MicroNAM

Header-only NAM inference with zero dependencies.

Using

1. #include "MicroNAM.h" in your C++ project
2. Generate weights header with nam_to_header.py <model.nam> <model.h> and #include it
3. Example code below for instantiation in app:

MicroNAM::NanoNet<1> mNamInstance; // audio block size 1
mNamInstance.load_weights(mModelWeights);
float inputBuffer[1];
float outputBuffer[1];
mNamInstance.forward(inputBuffer, outputBuffer);

Awknowledgements

MicroNAM is based on an implementation shared by nadavb in an Electrosmith Daisy forum thread.