On building phonetically and prosodically rich speech corpus for text-to-speech synthesis

This paper proposes a way of preparing and recording a speech corpus for unit selection text-to-speech speech synthesis driven by symbolic prosody. The research is focused on a phonetically and prosodically rich sentence selection algorithm. Symbolic description on a deep prosody level is used to enrich the phonetic representation of sentences (by respecting the prosodeme types phones appear in). The resulting algorithm then selects sentences with respect to both phonetic and prosodic criteria. To cover supra-sentential prosody phenomena, paragraphs were selected at random and recorded as well. The new speech corpus can be utilised in unit selection speech synthesis and also for training a data-driven prosodic parser.