基于音频内容的语义级场景检索

Audio Semantic_level Context Detection

本文对基于内容的音频检索提出了一种分级方法,第一级:用HMM对音频事件的统计特性建模;第二级:用SVM结合一些音频事件对特定语义场景建模,完成对语义场景的检索。实验证明,HMM和SVM的结合对音频语义级场景的检索达到比较理想的效果。

In this paper,Audio semantic-level content analysis, we propose a hierarchical approach for content-based retrieval of audio.The first, HMMs are used to model the statistical characteristics of several audio events.And the second, SVMs are used to fuse the characteristics of various audio events related to some specific semantic concepts.,then model semantic context. The experimental results show that the approach is effective in detecting semantic context.