nam.cpp

An experimental Neural Amp Modeler (NAM) inference implementation using ggml.

It currently supports classic A1 WaveNet models and offline processing of mono float32 WAV files only.

Build

Requires json-c, pkg-config, and a local ggml source checkout.

cmake -B build -DGGML_DIR=/path/to/ggml
cmake --build build -j

Usage

./build/nam [--threads N] [--stream-block N] model.nam input.wav output.wav

By default, the complete file is processed in one offline graph. Passing a positive --stream-block size uses a reusable fixed-size graph and carries the model's input history between blocks. This bounds inference memory while producing the same causal output; for example:

./build/nam --stream-block 65536 model.nam input.wav output.wav