Real-time voice conversion software aimed at VTubers called "Voidol"

[uncreepy]

I am hoping that they make the Windows version's price the same as the Mac and iPhone versions. It's normally about $16 for the base software, but they had a campaign going where it was half off at $8.

Each voice appears to be a different price, I can't confirm them myself, because they are only buyable on iPhone (I don't have one). (The Mac version only has CRIMMZOH and Iroha so far, who are free.)

But based on old Twitter posts and some articles, I believe some of the voices are normally about $8 (Yoshida-kun, Miranda, Tsukasa) but were half off at $4 for the campaign. Some of the other voices are VERY expensive compared to those prices (Zunko and Tsubasa were shown to be $35 and for some unforsaken reason, Peroro is $29).

 

[Prism]

It kinda makes sense because the more expensive voice acting but it's dramatically higher.

 

[uncreepy]

Voidol posted some news today, but it's disappointing, so prepare yourself before continuing to read. (Kind of dramatic, but whatever... This is my thread.)

The sale for the Mac App Store has ended, so it's back to normal price now (about $18).
They said they continue to develop the Windows version with the goal of being done during April.


Last time, they said that they were planning on releasing the new voices on August 9th, so I wonder if they will follow through or not? Maybe only the Mac people will be able to buy it (that's okay, they've been waiting since the start of April... I can just stare longingly at them if they put them on Amazon). :{

Quote of day:

Give us the Windows version…_:(´ཀ`」 ∠):

 

[uncreepy]

COST: $10 (until last day of October) or free trial
The Windows version is now on sale! It normally about $20, but there is a half off campaign right now (so it's currently about $10 until the last day of October).

For people who want to try it out without paying, there is a free trial version. You can only use the 3 default voices (Otomiya Iroha, CRIMMZOH, and Kanade Minato) for about 5 minutes before it cuts you off. I tested it and could NOT get it to work any longer than this, you only seem to get ONE attempt to use it? (You have to click "Bypass" to change the voice, but I can't remember if I clicked it properly or not after closing and opening the program after the timer ran out, but I don't care because I bought the license already.)

The full version also lets you download > drag and drop "voice conversion models" (this is what they call the add-on voices), but you can not do this in the trial version. There is a warning in the guide that they assume you will try the trail version before purchasing, so they will not issue refunds if you're unhappy with the quality of the full version.

Where to buy + Amazon guide
You can purchase it for download only from Rakuten, Mac App Store, and the Japanese version of Amazon. (I will be talking about buying from Amazon, because I am getting it there.)

Here is the Amazon link, you can change the interface to be English (circled in green):


In order to purchase anything (including downloads) from Amazon Japan, you need to have a Japanese address linked to your account. In the past, I just Googled something like "generate Japanese address" to get one.

When you install Voidol, it detects your system language, so my installer and interface were in English automatically. In the menu of Voidol, if you go to Help, there is even an English language guide.

New add-on voices!
Several new voices were teased in the trailer below, it says they will be gradually releasing more voices for purchase.

Instead of only coming with Otomiya Iroha and CRIMMZOH, there is a new male voice for free: Kanade Minato.

New voices I spotted in the trailer:

Edit: I messed up and realized his name is Gate Jobs, not Get Jobs.

I am not sure if all of these "Characters" are real or fake. The bottom one had a talking sample play (noted by the rainbow markings under the name), so that one is probably real?

I tried to look up who "Gate Jobs" is, but I couldn't find anything.

I'm not sure how they will spell CRIMMsonna, so I put a question mark by it (great, who asked for another one of these guys).

I found out that Senkin Ichiru is a VTuber.
Here is them using Voidol (I'm not sure if it's the actual Ichiru voice or if it's Iroha or something):

They are actually a guy who sounds like this (I guess Ichiru is supposed to be their video game character. When they are her, she has a bow, but when it's him, she has no bow and her shirt is closed/with no bazongas.):

When they uploaded the Voidol test, they used it a lot up until about 3 months ago. But lately, they haven't been using Voidol.

Note: You need to contact Crimson Technology for permission to use the voices for commercial purposes.

Testing out Voidol:
I will be writing another post with audio samples later in the day. It sounds pretty rugged, but I'm commited to trying out Voidol. I will be testing it out with me talking in real time, audio clips from stuff, and trying out things like clips from CeVIO and Vocaloid.

 

[Prism]

Finally super excited

 

[uncreepy]

My recording setup:
I have been testing it out and it's actually kind of awkward to record. If anyone has any suggestions for how to properly record the audio, feel free to make suggestions. This is my setup:

Audio box (Behringer brand) + cardiod mic plugged in to it (see settings below)

Open both Voidol and Audacity, make sure Voidol is not on Mute.

I used these settings for Audacity ("What U Hear" might be called "Stereo Mix" on your computer, it's basically a way to record audio with a sound driver):


I messed around with the types of halls/rooms, space (makes it sound echoey), and noise gate. This is the amount that I personally thought sounded good. Feel free to suggest your settings if you find mine didn't work for you. I found that having too much echo ("space") sounds too well, outer spacey, but having 0 makes the errors stand out more, so having a little bit helps cover up the errors it outputs.


Place the mic about keyboard distance from you (NOT really close up to your face). Press record in Audacity, start talking in a relatively loud voice so Voidol can convert your voice. In my setup, I have the voice playing out loud and the audio being captured by the computer's built in sound driver (What U Hear / Stereo Mix). You'll have to practice a lot, because it's really trippy trying to talk normally while the Voidol voice is coming out (it's kind of distracting and makes you want to talk slower).

My thoughts:
Kanade Minato: His voice is the worst out of the 3 free voices. He is really quiet and seems to pronounce some phonemes strangely. He talks extremely quietly and sounds evil/unsettling (even though is picture is cute and youthful looking). I had trouble getting him to finish sentences, he just drops out volume-wise randomly. Maybe he wasn't developed long enough compared to the other two voices, which we have known about since April?

Otomiya Iroha: I was surprised, but I actually preferred her voice the most. Her voice actually sounds different depending on which model you choose. I'm a girl, so I tried all of the female ones.

Female 1 and 2 sound much more mature, almost like an announcer for trains or something. The "High Pitch Conversion" ones sound very high-pitched and "anime-like". Suprisingly, you can speak both English or Japanese with her (CRIMMZOH can as well, but it doesn't work good with Minato).

CRIMMZOH: He's alright. Better than Minato, but not as cute as Iroha. He's a little bit wonked up, but definitely no where near as much as Minato.

Sample clips:
I tried to play a VSQx out loud and hold the mic up to the speaker, but it made Voidol freak out. It just made it screech with fuzzy sounds, I could not get it to work at all. I haven't tried doing something like holding up a recording from my phone to the mic, might try that out later. So I will only comment on converting my voice in real time to the Voidol voices.

Please note that the official Zunko Twitter said that Voidol frankly needs more development (and that Zunko will be released for it in the future, which was estimated to be some time in September by Crimson Technology). But I imagine that the more people contribute monitarily to the project by purchasing Voidol, the faster they will be able to update the software and add more voices. I guess it would be good to think of this as being an early version of the software.

(I will upload some clips and share them in a while. I will either edit this post or make a new one to share my recordings.)

 

[Voidquestions]

Note: You need to contact Crimson Technology for permission to use the voices for commercial purposes.

You keep saying this but I don't see it anywhere on their website?

Is it not good enough to simply buy the software to use it for commercial purpose? If not, then what is the point of buying Voidol for 10$ (20$ later) if you can't use it for commercial purposes?

Is it even possible to form a partnership over-seas? They assume the person who bought this lives in Japan, because you need a Japanese address in Amazon JP.

How much would the license even cost? Because someone starting out obviously wouldn't have enough for expensive licenses. They must build a fanbase first, ya know?

 

[uncreepy]

You keep saying this but I don't see it anywhere on their website?

It's in the EULA <-- this is the link to the English version of it, there's a Japanese version on their site.

EULA:

If the User wish to use the Converted Voice in any of the following ways or purposes, it is necessary to make another license agreement separately. In that case, please Inquire to “Contact” attached to the bottom of EULA.

(1) In the case where individuals use the Converted Voice for commercial purposes.
(3) In the case where the Converted Voice is used commercially for any of the following usage, coupled with the Product name “Voidol” or each Character’s name or other similar expression ("Contract Representation").

(a) Selling products on which name(s) are credited as voice actor, singer, artist, musical instrument and the like, together with the Contract Representation.

(b) Selling products / services of which the Contract Representation are indicated, in all advertising activities such as packaging and promotional materials.

(c) Audiovisual works with Contract Representation shown in end-title roll or otherwise visible to the public.

(4) In the case where the User uses the Converted Voice commercially as telephone ringtones or as audio alerts or signals in telephone or telecommunication equipment, except when used in combination with other musical instruments or sounds within a musical composition.

(5) In the case where the User uses the Converted Voice as built-in voice or sound for consumer electronics merchandise, robots, car navigation system, gambling machines, musical instruments, computer software, game software, or any other hardware and software system.

(6) In the case where the User uses the Converted Voice in such a way that it can be taken as if the Character in commercial video works are singing or performing. (including but not limited to animations, commercial films, games etc.)

(7) In the case where the User uses the Converted Voice for purposes or forms that CRIMSON deems commercial use. (not limited to cases referred above.)

tl;dr: They just want you to contact them if you plan to use the converted voices for something you plan to sell. I haven't contacted them before, so I don't know if they simply want to make sure what you're monetizing is something they approve or or not, or if they want you to buy a character usage license or something.

It sounds like this is referring to "promoting" your commercialized product by saying "Hey world, I used VOIDOL and this is OTOMIYA IROHA singing in my song" (for example). Kind of like how Vocaloid does not want you to say "Vocaloid" in YouTube titles (you have to pay to use the "Vocaloid" trademark) or how you can't make wads of cash by using Miku's name/image for commercial purposes. Basically, using the company's product name (Vocaloid/Voidol) and the character name (Miku/Otomiya Iroha) helps people find your song/animation/game/whatever, people wouldn't be able to find your stuff as easily if you weren't promoting your work by using their character, so some companies want a character usage fee.

I do not know if Voidol is okay with you using the voices and just creating your own original character with its own name (ex: Tina the Virtual Idol who looks like a pink haired girl instead of being named Iroha and looking like Iroha). Some people make OCs in that way so they can publish Vocaloid CDs with a different character on the cover and don't mention it being sung by a certain Vocaloid.

Hopefully what I wrote makes sense. I'm just reporting all that I know about Voidol currently, I'm not a representative or something, so I don't know the answers to your questions. I would have to contact them about the questions in order to get proper answers. But I don't have an interest in the answers right now, because I am not planning to monetize anything with Voidol (I'm not good at making music and I'm not a VTuber, and spending $10 to mess around with software seems reasonable enough to me).

 

[Voidquestions]

It's in the EULA <-- this is the link to the English version of it, there's a Japanese version on their site.

Thank you for the clarification. Looking into it, I found this. Specifically section 4. Apparently there is a fine line between Crimson voices and voices from other companies.

 

[mobius017]

Sample clips:
I tried to play a VSQx out loud and hold the mic up to the speaker, but it made Voidol freak out. It just made it screech with fuzzy sounds, I could not get it to work at all. I haven't tried doing something like holding up a recording from my phone to the mic, might try that out later. So I will only comment on converting my voice in real time to the Voidol voices.

You might have better luck if you can find a way to do this by cable, rather than holding a speaker up to a microphone. Speaker-microphone definitely can work, but if they're too close, you can get electromagnetic interference (buzzing/humming/squealing/etc.).

Hypothetically, if you could connect a double-ended headphone cable (cable with one of the 1/4" headphone jack ends on each end) from the headphone jack on your speakers to something like a Line-In port on your computer's audio card, and then set Voidol's input source to the Line-In port, you could get the audio into Voidol via cable. (Possibly, setting Voidol's input source could be accomplished with software your computer came with to manage the audio card, too, if Voidol doesn't have a setting like that. At least that was how it was on a PC I used awhile back. But whether or not something like that is available is probably going to vary a lot between different computers/audio cards.) I think I've even tried going from the speakers directly into the microphone jack...and I think it worked...though it sounds risky, so I'd definitely recommend keeping any volume low at first if you try that.

 

[uncreepy]

I did some speaking tests with the 3 voices. I used each option for each voice. I wish that Voidol didn't have crackling and wasn't so echoey, it seems like other people have the same problem, I'm not sure if there's a fix for it or not. (I think the official demos seem to have less crackling, but I'm not sure if it's due to them having access to professional hardware, or if they edited it after or not. But I think my results are the "normal" results people should expect to get.)

Japanese tests:
They are all reading the lines
"......Eh? "Mitsuketa" tte...... Ano, watashi no koto o shitte-iru n desu ka?"

CRIMMZOH: Female 1 > Female 2 > Male 1 > Male 2


Otomiya Iroha: Female 1 > Female 1 (High Pitch Conversion) > Female 2 > Female 2 (High Pitch Conversion) > Male 1 > Male 2


Kanade Minato: Female 1 > Female 1 (High Pitch Conversion) > Female 2 > Female 2 (High Pitch Conversion) > Male 1 > Male 2


=====================
English tests:
They are all reading the lines
"......Huh? What does he mean by "found me"......? Um, do you know about me?"

CRIMMZOH: Female 1 > Female 2 > Male 1 > Male 2


Otomiya Iroha: Female 1 > Female 1 (High Pitch Conversion) > Female 2 > Female 2 (High Pitch Conversion) > Male 1 > Male 2


Kanade Minato: Female 1 > Female 1 (High Pitch Conversion) > Female 2 > Female 2 (High Pitch Conversion) > Male 1 > Male 2
I had problems with him sounding demonic in Japanese. He's more normal-sounding in English, though.


=====================
I will upload some singing tests, but I'm not sure if I have time today for it or not. Here's a sample song with Iroha, though:


If anyone tries out Voidol, please feel free to share your tests here.

 

[uncreepy]

Okay, as a "control" in the Voidol experiment, I added a clip of me saying the lines in JP and then ENG.

Sorry my voice is a bit weird (maybe the random growlyness affects Voidol, but I have a hard time not talking with it). But I want to show you that I don't think my mic sounds crackly before getting converted through Voidol.

 

[Wario94]

I have an idea: what would happen if you could use actual Vocaloid voicebanks via voidol?

 

[uncreepy]

I guess you kind of already can if you think Tohoku Zunko counts, I think they just ported her voice files in to Voidol. (Can't imagine they bothered to re-record.)

Not that being able to use every Vocaloid ever in Voidol would improve the quality it outputs. XD;

 

[Wario94]

@uncreepy Actually, I meant you could use your own Vocaloid voicebanks as the test for voidol.

 

[uncreepy]

I found out "Bohemian voice" has an icon that is a caricature of Freddie Mercury (click time-stamped video to see):

He sings the song at about 0:56. (Kind of an unfair demo because the Queen cover band singer is singing at the same time as the Bohemian voice, so obviously it will sound good.)

I like the taped on "Soon" on the red poster covering up the availability. It probably said "Now" or something.

 

[uncreepy]

Some tests of Voidol by people other than me so you can get a better idea of the quality to expect.





^ Converted voice, then his original voice, then both overlapped.

 

[uncreepy]

Crimson Technology will be in Shanghai for an event called "Music China" (relating to musical instruments). Not sure if it will be all of the days or not (10th-13th).


The link in the tweet is supposed to show where they will be at the Shanghai New International Expo Centre (SNIEC), but the link leads to a 404. I checked their news section and it contains no articles relating to the event and hasn't been updated regarding Voidol since August 30th.

Link to the venue: https://music-china.hk.messefrankfurt.com/shanghai/en.html
(I couldn't find a map of the booths, but they have schedules for special events listed.)

This must be the event they mentioned way back on post #8 about wanting to exhibit in China so they could have characters to convert Chinese language.

The add-on voices were estimated to be released in September, but they are, yet again, late. It seems kind of lame to exhibit software that still doesn't have the add-on voices. It's so annoying that iPhone users can get the add-on voices, but the rest of us have been waiting since May. I wonder if it could have something to do with the bad reviews they received on Amazon JP (basically 2/5 stars).

 

[Wario94]

The add-on voices were estimated to be released in September, but they are, yet again, late. It seems kind of lame to exhibit software that still doesn't have the add-on voices. It's so annoying that iPhone users can get the add-on voices, but the rest of us have been waiting since May. I wonder if it could have something to do with the bad reviews they received on Amazon JP (basically 2/5 stars).

Since you mentioned it, I did made a research for "Voidol had the bad reviews on Amazon JP" and yeah, you were right about that! Here's proof:
5 stars: 11%
4 stars: 5%
3 stars: 26%
2 stars: 0%
1 star: 58%
Yeah. I believe that Voidol had the most dislikes than likes, many of which that said the PC version had bugs.
Because of that, do you believe that you're have to wait until the new Piapro Studio (which includes Cherry Pie) come out in 2020?

 

[mobius017]

Because of that, do you believe that you're have to wait until the new Piapro Studio (which includes Cherry Pie) come out in 2020?

I suppose that would be more a matter of opinion--each piece of software is probably going to have features or voices that someone might prefer, or they might want to use both but for different things.