Pronunciation issue with ibmtts

[Tomas Cerha]

Olivier BERT wrote:
> I'm experiencing a small issue while using ibmtts with speech diispatcher. 
> It is not specific to the use of Speech Dispatcher, but I think it will never 
> be fixed in the ibmtts engine itself. 
> 
> In french, some punctuation characters are preceded by a 'non-break space' 
> character (0xc2 0xa0 in utf-8). When Speech Dispatcher processes a text
> string for ibmtts, it converts it from utf-8 to cp1252 (I think) because
> ibmtts doesn't understand utf-8. After this conversion step, non-break 
> spaces are replaced with the code 0xa0. I finally come to the problem: 
> when ibmtts encounters a word followed by this character (0xa0), it spells
> this word instead of pronuncing it normally. 
> 
> I'm currently looking for a way to solve this issue in a clean way. 
> Is there a way to replace every non-break space by normal space in the 
> Speech Dispatcher code ?

Hi Olivier,

yes I think your direction is the right one.  As far as I remember, the
ibmtts module already does some conversions to work around bugs in the
engine.

Best regards, Tomas