Slovenian text-to-speech system

This paper presents a text-to-speech system that is capable of synthesising continuous Slovenian speech. Input text is processed by a series of independent modules. That enables easy improvements of separate parts of the system. The first two modules (text normalization and grapheme-to-phoneme conversion) comprise tasks such as end-of-sentence detection, abbreviation and number expansion, special formats conversion, morphological analysis and phonological modeling, accentuation and phrasing. In order to generate rules for our synthesis scheme, data was collected by analysing the readings of ten speakers, five males and five females. A two-level approach has been used for duration modelling, and so-called superpositional approach at pitch modelling. Speech waveform is synthesized using a concatenative TD-PSOLA technique. Acoustic segments inventory consists of diphones and some frequently used polyphones