Idea: language plugins

[Bohdan R . Rau]

Hi all

What do you think about it?

Let's assume there is a library with only one public function:

char *ortographic(
        const char *client, /* client name */
        const char *module, /* module name */
        const char *string);

For each supported language we should have separated library (like
libspdplugin_pl.so for Polish). Library is dynamically linked at first
attempt to say something in particular language.

What is it for?

Function should create plain ortographic text from give string. Examples
for Polish language:

Words like "dd:dd" looks like hour:minute. So we should pronounce
"20:07" as "dwudziesta zero siedem". But within sentence "o godzinie
20:07" it should be pronounced as "dwudziestej zero siedem".

Roman number "XIX" is pronounced as "dziewi?tna?cie". But within
sentence "w XIX wieku" (or even "w XIX w.") it should be pronounced as
"dziewi?tnastym", within "od XIX wieku" as "dziewi?tnastego" etc.

This library would also use user-defined pronunciation vocabulary.

I think it is not possible in language-independent speech-dispatcher,
nor in multilanguage synthesizer (like espeak).

Module name (may be null) allows to use synthesizer-dependent features
(for example letter separators and stress markers in Ivona).

Client name allows to use different rules (defined in user's
configuration file) for particular clients.

Of course the library must be written by native speaker :)

IMHO this plugin would improve speech quality...

...or not?

Bohdan