Neural Amp Modeler (NAM) is an open-source neural network emulator for guitar amplifiers and effects. Record your real amp or pedal, train a model on the recordings, and NAM learns to reproduce that exact tone, which you can then play through in real time, share, and load anywhere the format is supported.
NAM is developed on GitHub, where the core project has earned more than 3,100 stars under the MIT license, and it remains actively maintained, with updates in August 2026. This repository handles the training side: it learns models from audio captures of your gear and exports them to portable .nam files. A partner project, NeuralAmpModelerPlugin, plays those models in real time as a standalone application or audio plugin inside your DAW. Around the two projects a substantial ecosystem has grown: the neuralampmodeler.com site, full documentation, and a large community sharing captures of sought-after amplifiers, boutique pedals, and entire signal chains. Because the models are small files, a captured amp can be backed up, traded, and recalled exactly, something physical gear cannot do. NAM’s capture quality made it a reference point in the amp-modeling world and pushed commercial vendors to raise their game. Training runs on standard Python tooling and works with ordinary audio interfaces, so any guitarist with a re-amp setup can profile their own rig.
Neural Amp Modeler is completely free and open source under the MIT license, including the training code and the playback plugin. Community model captures are also shared freely.
NAM is for guitarists and producers who want their physical amps available inside the box, touring musicians backing up their rigs, and audio machine-learning enthusiasts interested in a mature, real-world application of neural audio modeling.