This reverts commit 433a0a0455.

The issue here appears to be that the process the eSpeak commands are
processed from to generate the speech differs from the one used to
modify the settings. This appears to be enough to cause it to not
get the updated settings until the process is restarted.

Thanks to Peter Vagner for locating the cause.

Fix bug, some letter names spoken wrongly (eg. accented letters).
Recognize Arabic full-stop (U+06D4).
Changes to languages: en, pt, ka.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@299 d46cf337-b52f-0410-862d-fd96e6ae7743

Conflicts:
	dictsource/dict_phonemes
	phsource/compile_report

Language changes: en, pt, da, an, fa.


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@298 d46cf337-b52f-0410-862d-fd96e6ae7743

Conflicts:
	src/voices.cpp

Conflicts:
	dictsource/dict_phonemes
	phsource/compile_report

Added languages: fa (Farsi), an (Aragon).
Changes to languages: pt, si, ne.



git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@297 d46cf337-b52f-0410-862d-fd96e6ae7743

This fixes the issue where installing will re-generate
the voices and may sometimes fail setting up the
shadow directories in the espeak-data folder.

Conflicts:
	dictsource/dict_phonemes
	dictsource/sl_list
	dictsource/sl_rules
	phsource/compile_report
	src/synthdata.cpp

Changes to langauges: ta, sl, pt, en


git-svn-id: https://espeak.svn.sourceforge.net/svnroot/espeak/trunk@296 d46cf337-b52f-0410-862d-fd96e6ae7743