DART: A Framework for Distributed Audio Analysis and Music Information Retrieval
Audio analysis algorithms and frameworks for Music Information Retrieval (MIR) are expanding rapidly, providing new ways to garnish non-trivial information from audio sources, beyond that which can be ascertained from unreliable metadata such as ID3 tags. The analysis component of MIR requires extensive computational resources. MIR is a broad field, and many aspects of the algorithms and analysis components that are used are more accurate given a larger dataset for analysis, and is often quite DSP/CPU intensive. A Desktop Grid based implementation would reduce computation time and provide access to potentially thousands of MP3 files on target machines, where the files analysed locally on clients’ machines, transferring back only the metadata/results of the analysis. This avoids legal issues, and saves bandwidth.
The DART application framework developed at Cardiff University focuses on the analysis of audio, with a particular interest into MIR. The existing application is designed and created in Triana, a graphical workflow-design environment that is used as a development test bed for the algorithms that will be distributed. The algorithms are programmed in a modular way, which allows only the relevant building blocks of the workflow to be converted into the standalone DART Java application, which is in turn converted into a JAR (multi-platform) executable, and is distributed to target machines across the Desktop Grid using BOINC or XtremWeb.
Eddie Al-Shakarchi, Cardiff University – UK