Development of rules for controlling the HLsyn speech synthesizer

In this paper we describe the development of rules to drive a quasi-articulatory speech synthesizer, HLsyn. HLsyn has 13 parameters, which are mapped to the parameters of a formant synthesizer. Its small number of parameters combined with the computational simplicity of a formant synthesizer make it a good basis for a text-to-speech system. An overview of the rule-driven system, called VHLsyn, is presented. The system assumes a phonetic string as input, and produces HLsyn parameter tracks as output. These parameter tracks are then used by HLsyn to produce synthesized speech. Recent work to improve the synthesis of consonants and suprasegmental effects is described, and is shown to improve the quality of the output speech. The improvements include refinement of release characteristics of stop consonants, methods for control of vocal-fold parameters for voiced and voiceless consonants, and rules for timing and intonation.