A Knowledge Representation Framework for Context-Dependent Audio Processing
This paper presents a general framework for using appropriately structured information about audio recordings in music processing, and shows how this framework can be utilised in multitrack music production tools. The information, often referred to as metadata, is commonly represented in a highly domain and application specific format. This prevents interoperability and its ubiquitous use across applications. In this paper, we address this issue. The basis for the formalism we use is provided by Semantic Web ontologies rooted in formal logic. A set of ontologies are used to describe structured representation of information such as tempo, the name of instruments or onset times extracted from audio. This information is linked to audio tracks in music production environments as well as processing blocks such as audio effects. We also present specific case studies, for example, the use of audio effects capable of processing and predicting metadata associated with the processed signals. We show how this increases the accuracy of description, and reduces the computational cost, by omitting repeated application of feature extraction algorithms.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.