In Text-to-Speech (hence forth) synthesis and related fields, attention is now being given to the research problem of replicating specific speaking styles. This has been facilitated by recent technological improvements in the capabilities of TTS systems, and the incorporation of the technology into real life applications and commercial products. This paper is concerned with describing and modelling the style of a radio news broadcaster for eventual inclusion in a speech synthesis system. A methodology for this task is specified. Pilot experimental work comparing a selected corpus of radio news broadcasts with read aloud versions of their edited transcriptions is reported. The pilot work described compares prosodic parameters of tone units, stresses, and speaking rate.
Bibliographic reference. Tams, Andy / Tatham, Mark / Page, Julian H. (1995): "Describing speech styles using prosody: a pilot study", In EUROSPEECH-1995, 2081-2084.