Video by Wampler Pedals via YouTube
Source
Links:
GuitarML
Download profiles: https://guitarml.com/tonelibrary/tonelib-pro.html
Facebook group:
https://www.facebook.com/groups/674031436584335/
Website and download link:
https://guitarml.com/
NAM:
Download NAM plugin:
https://github.com/sdatkinson/NeuralAmpModelerPlugin/releases/tag/v0.7.1
Download Profiles
https://www.facebook.com/groups/5669559883092788/files/files
https://tonehunt.org/
https://github.com/pelennor2170/NAM_models
https://drive.google.com/drive/folders/1nyO0D-2hREpq5lOtBRWwkRhpxe1m9enp
Facebook group:
https://www.facebook.com/groups/5669559883092788
Download link:
https://github.com/sdatkinson/NeuralAmpModelerPlugin/releases/tag/v0.7.1
____
STEP BY STEP INSTRUCTIONS, COURTESY OF JASON ZDORA (see his video here: https://www.youtube.com/watch?v=lrvuODtk9W0 )
NAM Colab page with "Easy Mode" script already loaded and ready to go:
https://colab.research.google.com/github/sdatkinson/neural-amp-modeler/blob/main/bin/train/easy_colab.ipynb
Training .wav file: https://drive.google.com/file/d/1v2xFXeQ9W2Ks05XrqsMCs2viQcKPAwBk/view
Profile loader VST3/AU: https://github.com/sdatkinson/NeuralAmpModelerPlugin/releases (download the installer)
Jason Zdora’s video on how to profile with NAM:
Step by step how to use NAM "Easy Mode"
1- Download the "V1_1_1.wav" here: https://drive.google.com/file/d/1v2xFXeQ9W2Ks05XrqsMCs2viQcKPAwBk/view
2- Play that audio file through your amp rig and record the output. Amp only! No Cabinet, unless you are wanting to use the entire signal chain including mics, which eliminates using IR’s for the most part for cab modeling)
3- Export your recorded wav file as 24bit 48khz, IN MONO, and name it "output.wav"
4- Make sure your output.wav is EXACTLY the same size as V1.wav (this has to be sample-perfect or it wont work)
5- Go to the Google Colab website https://colab.research.google.com/github/sdatkinson/neural-amp-modeler/blob/main/bin/train/easy_colab.ipynb
6- Click on the folder icon (left side of screen under the spyglass and the "{x}" icons)
7- Drag and drop your audio files (V1.wav and output.wav) into the blank area at that opens up when you click on the folder icon
8- Scroll down to "Step 2: Installation" and click on the "[ ]" inside the darker box inside of the Step 2 area. Should see a bunch of script stuff fill the page. Scroll down.
9- Underneath "Step 3: Train!" find the 2nd darker box. If you want to increase the epochs, find where it says "run(epochs=100)" and type in whatever you want. "run(epochs=1000)" is great but remember that it takes about 8 seconds per-epoch so plan accordingly.
10- Click on the "[ ]" inside of Step 3 to begin running the training. If you see a red "play" button, find the error message below and report it on the user group. If its running correctly, you will see a loading bar filling up over and over with an Epoch number in front of it.
11- When its done, you will see a new folder in the folder section called "exported_models". Open up the sub-folder that has 6 different files in it. Right-click and download "config.json" and "weights.npy". Its a good idea to put them both inside a folder on your computer and name that folder after the hardware you modeled and the date you made the model or other details (easier reference if you make multiple versions).
12- Download the NeuralAmpModeler.vst3/AU using the installer and put it in your VST3 folder or if on a mac the AU will install where it needs to: https://github.com/sdatkinson/NeuralAmpModelerPlugin/releases
13- Start up your DAW and load up NAM Snapshot (folder is called "Steve Atkinson" in your VST3/AU finder).
14- Click the "Choose Model" button on the GUI and find the folder with your config.json and weights.npy files (the newest version of NAM just saves as “model.nam” which is the file you load into your plugin)
15- Add an IR after the NAM Snapshot plugin
16- Enjoy If you want to adjust the epochs or the number of channels run(epochs=X, stage_1_channels=Y) Where: X = 100 for quick and decently accurate X = 1000 for slow and very accurate Y = 12 for lightweight VST3 CPU cost but less articulate Y = 16 for typical kinda heavy CPU cost (this is the default) Y = 20 for very heavy CPU cost but more articulate