A concatenative speech synthesis for monosyllabic languages with limited data

Quality of unit-based concatenative speech synthesis is low while that of corpus-based concatenative speech synthesis with unit selection is great natural. However, unit selection requires a huge data for concatenation that reduces the range of its applications. In this paper, by using temporal decomposition for modeling contextual effects intra-syllable and inter-syllables, we propose a context-fitting unit modification method and a context-matching unit selection method. The two proposed context-specific methods are used in our proposed syllable-based concatenative speech synthesis applied for monosyllabic languages. The experimental results with a Vietnamese speech synthesis using a small corpus support that the proposed methods are efficient. As a consequence, the naturalness and intelligibility of the proposed speech synthesis is high even when we have only limited data for concatenation.