Make the TTS a better place

Instead of having shitload of junk space, you should first record all the syllables needed and configure them. Make a program that looks at the configuration file and chops off pieces of the audio in there and puts them in a different directory. The files that will be made will be the space in the middle of left and right offset(wherever it is specified in the oto). It will also glue them together in such a way that there's no space between the samples and will add the place of every file in the sound file itself as metadata(if the audio goes: fa, fe, fo, of, fu the metadata part will go: 1(leftOffset,rightOffset);2(leftOffset,rightOffset);3(leftOffset,rightOffset);4(leftOffset,rightOffset);5(leftOffset,rightOffset);). The right/left offset parts aren't going to be copied from the oto, they will be telling at how many milliseconds which sound starts and ends. It's going to make it's own configuration file by looking at the oto.ini. What it will do is it will rewrite all the lines as the samples go in the audio file it just made(it's all in the order in which it was in the oto.ini(meaning both the samples and configuration) no matter which order is in the directory). So, if we have in the oto:
fa.wav=alias,leftOffset,consonantField,rightOffset,preutterance,overlap
fe.wav=alias,leftOffset,consonantField,rightOffset,preutterance,overlap
fof.wav=alias,leftOffset,consonantField,rightOffset,preutterance,overlap
fof.wav=alias,leftOffset,consonantField,rightOffset,preutterance,overlap
fu.wav=alias,leftOffset,consonantField,rightOffset,preutterance,overlap
in the configuration file for YARE it will be a little bit different(note: It will look at the aliases and name the files in the audio file(as metadata) after them so if you don't want them to be kana make sure you have the right aliases before doing this.):
1(fa,alias,consonantField,preutterance,overlap)
2(fe,alias,consonantField,preutterance,overlap)
3(fo,alias,consonantField,preutterance,overlap)
4(of,alias,consonantField,preutterance,overlap)
5(fu,alias,consonantField,preutterance,overlap)
The last 3 values will be looked at as parameters, all the ones before them are going to be looked at as aliases.
Where label(first value) is the name of the sound and alias is the additional alias for the same set of settings to save time and space. The label fields aren't limited in numbers. For all I care you can make a full voicebank out of one sample trough label fields. Something that's not doable for other fields is giving them anything other than numbers. The reason it will ask where to look (WAVE files' name or alias in the oto.ini) is because there are stringed voicebanks and they usually just give you grief in the end. It will also look where each configuration line in the oto is in the file and if two use the same space, it will not copy the space that's used twice, just once.