Secondary education

The present paper focuses on the investigation of various audio pattern classifiers in broadcast-audio semantic analysis, using radio-programme-adaptive classification strategies with supervised training. Multiple neural network topologies and training configurations are evaluated and compared in combination with feature-extraction, ranking and feature-selection procedures. Different pattern classification taxonomies are implemented, using programme-adapted multi-class definitions and hierarchical schemes. Hierarchical and hybrid classification taxonomies are deployed in speech analysis tasks, facilitating efficient speaker recognition/identification, speech/music discrimination, and generally speech/non-speech detection-segmentation. Exhaustive qualitative and quantitative evaluation is conducted, including indicative comparison with non-neural approaches. Hierarchical approaches offer classification-similarities for easy adaptation to generic radio-broadcast semantic analysis tasks. The proposed strategy exhibits increased efficiency in radio-programme content segmentation and classification, which is one of the most demanding audio semantics tasks. This strategy can be easily adapted in broader audio detection and classification problems, including additional real-world speech-communication demanding scenarios.► Use of audio pattern classification in radio broadcast semantic analysis concepts. ► Investigation of programme-adaptive modules in real-world demanding scenarios. ► Formulation of direct, hierarchical and hybrid pattern classification schemes. ► Implementation of speech/non-speech segmentation and speaker recognition models. ► Training/evaluation of multiple pattern classifiers utilizing feature ranking.

The current paper focuses on the evaluation of the contribution of stereo information to feature-based pattern classification for audio semantic analysis. In radio productions, where multiple speakers' voices and music coexist, it is quite important to detect voice and music patterns, in order to achieve efficient classification of the audio content. Recent research confirmed that the process of audio feature extraction and selection is critical for the overall system performance. In the current work, the feature extraction procedure is applied in each audio channel of stereo radio programmes separately in order to exploit the differences of audio information between channels. Feature ranking algorithms are employed, aiming to investigate the saliency of the feature differences. The classification performance of the implemented artificial neural systems revealed that the stereo differences have to be taken into consideration during audio semantic analysis.