VCV oto.ini generator (beta version)

D

Dojan

Momo's Minion
Pianodream link said:
Any news on this plugin? I'd like to know if you're planning on making a kana-supporting version ^^

Or at least if you're working on it... (Sorry for sounding like an ass)
Click to expand...

It does kana aliases already.
 
kirrinx

kirrinx

Teto's Territory
Posting this response for Tady ~

"I'm very sorry that I haven't replied to this thred in a while, but I've been having troubles to access UTAforum, so I'm asking someone else to post it for me. Regarding the program itself, as of now, I haven't got much progress on it, and still coudn't figure out a way to work with kana encoded files, but I'll try to research a bit more. When I get access back I'll answer better the questions that were made"
 
T

tady159

Ruko's Ruffians
Thread starter
Hi, and I'm very sorry that I couldn't respond any time before, but I've been having (actually, I still have) issues to access the forums, and just now I found a way that would allow me access it AND log in.

So... I'm gonna quickly answer some questions, and do a short report on the program itself:

1. The program does not run on XP, I'm already trying to fix this, re-writing the code on a Windows XP, but new issues appeared.

2. No, this program does NOT generate CV otos, and I will NOT make a program that generates one, cause this doesn't make sense. UNLESS that you only want aliases, but there are already programs that do this. A VCV oto generator is possible, because VCV uses multiple aliases for one file, which follows a logic. And VCV (normally) uses fixed parameters for all aliases, so that is how the program works. Since CV doesn't have those two characteristics I've just said, it is not possible to make a program that generates a CV oto (unless that something very complex was used, like intensity and formant variation recognition, to estimate parameters, but as mentioned previously on this thread, UTAU already has this feature).

Report:
I'm currently re-writing the source code of the program on Windows XP to ensure the compability with this OS. I'm still researching a way to make the program read kana encoded lists, but I still didn't have much progress on this part. I think I'll have to make two versions of the program (one for kana encoded lists and another one for romaji encoded lists), just to avoid recognition problems and some other issues...
I'm adding the offset value, as it was asked, and it is going to increase accordingly to a value given by the user.
Some complications on the code appeared, and I think that I will have to rewrite the program from scratch and make it less messy (this is mostly developer talk, and not so relevant to final users, I think).

At the moment, I have some other priorities, related to my college, and health, so I'm not going to work on the program soon, but I do plan on continuing this project and have it finished (I hope that still in this year).
 
T

tady159

Ruko's Ruffians
Thread starter
The VCV generator can't make CVVC otos, because it creates triphones combinations, while CVVC works with diphones. I believe that CVVC would be much harder to work with, because it mixes with CV, and there is a great variety of lists, which makes the prediction of how the program should generate the oto much harder.
 
L

Lightzaver

Teto's Territory
Okay, Thanks!

Good luck with this, I haven't actually had a chance to test but I think it should be "incorparated" into UTAU it'self !
 
Karma Drak-pa

Karma Drak-pa

Teto's Territory
I just would have a question : does this plugin work only for Japanese VCV or we can manually add other vowels for other languages ? If so, it would REALLY interest me.
 
NegiStick123

NegiStick123

Momo's Minion
Inputting something like 300.00 makes the program go crazy and makes weird otos, but when using regular 300. When otoing I use .00 on my parameters because it makes tracking my progress much easier as UTAU automatically gets rid of .00 when you edit it.

Insert the desired consonant value: 450.00
Insert the desired cutoff value: Insert the desired preutterance value: Insert the desired overlap value: oto.ini file generated successfully, press any key to exit
Click to expand...
As you can see the rest of the text gets weird. That is really how it looked. I never get the chance to input anything else besides when I did 450.00

oto.ini
ka_ka_ki_ka_ku_ke_ka.wav=- か,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=a か,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=a き,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=i か,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=a く,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=u け,0,300,10226768,1968630880,2686584
ka_ka_ki_ka_ku_ke_ka.wav=e か,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=- き,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=i き,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=i く,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=u か,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=a け,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=e き,0,300,10226768,1968630880,2686584
ki_ki_ku_ka_ke_ki_ke.wav=i け,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=- く,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=u く,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=u き,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=i こ,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=o か,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=a こ,0,300,10226768,1968630880,2686584
ku_ku_ki_ko_ka_ko_ki.wav=o き,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=- け,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=e け,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=e く,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=u こ,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=o け,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=e こ,0,300,10226768,1968630880,2686584
ke_ke_ku_ko_ke_ko_ko.wav=o こ,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=- ん,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=n き,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=i ん,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=n け,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=e ん,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=n こ,0,300,10226768,1968630880,2686584
n_ki_n_ke_n_ko_n.wav=o ん,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=- こ,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=o く,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=u ん,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=n か,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=a ん,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=n ん,0,300,10226768,1968630880,2686584
ko_ku_n_ka_n_n_ku.wav=n く,0,300,10226768,1968630880,2686584
 
T

tady159

Ruko's Ruffians
Thread starter
For now, this beta version only works with the five vowels (a,i,u,e,o) and th syllabic "n". However, I plan on making it more generic on a next version.

About using ".00" on the parameters: The program expects the user to input INTEGER numbers only for the parameters, and by adding ".00" at the end, it makes the program interpret them as floating-point numbers, and makes a bad convertion. I could change this on the next version.

Unfortunately, college is demanding much more time from me, and then I can't spend a lot of time on UTAU-related stuff. I'm also taking my time with this, because I'm waiting to see if I learn better ways to write the code for the program.

Ah, I'm also planning on giving it a Graphical User Interface as well. But this could take some more time, and depending on the programming language I choose, it'd require some additional software to be installed on the computer to be able to run the program.
 
Mugi

Mugi

Teto's Territory
Aahh yes that's right. A lot of my friends record VCV in romaji and right now I'm just imaging the mess of an oto setparam would generate for that.

Man, if you manage to make a fully functional oto generator, I would love to see it! I'm on board for this project. :wink:
 
T

tady159

Ruko's Ruffians
Thread starter
There's no point in making such generator, because usually all parameters are set with different values for each sample. And usually, one .wav file yields only one sample.
Unless you want an automatic aliaser. This one already exists, and I believe it was made by masao.
 
You must log in or register to reply here.
Top Bottom