Desk Research: We try to obtain as much as possible different MP3 encoders and preclassify them by the encoding libraries they use. Besides, we set up a database of MP3 files producted by different encoders.
Feature extraction
Data Exploration: We assemble a list of typical features of the respective MP3 encoders. Here, we differ between deterministic "e;safe"e; features (e.g. file headers) and probabilistc features like distributions and other statistical measures.
Classification prototype
Implementation: Based on the feature list we assemble a basic clasifier framework which will rest extensible for future developments and insights. All results will be published or released under the GPL.