Second European Conference on Speech Communication and Technology

Genova, Italy
September 24-26, 1991


Assessment of Intonation in Text-to-Speech Synthesis Systems - A Pilot Test in English and Italian

Martine Griee (1), Kiki Vagges (2), Daniel Hirst (3)

(1) Department of Phonetics and Linguistics, University College London, UK
(2) Centro di Studio per le Ricerche di Fonetica-CNR, Padova, Italy
(3) Institut de Phonetique d'Aix, France

Tests of prosodic form are seen as the first stage of a programme to evaluate the quality of intonation algorithms in text-to-speech synthesis. Before we can assess the functional appropriateness of an intonation contour , we must first assess whether the synthesiser's realisation of this contour is adequate. A pilot test to evaluate the realisation of falls and rises over stretches of differing segmental complexity was carried out in English and Italian. For each segmental context, scores for these two contours were calculated relative to the monotone realisation. Keywords: Text-to-Speech Synthesis, Speech Output, Assessment, Intonation, Prosody

Full Paper

Bibliographic reference.  Griee, Martine / Vagges, Kiki / Hirst, Daniel (1991): "Assessment of intonation in text-to-speech synthesis systems - a pilot test in English and Italian", In EUROSPEECH-1991, 879-882.