eSpeak: speech synthesis

Hello!

Ralf Herzog asked me here http://www.voxforge.org/home/forums/message-boards/acoustic-model-discussions/creating-new-model-with-the-use-of-germany-voxforge-model the following question "which eSpeak phonemes correspond with which Polish IPA phonemes". I think I don't quite get it. I found this http://espeak.sourceforge.net/phontab.html (by typing "eSpeak phonemes" on google) where I see an example of defining one phoneme aI. If I'd like to say which Polish IPA phonemes correspond to which eSpeak phonenes, I need the list of all eSpeak phonemes. On the page with Polish IPA phonemes (http://en.wikipedia.org/wiki/Wikipedia:IPAforPolish) there is list of all relations between written letter and spoken sound. So I guess just by comparing it with English (and other languages which are available in eSpeak project) phonemes it can be possible to find all (or almost all) phonemes. So for example if Polish 'b' letter is the same as 'b' in English word 'bike', there is only need to copy those numerical values (like here for aI http://espeak.sourceforge.net/phontab.html) from English language to Polish language. The same for all other sounds which are common for both Polish and English. And if there are some sounds which are present in Polish but not used in English, I guess they can be found in other, already existing languages in eSpeak, e.g. Italian, Spanish or others. So what exactly does he expect me to do by writing "which eSpeak phonemes correspond with which Polish IPA phonemes"? I couldn't send private message so I decided to post it here.

Greetings :-)!

I'm sorry I didn't reply to your question earlier. I didn't receive an email. Perhaps I became un-subscribed to this forum.

eSpeak has a common set of phonemes which different languages can use. Also additional phonemes can be defined for a language.

To find a list of which phoneme names the eSpeak Polish voice uses, look at the text file: dictsource/dict_phonemes. Find the section "dictionary pl_dict".

You can find which phonemes eSpeak uses for a Polish word by using the command: espeak -x -v pl "test"

and replace "test" with your word.

You can see the rules which eSpeak uses to translate Polish text into phonemes by looking at the text file: dictsource/pl_rules

The format of this *_rules files is described in http://espeak.sf.net/dictionary.html