A Data-Driven Synthesis Approach For Indian Languages using Syllable as Basic Unit

This paper describes a data-driven synthesis method for Indian languages using syllables as basic units for concatenation. Unit selection algorithm of this method exploits the advantages of a prosodic matching function, which is capable of implicitly selecting a larger sequence such as words, phrases and even sentences. We show that the proposed method generates high quality speech for Indian languages. 1 Text to Speech Systems The function of Text-To-Speech (TTS) system is to convert an arbitrary text to a spoken waveform. This generally involves two steps: 1) text processing: converting the given text to a sequence of synthesis units and 2) speech generation: generation of an acoustic wave form corresponding to each of these units in the sequence. [Klatt, 1987] [Yegnanarayana et al., 1994]. Current TTS systems employing data-driven synthesis techniques are shown to generate more natural speech than the conventional approaches [Hunt and Black, 1996] [Black and Taylor, 1997] [Kenney Ng, 1998]. Data-driven synthesis techniques employ a large speech corpus containing multiple realizations of each unit with diering prosody. During synthesis, a particular manifestation of a unit is selecte