Speech Analytics
I. Dan Melamed, Mazin Gilbert · 2011
This chapter describes the ways in which speech analytics can be helpful. It discusses the inherent technical challenges and some of their possible solutions, with reference to the few pieces of relevant research that tha author have been able to find. The chapter pays particular attention to user interface issues, since a speech analytics system is only as useful as an analyst’s ability to understand and control it. Usually, however, transcription is done by an automatic speech recognition (ASR) system. In this case, quality of analyses done by the speech analytics system depends on the quality of its ASR system. At a minimum, the text feature extraction module records which words occur in which transcripts. Acoustic information is what distinguishes speech analytics from text analytics. The job of the relational feature extraction module is to convert metadata into a form suitable for storing in a relational database management system (DBMS). Controlled Vocabulary Terms database management systems; speech recognition equipment; speech synthesis