Key points are not available for this paper at this time.
We present case studies of three different approaches to voice search for business information: premium DA, fully automated business search, and multimodal search. In each case we describe the user demographics, business economics and technology limitations that drive user interface and system design. Findings from statistical data analysis of real user utterances are highlighted with discussion to their implications in speech recognition, back-end search and voice user interface design.
Chang et al. (Sat,) studied this question.