RDFx: Audio Effects Utilising Musical Metadata

We present a system of audio effects capable of processing and predicting metadata associated with the processed signals. This new class of audio effects has additional inputs and outputs describing features in RDF using the Audio Features Ontology. This data is processed in parallel to the audio signals. We show that by deriving the changes in metadata resulting from audio processing directly from the effect parameters, we can increase the accuracy of descriptors and reduce the computational cost by omitting repeated application of feature extraction algorithms on the output signal. Our demonstration uses a modified version of Sonic Visualiser controlled via the Open Sound Control protocol (OSC) to display metadata at various stages of processing.