https://github.com/aisk/nam.cpp

Neural Amp Modeler inference in GGML
https://github.com/aisk/nam.cpp

Last synced: 2 days ago
JSON representation

Neural Amp Modeler inference in GGML

• Host: GitHub
• URL: https://github.com/aisk/nam.cpp
• Owner: aisk
• License: mit
• Created: 2026-07-12T09:42:28.000Z (12 days ago)
• Default Branch: main
• Last Pushed: 2026-07-12T11:57:48.000Z (12 days ago)
• Last Synced: 2026-07-12T13:12:32.434Z (12 days ago)
• Language: C++
• Size: 5.86 KB
• Stars: 1
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# nam.cpp

![](https://repository-images.githubusercontent.com/1298156726/b39879cc-f47d-433e-907f-94f1b2bbb84c)

An experimental Neural Amp Modeler (NAM) inference implementation using [ggml](https://github.com/ggml-org/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.

```sh

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

cmake --build build -j

```

## Usage

```sh

./build/nam [-t N] [-b N] model.nam input.wav output.wav

```

By default, the complete file is processed in one offline graph. Passing a positive `-b`/`--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:

```sh

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

```