Abstract

Data vocalization is the process of summarizing data via voice output. We present CiceroDB, a novel database system, designed from the ground up for vocal output of query results. It is targeted at scenarios in which visual output is either impossible or undesirable. CiceroDB exploits the particularities of vocal output to reduce data processing overheads. For instance, it uses sampling to converge to highlevel voice descriptions and overlaps voice output with background processing tasks. In this paper, we describe preliminary results and ongoing research efforts.