Classification of Audio Codecs with Variable Bit- Rates

Abstract

A large portion of the Internet bandwidth is used for transmission of multimedia such as audio data. For eavesdropping or network surveillance purposes, the first step of a sniffer may be to determine the codec by which a fragment is generated. This problem is usually modeled as a multi-class classification problem. The basic methods for determining the codec type of each fragment rely on the metadata in the corresponding file header. However, in a non-cooperative context, the whole file is not available. So, generally, statistical features extracted from the fragments combined with machine learning algorithms are used for solving this multi-class classification problem. To date, almost all frameworks implicitly assume fixed and known bit-rates for the employed codecs. However, in practical situations, various rates of a specific codec may be used in a network. In this situation, as it is shown in this paper, the classifiers trained by codecs with fixed bit-rates perform poorly when the test data is generated by various rates of the codecs. In this paper, the classification of audio codec fragments with variable bit-rates is considered, simulated, and analyzed. According to the simulation results, for 1 Kbyte fragments, the accuracy of the proposed random forest classifier is about 89%.