From: [IDC]Dragon <idc-dragon_at_gmx.de>
Date: Sat, 3 Apr 2004 23:32:18 +0200 (MEST)

Hello Voiceboxers,

The existing voice implementation for simplicity had an internal drawback,
it consumed memory in the string buffer also for words which are only spoken,
like e.g. the numbers. This was the main reason for me not to implement
spelling, because the letters would have introduced a lot more of these words.

Now I've changed this (mildly pushed by Daniel, thanks for the genlang
part), voice only words are independent by having their own ID space and table.
Current drawback for the users is that I had to break compatibility and change
the voice file format. From tomorrow's build on Rockbox won't load the old
format, you have to update the voice file to maintain the function.

I have prepared 2 voice files in the new format here:
http://joerg.hohensohn.bei.t-online.de/archos/speech/voicefont/

It's Microsoft Mary and AT&T Crystal speaking, your choice. You'll have to
copy and rename the one you want to "english.voice" on your box. (Of course,
in the .rockbox/langs/ directory.)
The files are now both 32kHz sample rate, so they will also play on the
digital output. On the long run, I doubt we can keep this high rate, it's getting
pretty tight with space already (had to force a high VBR compression to fit
it into 1.5 MB).
The authoring tools there are also updated.

Sorry for this little hassle,
Jörg

-- 
+++ NEU bei GMX und erstmalig in Deutschland: TÜV-geprüfter Virenschutz +++
100% Virenerkennung nach Wildlist. Infos: http://www.gmx.net/virenschutz
_______________________________________________
http://cool.haxx.se/mailman/listinfo/rockbox

Received on 2004-04-03