Affordable Access

Information fusion approaches to the automatic pronunciation of print by analogy

Publication Date
  • Computer Science
  • Medicine


doi:10.1016/j.inffus.2004.08.002 Information fusion appro f p ,*, Y chool Southampton SO17 1BJ, UK b Institute for Biodiagnostics (Atlantic), National Research Council Canada, Neuroimaging Research Laboratory, 1796 Summer Street, Suite 3900, Halifax, Nova Scotia, Canada B3H 3A7 c Faculty of Computer Science, Dalhousie University, Halifax, Nova Scotia, Canada B3H 1W5 riority of rank fusion over the other methods. � 2004 Elsevier B.V. All rights reserved. rather complex and indirect. Since printed text is not al- ways a very direct specification of pronunciation, it is sensible to convert it to something much closer to a rep- resentation of the corresponding sound sequence. Lin- guists have long used the phoneme as an abstract unit . * Corresponding author. Tel.: +44 023 80 594577; fax: +44 023 80 594498. E-mail addresses: [email protected] (R.I. Damper), yannick. [email protected] (Y. Marchand). Information Fusion 7 (200 1566-2535/$ - see front matter � 2004 Elsevier B.V. All rights reserved Keywords: Score fusion; Rank fusion; Automatic pronunciation; Analogical reasoning; Speech synthesis 1. Introduction Text-to-speech (TTS) synthesis is an emerging tech- nology with many potential applications to next-genera- tion computer and information systems [1,2]. A very important sub-problem within TTS synthesis is the automatic generation of word pronunciations from tex- tual input, or �print�. Unless we are able to derive a good specification of pronunciation of the individual words in the input, we cannot hope to produce a satisfactory TTS system. Yet for many languages, such as French and English, the relation between letters and sounds can be Received 29 March 2004; received in revised form 5 August 2004; accepted 5 August 2004 Available online 11 September 2004 Abstract Automatic pronunciation of words from their spelling alone is a hard computational problem, especially for languages like Eng- lish and French where there is onl

There are no comments yet on this publication. Be the first to share your thoughts.


Seen <100 times