Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Linguistic-Prosodic Processing for Text-To-Speech Synthesis in Italian

Philippe Boula de MareŘil

Elan Informatique, Toulouse, France

The linguistic-prosodic processing applied to text-to-speech synthesis in Italian is described. It proceeds in 5 steps: tokenisation and normalisation of abbreviations, numbers, etc.; part-of-speech tagging, based on function words, terminations and contextual heuristics; shallow parsing, based on a chunk grammar; grapheme-to-phoneme conversion, lexical stress assignment and syllabification by an expert system, particular attention being paid to loanwords and proper names; at least, prosody generation. The last three steps are emphasised.


Full Paper

Bibliographic reference.  Boula_de_MareŘil, Philippe (2000): "Linguistic-prosodic processing for text-to-speech synthesis in italian", In ICSLP-2000, vol.1, 697-700.