Design of speech corpus for text-to-speech synthesis

Abstract

This paper deals with the design of a speech corpus for a concatenation-based text-to-speech (TTS) synthesis. Several aspects of the design process are discussed here. We propose a sentence selection algorithm to choose sentences (from a large text corpus) which will be read and stored in a speech corpus. The selected sentences should include all possible triphones in a sufficient number of occurrences. Some notes on recording the speech are also discussed to ensure a quality speech corpus. As some popular speech synthesis techniques require knowing the moments of principal excitation of vocal tract during the speech, pitch-mark detection is also a subject of our attention. Several automatic pitch-mark detection methods are discussed here and a comparison test is performed to find out the best method.