A parametric monophone speech synthesis system

[摘要] Speech is the primary and most natural means of communication between human beings.With the rapid spread of technology across the globe and the increased number of personaland public applications for digital equipment in recent years, the need for human/machineinteraction has increased dramatically. Synthetic speech is audible speech produced by amachine automatically. A text-to-speech (TTS) system is one that converts bodies of textinto digital speech signals which can be heard and understood by a person.Current TTS systems generally require large annotated speech corpora in the languagesfor which they are developed. For many languages these resources are not available. In theirabsence, a TTS system generates synthetic speech by means of mathematical algorithmsconstrained by certain rules.This thesis describes the design and implementation of a rule-based speech generationalgorithm for use in a TTS system. The system allows the type, emphasis, pitch and otherparameters associated with a sound and its particular mode of articulation to be specified.However, no attempt is made to model prosodic and other higher-level information. Instead,this is assumed known. The algorithm uses linear predictive (LP) models of monophonespeech units, which greatly reduces the amount of data required for development in a newlanguage. A novel approach to the interpolation of monophone speech units is presentedto allow realistic transitions between monophone units. Additionally, novel algorithms forestimation and modelling of the harmonic and stochastic content of an excitation signal arepresented. This is used to determine the amount of voiced and unvoiced energy present inindividual speech sounds.Promising results were obtained when evaluating the developed system's South AfricanEnglish speech output using two widely used speech intelligibility tests, namely the modifiedrhyme test (MRT) and semantically unpredictable sentences (SUS).

[发布机构] Stellenbosch University