The objective of this paper is to determine which of the MPEG-7 standard low-level sound descriptors are the most significant in the process of automatic classification of musical instrument sounds. First, pitch detection is performed. Then, the parametrization stage of musical sounds based on descriptors contained in the MPEG-7 standard is carried out. Next, a thorough statistical analysis of the feature vectors obtained is performed. For the purpose of automatic classification, two decision systems based on artificial neural networks (ANNs) and rough sets, are used. Both decision systems are trained with feature vectors consisted mostly of parameters contained in the MPEG-7 standard, however their content being reduced after statistical analyses. In addition, a comparison of results obtained by these decision systems with the results got from the nearest neighbor algorithm is made.
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 free for AES members and E-Library subscribers.