There is an immense amount of audio data available currently whose content is unspecified and the problem of classification and generation of metadata poses a significant and challenging research problem. We present a review of past and current work in this field; specifically in the three principal areas of segmentation, feature extraction, and classification and give an overview and critical appraisal of techniques currently in use. One of the major impediments to progress in the field has been specialism and the inability of classifiers to generalize, and we propose a non exclusive generalized open architecture framework for classification of audio data that will accommodate third party plugins and work with multi-dimensional feature/descriptor space as input.
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.