Low-level musical descriptors for MPEG-7

摘要

This paper addresses the definition of a minimum set of audio descriptors for MPEG-7. These descriptors are obviously mandatory for MPEG-7 since they describe well-known characteristics that are often used for indexing sound files. These descriptors were defined during the first phase of MPEG-7, and especially for being evaluated during the Lancaster evaluation meeting of February 1999. This set of audio descriptors is the result of a collaborative work, which took place in the audio group within the consortium for low-level audio descriptors. We will show how a top-down approach can yield a multilevel representation of audio files according to their format characteristics and their content. This paper also provides examples of audio descriptions based on the proposed descriptors.