5th International Conference on Spoken Language Processing
In spoken dialogue systems, the minimal unit of analysis does not necessarily correspond to a full sentence. A possible approach for language processing is that of splitting the sentence in a sequence of units that can be successively processed by linguistic modules. The goal of the Semantic Boundary (SB) detector is to locate boundaries inside a sentence in order to obtain such minimal units. Useful information for SB detection can be extracted both from the acoustic signal of the utterance and from its corresponding word sequence. In the paper techniques for semantic boundary prediction, based on both acoustic and lexical knowledge, will be presented. Furthermore, a method for combining the two knowledge sources will be proposed. Finally, performance obtained on a corpus of hundreds of person-to-person dialogues will be provided. Best result gives 62.8% recall and 71.8% precision.
Bibliographic reference. Cettolo, Mauro / Falavigna, Daniele (1998): "Automatic detection of semantic boundaries based on acoustic and lexical knowledge", In ICSLP-1998, paper 0333.