Expressive & Conversational Speech Synthesis
In this collaborative seedling project, with Cereproc SDK, and University of Edinburgh's CSTR, we investigated approaches to conversational speech synthesis. Work focussed on building a unit selection synthesis voice that includes units from conversational rather than just read speech. Additional work looked at a generative model of speech disfluencies, developed from analysis of natural speech data and models of detection of speech disfluencies.
2009
US Army
Publications
-
Automatic detection of unnatural word-level segments in unit-selection speech synthesis. In: IEEE Automatic Speech Recognition and Understanding Workshop (ASRU). Big Island, Hawaii, USA; 2011.
-
Full TextPrediction and realisation of conversational characteristics by utilising spontaneous speech for unit selection. In: Speech Prosody. Chicago, Illinois; 2010. Abstract
-
Cross-domain speech disfluency detection. In: 11th Annual SIGdial Meeting on Discourse and Dialogue. Tokyo, Japan; 2010.
-
Full TextUsing Integer Linear Programming for Detecting Speech Disfluencies. In: North American Chapter of the Association for Computational Linguistics - Human Language Technologies (NAACL-HLT). Boulder, Colorado; 2009. Abstract