
Voice inventories, which are grouped in subdirectories of this directory,
are distributed separately as their development is fully independent
of the development of Epos itself: they differ from Epos and from each other
in authorship, licensing conditions, testing methodology, size and some
special features.

There is a single exception, the kubec-vq and vichova-vq voices.
It is possible to include them because they are licensed under GPL by their
copyright owner (Institute for Radio Engineering and Electronics of the
Czech Academy of Sciences), and because they are very small
(25 kB uncompressed each).
These voices can be used by the impatient to hear some speech before
downloading any better voices, but there is the expectable relation
between size and quality.

Additional voices are available to the on-line users of Epos without
being downloaded, namely the "remote" voices which use a public
TTSCP server to do the synthesis and therefore need no local speech
inventories.  This is useful when trying out Epos at an on-line
computer.

Questions concerning the availability of specific high quality
inventories should be sent to epos@speech.cz.

Please refer to */README* files concerning the licensing conditions
of files in subdirectories of this directory.


