Tensei Koe Uranai: Beta real-time voice conversion with 13 voices

[uncreepy]

It's awkward to translate, but the software's name is literally "Reincarnation Voice Fortune Telling". It's aimed at VTubers who want to speak in the voice they always wanted to match their avatars.

Loose translation of their shtick:

I wonder why I'm living this kind of life?
I wanted to be born a boy! I wanted to be born a girl!
I think you've probably had many worries about things you can't choose.

If you could become the "you you want to be", what would you do?
What would you say to someone?

GREE VR Studio Lab developed this, they focus on VR games and technology. Tensei Koe Uranai uses WORLD, a voice analysis synthesis (vocoder) by M.Morise that seems to be able to be downloaded and used by anyone.

It's in beta right now and I don't know much of the details. But you can use the website to select a voice, talk with your mic, and upload the vocoded voice so it creates a sharable link to show other people. ALLEGEDLY, it said you can speak any language with this and used #English in the tweet that said so. (I haven't confirmed this, will test later.)

There are 13 voices to chose from.

Warning: The order of characters changes every time you visit the page.

Edit: Crud, I spelled "kawaii" wrong on the pic, might fix later.

There are hyperlinks that leads to a sample of each characters voice from the official twitter, so be sure to click each name to hear. (No example of gorilla or onii-san yet.)
Dark voice, soprano, dual voice type (not sure what this means)
Kawabo (short for "kawaii voice"), madame (yes, this voice sounds masculine, the link isn't wrong), monsieur
Elementary student, damibo (short for "dami-goe / coarse voice"), dandy
Onee-san (big sister), male middle schooler, onii-san (big brother)
Gorilla (It says that this is for virtual primates, it must be referencing "Virtual Gorilla", I think the avatars seem to have been made in VRoid, so that's why this is a human avatar instead of a gorilla?)

I THINK what this site is doing is getting people to upload their audio clips so that the algorithm can learn from their sentences. When you go to test out a voice, it suggest a line (serif) for you to read but leaves the name blank so you can put whatever you want in there.

I don't have time right now, but I want to test out the voices for myself and post more about it on here. I also have to check the usage agreement.

I do not know if the software will cost money or be free in the future after it's out of beta (have to research more). However, I am suprised and this came out of nowhere. Another real time voice converter aimed at VTubers. It seems the competition is heating up.

 

[mobius017]

Gorilla (It says that this is for virtual primates, it must be referencing "Virtual Gorilla", I think the avatars seem to have been made in VRoid, so that's why this is a human avatar instead of a gorilla?)

Ahh.... Alrighty then. That would make logical sense, then (I assume VRoid doesn't have non-human avatars, or at least gorilla ones.). They should still probably change that pic at some point....

...

(But humans are classified as primates.... So all the pics/voices were intended for virtual primates....)

...

Nevermind my irrelevant ramblings. Continue please.

 

[akihiko]

Hi,
Thanks to find Koeuranai. Now English version is ready.
「転声こえうらない」
This is experimental service, to find user needs and usages.
Avatars are comes from REALITY.
REALITY｜VTuberと未体験の交流ができるライブ配信アプリ
Ask me if you have any question.
or official Twitter

 

[frankensalad]

Thanks for stopping by. I started playing around with Koeuranai a bit. It's a lot of fun.