Intonation for synthesis of speaking styles

In this paper we describe the adaptation of the Fujisaki model for English intonation of speaking styles. This is for a prosodic model of a radio news broadcaster. Intonational phonological style models are unable to capture the fine threaded realisation differences of individual read-aloud styles, and a quantitative approach is required. The model is used to analyse a corpus of read-aloud speech styles, concentrating on real rather than laboratory speech. Progress towards a novel synthesis model, drawing on modern theories of speech production is described