The paper focuses on the investigation of salient audio features for pattern-based semantic analysis of radio programmes. Most news and music radio programmes have many structure similarities with respect to the appearance of different content types. Speech and music are continuously interchanged and overlapped, whereas specific speakers and voice patterns are more important to recognize. Recent research showed that various taxonomies and hierarchical classification schemes can be effectively deployed in combination with supervised and unsupervised training for semantic audio content analysis. Undoubtedly, audio feature extraction and selection is very important for the success of the finally trained expert system. The current paper employs feature ranking algorithms, investigating audio features saliency in various classification taxonomies of radio production content.
(See document for exact affiliation information.)
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 Join the AES. If you need to check your member status, login to the Member Portal.
