| Summary: | 碩士 === 國立嘉義大學 === 資訊工程學系研究所 === 99 === It is one of most essential issues to extract the keywords from conversational speech for understanding the utterances from speakers. This thesis aims at keyword spotting from spontaneous speech for semantic object detecting. We proposed information structure based approach with prosodic features that are used for semantic object detection. The prosody words are segmented from speaker’s utterance according to the pre-training decision tree. The supported vector machine is further used as the classifier to judge the prosody word is semantic object or not.
This thesis mainly consists of three parts; information structure, prosody word segmentation, and semantic object detection are included. We first describe information structure originated from cognitive psychology. Instead of syntactic analysis, the pragmatics viewpoint is used to observe the content of conversation here. We can divide the conversation content into focus and topic parts in the utterance. It is more robust by information structure for semantic object detection from ungrammatical spontaneous speech compared to syntactic analysis. In the second part, the prosody word boundary segmentation algorithm based on decision tree is illustrated. Besides the data driven feature, the knowledge obtained from the corpus observation is integrated in the decision tree. Finally, the semantic objects in the focus part are extracted using prosody features by sported vector machine (SVM). According to the experimental results, we can find the proposed method outperform the phone verification approach especially in recall and accuracy. This shows the proposed approach is operative for semantic object detecting.
|
|---|
