Neural-network-based F0 text-to-speech synthesiser for Mandarin

Shaw-Hwa Hwang*, Sin-Horng Chen

*Corresponding author for this work

Research output: Contribution to journalArticle

10 Scopus citations

Abstract

A neural-network-based approach to synthesising F0 information for Mandarin text-to-speech is discussed. The basic idea is to use neural networks to model the relationship between linguistic features, extracted from input text and parameters representing the pitch contour of syllables. Two MLPs are used to separately synthesise the mean and shape of pitch contour, using different linguistic features. A large set of utterances is employed to train these MLPs using the well known back-propagation algorithm. Pronunciation rules for generating F0 information are automatically learned and implicitly memorised by the MLPs. In the synthesis, parameters representing the mean and shape of the pitch contour of each syllable are generated using linguistic features extracted from the given input text. Simulation results confirmed that this is a promising approach for F0 synthesis. The resulting synthesised pitch contours of syllables match well with their original counterparts. Average root mean square errors of 0.94 ms/frame and 1.00 ms/frame were achieved.

Original languageEnglish
Pages (from-to)384-390
Number of pages7
JournalIEE Proceedings: Vision, Image and Signal Processing
Volume141
Issue number6
DOIs
StatePublished - 1 Dec 1994

Fingerprint Dive into the research topics of 'Neural-network-based F0 text-to-speech synthesiser for Mandarin'. Together they form a unique fingerprint.

  • Cite this