java sound

[nikkka]

i want to write a program which will transfer text into speech. i suppose, i'll need to record mp3/wav files and connect them to file, which will activate them one by one. all i ask, is some literature/documentation related with this subject.
kind regards,
nick

[JosAH]

i want to write a program which will transfer text into speech. i suppose, i'll need to record mp3/wav files and connect them to file, which will activate them one by one. all i ask, is some literature/documentation related with this subject.
kind regards,
nick

For starters check the FreeTTS package; it works fine with the default voice (named 'Kevin'). If you can manage to get the 'mbrola' voices working, drop us a note. It can process ordinary text Strings.

kind regards,

Jos

[nikkka]

check the FreeTTS package

as i can see, it works only for english words. i want to write a new one, for georgian language, so i will need to do everything from scratch :S

[JosAH]

as i can see, it works only for english words. i want to write a new one, for georgian language, so i will need to do everything from scratch :S

Yep, FreeTTS can handle English (American English) only as far as I know. There's another package, I forgot the exact name, that contains the name 'Mary'. It can handle quite a few languages; I don't know about Georgian. As always Google is your friend. Writing a TTS package from scratch is no option.

kind regards,

Jos

[nikkka]

Writing a TTS package from scratch is no option.

in fact, that is exactly what i want to do. do you mean it's too hard? again, i dont need anything complicated like changing intonation and stuff like this. just turning on/off sound files according to given character :S

[JosAH]

in fact, that is exactly what i want to do. do you mean it's too hard? again, i dont need anything complicated like changing intonation and stuff like this. just turning on/off sound files according to given character :S

Your final remark tells that it'd be too hard for you because speech isn't really character based; it works on phonemes and the parsing from text to the generation of a sequence of phonemes is really context sensitive (depending on the language parsed). Change of pitch (frequency) and intonation is the easier part of the pie here. I'd try that <mumble>Mary package if I were you.

kind regards,

Jos

[nikkka]

Your final remark tells that it'd be too hard for you because speech isn't really character based; it works on phonemes and the parsing from text to the generation of a sequence of phonemes is really context sensitive (depending on the language parsed). Change of pitch (frequency) and intonation is the easier part of the pie here. I'd try that <mumble>Mary package if I were you.

what's great about georgian language, is that you read it EXACTLY as you spell it. there are no combinations of characters that sound differently (like English "oo" or "th"). in every word, every character sounds the way it sounds in the other. is it still hard? :S

[JosAH]

what's great about georgian language, is that you read it EXACTLY as you spell it. there are no combinations of characters that sound differently (like English "oo" or "th"). in every word, every character sounds the way it sounds in the other. is it still hard? :S

So Georgian has the phonemic orthography property? I didn't know that. You could try to find sound files corresponding to each letter and 'glue' them together as you process the text but I won't hold my breath ...

kind regards,

Jos

[nikkka]

Georgian has the phonemic orthography property?

Yes, one of the very few such languages. So, it's much simpler than English.

You could try to find sound files corresponding to each letter and 'glue' them together as you process the text

That's exactly what I wanted to do. Please, would you be so kind to advice me some literature/documentation?

[JosAH]

Yes, one of the very few such languages. So, it's much simpler than English.

That's exactly what I wanted to do. Please, would you be so kind to advice me some literature/documentation?

I don't know if it can be done, but I'd check a local sound studio (national radio?) and ask if they have those 33 small sound files available. Next I'd start to play a bit with them to check whether or not the approach is feasible ... For literature, all I can recommend is Google because I've never played with language like that before ...

kind regards,

Jos

[nikkka]

I don't know if it can be done, but I'd check a local sound studio (national radio?) and ask if they have those 33 small sound files available. Next I'd start to play a bit with them to check whether or not the approach is feasible ... For literature, all I can recommend is Google because I've never played with language like that before ...

Anyway, thank you :)
And hey, how do you know that there are 33 letters in Georgian alphabet? :D

[JosAH]

Anyway, thank you :)
And hey, how do you know that there are 33 letters in Georgian alphabet? :D

Google; in the Dutch language there are 26 letters in the alphabet but we use 27; originally we have the y with dots on top of it; and we didn't have the y without the dots. The one with the dots originally was a 'diphtongue' that came from 'ii' (pronounced as 'eeee!' in English) but just because the Dutch steal every word on the planet we had to use the y (without the dots) as well. Until shortly we also used the e with dots on top of it, but not anymore. Russian (if I'm not mistaken) also has 33 different letters.

kind regards,

Jos