A SELF-LEARNING SPEECH SYNTHESIS SYSTEM
C.S. Blackburn and S.J. Young
March 1996
We describe a self-organising pseudo-articulatory speech production model (SPM), and present recent results when training the system on an X-ray microbeam database. The SPM extracts statistics describing articulator positions and curvatures during the production of continuous speech, then applies an explicit co-articulation model to generate synthetic articulator trajectories corresponding to time-aligned phonemic strings. A set of artificial neural networks estimates parameterised speech vectors from the synthetic articulator traces. We present an analysis of the articulatory information in the X-ray microbeam database used, and demonstrate the improvements in articulatory and acoustic modelling accuracy obtained using our co-articulation system.
If you have difficulty viewing files that end '.gz',
which are gzip compressed, then you may be able to find
tools to uncompress them at the gzip
web site.
If you have difficulty viewing files that are in PostScript, (ending
'.ps' or '.ps.gz'), then you may be able to
find tools to view them at
the gsview
web site.
We have attempted to provide automatically generated PDF copies of documents for which only PostScript versions have previously been available. These are clearly marked in the database - due to the nature of the automatic conversion process, they are likely to be badly aliased when viewed at default resolution on screen by acroread.