communities?

I’ve only recently started my own journey into creating captures, but I can assure it’s more complicated than just “dialing in an amp” as some of us might have assumed.

Dialed in for which guitar? Which pickups? Cab? Mic? Room? Direct? If direct, what load?

I have little faith that anything I’ve captured has translated outside my recording/listening environment. It’s very apparent that profiling is still in its infancy in the sense that there is a total lack of standardization. To embrace profiling either as a content provider or a consumer requires patience and perseverance. It ain’t for everyone.
 
What levels are you guys recording your reamps back into your DAWs?

I tried making a capture tonight, after being out of the loop on the newest capturing advances. The normalize slider brought the volume up about 5dB, so I am wondering if I’m hitting my DAW too quietly compared to others. My WAV peaks at -12dBFS based on how I have my rig dialed in at the moment.
 
What levels are you guys recording your reamps back into your DAWs?

I tried making a capture tonight, after being out of the loop on the newest capturing advances. The normalize slider brought the volume up about 5dB, so I am wondering if I’m hitting my DAW too quietly compared to others. My WAV peaks at -12dBFS based on how I have my rig dialed in at the moment.

It doesn't matter too much as long as your SNR is good. You can choose to normalize before exporting or just leave it and let NAM do it. I usually render around -9 dB for self consistency I think, but its kind of arbitrary
 
I’m curious to see if my captured input levels translate to other systems, if you have the time.

https://app.box.com/s/b9ckc6gtsom43v1y7yt4fry5ow9ilnkr

I used my Plexi-inspired 50W head, with 5000pF bright cap, with the volume at around 2:00 for a generic crunch. The bass was set low, perhaps a bit too much for some tastes. The amp was connected to a Suhr Reactive Load.

I had first set my interface input for -12dBFS in response to a 0.707V RMS 440Hz signal, and set my reamp box to unity gain based on that input. Default 100 Epoch capture.

My amp is kind of noisy with the big bright cap. It’s interesting that the capture doesn’t have the background hum. At the same time, I wonder if I can maybe try to improve gain staging or isolation in my chain.
 
@northern_fox any tips on queuing up a load of files to train all at once so I can leave this thing training overnight?? ideally I'd like it to spit out the graphs for each one at the same time.
 
@northern_fox any tips on queuing up a load of files to train all at once so I can leave this thing training overnight?? ideally I'd like it to spit out the graphs for each one at the same time.
Not him but I seem to remember Steve saying you can select multiple files and disable the graph checks and it will run them all. Haven't tried it.
 
@northern_fox any tips on queuing up a load of files to train all at once so I can leave this thing training overnight?? ideally I'd like it to spit out the graphs for each one at the same time.

What @Deadpan said. I actually was the one who added that functionality way back! Select as many files as you want on the output audio field. It will train all of them, and do silent run + ignore checks if you want it to run fully automatic with no user input. It'll save an ESR plot for each one as it goes as well, but you can also watch the results in realtime on tensorboard

Metadata will be shared across all of them, as well as any advanced options

1691427247953.png
 
I typically just drop this batch script next to the lightning_logs folder and it'll pull up the tensorboards really easily (change the env name to be correct)

 
Fwiw, Melda added .nam support to their MGuitarArchitect - and in case you don't know their plugins already, now is the time to change that, there's some very excellent ones among them, some killer free ones, too.
 
What @Deadpan said. I actually was the one who added that functionality way back! Select as many files as you want on the output audio field. It will train all of them, and do silent run + ignore checks if you want it to run fully automatic with no user input. It'll save an ESR plot for each one as it goes as well, but you can also watch the results in realtime on tensorboard

Metadata will be shared across all of them, as well as any advanced options

View attachment 9634
Ahhhh sick! I don't have to write it myself then! Nice one!
 
1691445813394.png

I've got 37 reamps here ready for training; plus one additional one from another session.

Mesa Boogie Dual Rectifier 2018 multi-watt version:

Green - Clean, Pushed Low, Pushed High
Orange - Raw, Vintage, Modern
Red - Raw, Vintage, Modern

Diezel VH4 2001:

Channel 1 - without bright switch and with bright switch
Channel 2 - without bright switch and with bright switch
Channel 3 - Nice chunky rhythm tone
Channel 4 - Nice chunky rhythm tone, but you can also use it for leads

Marshall JVM410HJS:

Clean - Green Low, Green High, Orange Low, Orange High, Red Low, Red High
Crunch - Green Low, Green High, Orange Low, Orange High, Red Low, Red High
OD - Green Low, Green High, Orange Low, Orange High, Red Low, Red High

Orange Rockerverb MKIII 100-watt:
Clean
Clean dimmed
Distortion Low, Mid, High

Will get these training either overnight, or tomorrow. Doing some 3D rendering right now so my GPU is already solidly in use.
 
By now, is there any way allowing me to load NAM captures on this aging Mac Pro still running macOS Mojave? Anyone knows?
 
Back
Top