In one sentence, it is a docker container image with a pre-installed version of yaafe. Yaafe can be quite complicated to build and the process varies from one os/distro to another. With this docker container you get a shared workspace between your host and container, all the yaafe commands are executed in the container and the results are saved in the shared workspace.
To build the docker container image, assuming that you have a working docker setup, do:
make build
As an example, we can extract the MFCCs and the SpectralFlatness from the provided sample wave files by doing:
./yaafe -c featureplan -r 44100 wav_samples/*
The output on the host side will look like this:
Kawai-K3-Strings-C5.wav
Kawai-K3-Strings-C5.wav.mfcc.csv
Kawai-K3-Strings-C5.wav.spectral_flatness.csv
Low-Conga-1.wav
Low-Conga-1.wav.mfcc.csv
Low-Conga-1.wav.spectral_flatness.csv
The featureplan config file contains the definitions of the features we want to extract from the audio files (the format is documented under http://yaafe.sourceforge.net/manual/quickstart.html#feature-definition-format):
mfcc: MFCC
spectral_flatness: SpectralFlatness