NAM tools

Cab embed

Bake a cabinet IR into an amp capture, so one NAM file plays the amp and the cab together in any player that loads A2 captures. Listen to the pair in your browser first, then train the new file on your own free Kaggle account.

Step 1: Choose your files

They stay on your computer until you start a training.

Step 2: Listen

Plays through Neural Amp Modeler's engine in your browser. Nothing is uploaded. Use headphones with live input to avoid feedback.

Your guitar signal

A dry guitar track, recorded without an amp.

What you hear

Amp + cab IR is what training will learn. Amp only lets you hear what the IR adds.

-6 dB

Measure the match needs the amp capture, the cabinet IR and a trained capture.

Step 3: Train on Kaggle

Kaggle's GPU learns the amp and the cab as one A2 capture, on your account and your free GPU hours.

Your Kaggle account

Create one on Kaggle under Settings → API. Your account must be phone-verified to use GPUs.

The new capture

Standard takes about three times as long as Draft, and High about six. Waiting for a GPU can add more.

Your Kaggle key stays private. It's saved only in this browser and sent securely with each request to this site's server, which uses it for that one request and forgets it. This page has no ads or third-party scripts. A Kaggle key can manage your whole Kaggle account, so delete it on Kaggle when you no longer need it.

What happens when you train
  • This site sends your capture and IR to Kaggle as one private notebook on your account, with a T4 GPU and internet access (to install NAM's trainer). Nothing is kept on this server.
  • The notebook plays NAM's official training signal through your capture and IR, the way NAM Mixer does, then trains an A2 capture on the result with neural-amp-modeler 0.13.
  • The new file keeps the capture's name, maker and calibration, and is marked as an amp + cab capture.
  • An A2 capture hears about 130 ms. The amp uses most of that, so a longer IR's tail is approximated.

Your trainings

Listed only in this browser. The page checks on them every 30 seconds while it's open, so you can close it and come back.

    No trainings yet.