Synthesis by generation and concatenation of multi-form segments

Abstract

A speech synthesis system and method is described. A speech segment database references speech segments having various different speech representational structures. A speech segment selector selects from the speech segment database a sequence of speech segment candidates corresponding to a target text. A speech segment sequencer generates from the speech segment candidates sequenced speech segments corresponding to the target text. A speech segment synthesizer combines the selected sequenced speech segments to produce a synthesized speech signal output corresponding to the target text.

Description

[0001] This patent claims priority to US Patent Application 11 / 838,609, filed August 14, 2007, which is hereby incorporated by reference. technical field [0002] The present invention relates to speech synthesis. Background technique [0003] Machine-generated speech can be produced in many different ways and used in many different applications, but there are currently two basic methods for synthesizing speech signals that are widely used. One approach attempts to use models to construct speech signals, while the other concatenates pre-stored speech segments. Model-based approaches tend to be efficient in terms of storage and flexibility, but produce rather unnatural-sounding speech. An example of model-based speech synthesis is e.g. "Simultaneous Modeling Of Spectrum, Pitch And Duration In HMM-Based Speech Synthesis", which is incorporated herein by reference. [0004] Another method of speech synthesis - segment concatenation is capable of producing very natural speech...

AI Technical Summary

Problems solved by technology

usually provide good segment quality volume, but may suffer from poor rhythm (prosody)

Consistent voice quality, but has an unnatural "handling The" feature

Images