Real Time Frequential Filtering

A framework to design audio filters in the time frequency domain.

Example

#include <iostream>

#include <Eigen/Core>

#include "rtff/filter.h"

...
std::error_code err;
rtff::Filter filter;
filter.Init(channel_number, err);
if (err) {
  std::cerr << "Error when initializing the filter" << std::endl;
  return -1;
}
filter.execute = [](std::vector<std::complex<float>*> data, uint32_t size) {
  for (auto channel_idx = 0; channel_idx < data.size(); channel_idx++) {
    auto buffer = Eigen::Map<Eigen::VectorXcf>(data[channel_idx], size);

    // Do stuff to your buffer here !

  }
}
rtff::AudioBuffer buffer(filter.block_size(), channel_number);

...
filter.ProcessBlock(&buffer);

Latency

Computing the short time fourier transform implies a latency. If you want to develop a plug-in, you need to inform the DAW about the number of frames of latency produced by your filter.
The AbstractFilter::FrameLatency() function gives you exactly what you need.

Documentation

The documentation is based on sphinx, breathe and readthedocs

If you ever want to build it, we provide Dockerfile

docker build -t rtff-docs -f docs/Dockerfile .
# Generate doxygen xmls
docker run -it --rm -v$(pwd):/code rtff-docs doxygen
# make sphinx documentation
docker run -it --rm -v$(pwd):/code rtff-docs make html

Cross compile

You can use the dockcross project to cross compile the library.
For example to build an arm64 cross compiled version do:

docker run --rm dockcross/linux-arm64 > dockcross
chmod +x dockcross
./dockcross cmake -H. -Bbuild -GNinja -Drtff_use_mkl=OFF -Drtff_fftw_extra_configure_flags=--host=arm-linux-gnueabi
./dockcross ninja -Cbuild

Note that we disable the use of the mkl and force a cross compile flag on the fftw.