espeak, mbrola, utf8, navit : problem with french accentuated letters

[Xavier Cremaschi]

Hi,

I am trying to use espeak+mbrola and navit to obtain a french-speaking GPS.

At the end I will need navit to produce french sentences 
(navit-locale-fr is not installed because I don't use 0.0.4-r1 but 
svn-1865), but for the moment espeak is the problem.


But when I run espeak -v fr napoleon.txt (napoleon.txt being an UTF-8 
Unicode text file containing the first paragraph of
http://fr.wikipedia.org/wiki/Napol%C3%A9on_Ier#La_formation_militaire ) 
  espeak tries to spell accentuated letters.
For example for the word "représentant" I hear :
     repr-e accent aigu-sentant (repr-e with accent-sentant)


If I do on my PC with Debian and espeak 1.37
     espeak -v mb-fr1 -f napoleon.txt
I can see things like :
e	56	 0 106 80 102 100 102
p	75

but on my FreeRunner with Om2008.12 and espeak 1.37 I see
e	81	 0 103 80 96 100 96
@	87	 0 96 80 92 100 92
_	1
a	47	 0 99 80 96 100 96
k	92
s	100
a~	59	 0 99 80 96 100 96
t	85
E	63	 0 99 80 96 100 96
g	65
y	148	 0 103 80 96 100 96
_	1
p	75


this "aksatEgy" is the "accent aigu" in "repr-e accent aigu-sentant"

napoleon.txt seems :
- ok while doing 'cat' on it (using ssh in Konsole - kde virtual terminal)
- '..' instead of accent in vi (using ssh in Konsole)
- some accents ok, some not in vim (using ssh in Konsole)
- no accents in openmoko-terminal2


What can I do to have a full utf8 environment able to support french 
language ?


Xavier Cremaschi.

PS :
$echo $LANG
fr_FR.UTF-8