Music composers, no more need to find a singer for your songs anymore?

[hima]

VOCALOID
http://www.tirkx.com/VOCALOID_2_CV_01.png[/img]

http://img.doppotan.com/thumbs/cqs1190265127k.jpg[/img]

Vocaloid is a program that allow you to create a singer voice for your music. The artist voice are created by Voice Synthesizing. All you have to is to put in the music note, and type the Lyric. And the the program will synthesize the voice according to the note and lyric you put in.

The program can be used with a Midi keyboard and Midi files. Currently there are two versions available. With different advantages and disadvantages.
1.) First version do better job at English lyric than Japanese. Also, the interface is in English.
2.) Second version improve the quality with Japanese lyric, but become more difficult to work with English lyric. And the interface is changed to become Japanese.

The voice is the character's above voice, named Hatsune Miku. The seiyuu or voice actress for Hatsune Miku is Fujita Saki.

Here are some example of songs done by Vocaloid.

http://www.youtube.com/watch?v=Clakc7ICEIo

http://www.youtube.com/watch?v=y4FwNdnwFTc
--------------------------------------------------------------------

So what do you think? Do you think this will be good for indy music composer who can't find a singer for them? Or do you think finding an actual singer is better? To me, I'm surprised by how good it sound. This is probably due to how Japanese language work. I don't think the same method can be used with my native language, which is Thai And I think it'll be good for those who want to practice creating music without having to pay too much.

Anyway, I think this is very interesting and something I should keep my eyes on. Seeing that the cover said Character Vocal Series 01, that means there must be more characters to come. One funny thing is no matter how dirty or ridiculous your lyric may means or sounds, she's still gonna sing it like nothing strange happen...which is kinda funny to me.

 

[Ceroscuro]

Sounds good for a synthesizer, but no where CLOSE to real.

 

[Reives]

That is rather surprisingly nice. I think it has some to do with the fact that it's sang in Japanese and hence I can't recognize any weird accents and etc., though. But it's definitely a handy tool. o: In terms of actual sound-wise, a real singer is still obviously better, for the freedom of expression. But this is definitely awesome for indie composers to get things mapped out.

 

[coloratura]

I'm with Reives! It does sound actually very nice, but being a trained singer myself, expression is so very important. (Probably less so in J-POP and and American pop though!) I don't know anything about Japanese, but I know in English, syllabic stress is very important to making beautiful music... because without good syllabic stress, a song will sound very robotic and lack fluidity. This program seems to be very flat in syllabic stress. (In English, for example, if you sing the word "beautiful", you would emphasize the "beau" and back of gently on the "ti" and "ful" to get a natural sound. Because when you talk, you don't say "BEAU-TI-FUL", you say "BEAU-tiful". :3 )

A real singer can do things a synth can't - and in music, where emotion and expression are the most important, you need a human to do it. Songs are so much prettier when you have dynamics, word painting (like for example, in a song today I sang, I had to sing the word "tattered" - my voice teacher said to do very strong, aspirated t's to make the word sound like it's meaning!), emotion... all that good stuff that makes music an art. :]

I'm sure it's a great tool for those who can't find a singer though! :] But being a singer myself, I'm obviously leaning towards "real singer" land!

 

[arcthemonkey]

It's also worth noting that Japanese has a relatively shallow pool of sounds to draw from. Outside of the normal alphabet, there are a surprisingly small number of extra sounds, and pronunciation of Japanese words is pretty easy besides, so making a sample base becomes a lot easier than other languages. I couldn't think of a better language for something like this.

I dunno how Tai works, though. I really doubt it would work in English (although, AT&T has a pretty cool TTS thing going on right now - it doesn't sing, though.)

Despite all this, I don't think having a fake singer is going to do you a whole lot of good on the credibility side of things.

Being able to blend syllables is important, too. There's a lot more to singing than just reading the words in pitch, as coloratu said.

Edit: Actually, it seems that either it can sorta do english, or there's an english version...

 

[hima]

Thai has a lot of weird sounds and I think it would be more difficult to do We also have a symbol mark on the alphabet so that you know the word's pitch. Same word with different pitch mean different things.

I agree with you that there's a lot more to singing than just reading the words in pitch and something that computer can't do,at least for now, is expression and emotion. I listen to Miku's version of "Tori No Uta" and it's nothing compare to the original, which is more powerful and more touching. She's probably suck singing Utada Hikaru's songs too. This program would work well with cute and pop song that require cute character's vocal, I guess. The creators are probably aware of this so I'll have to see what will they present us in the future version

 

[Diedrupo]

There actually is an English version, though I don't recall the name. It isn't as good as the Japanese one obviously.

But yeah I saw that leekspin song done to the japanese vocaloid and it was pretty cool (especially considering she wasn't singing in japanese).

 

[rexxzecutioner]

this is terrible.

 

[Anonymous]

the english version is called vocaloid too.

that said, vocaloid is awful. maybe it's better for japanese but for english - for anything other than OOOOOOOOOOHH smothered in reverb - it is as bad as any other text-to-speech synthesis. voice synthesis is not ready for english.

 

[Cruelty]

the only use i could see for this is for a robot-ish sample in a dance song. even then, I hate that damn anime voice.

 

[Danton Lavergne]

Meh, that's not very good at all. There's so many better VSTs like Voices of Passion, and Symphonic Choirs for example.

 

[Sabao]

Actually, even for Japanese popstar standards, the Miku one still sounds rather robotic. It's nowhere near the real thing, but it's still pretty amazing.

 

[arcthemonkey]

voice synthesis is not ready for english.

http://www.research.att.com/~ttsweb/tts/demo.php

Just a little fiddling with that thing (Especially the "Rich" voice) can make some damned good digi-speak. Just try making Rich say "Hot damn". Put a clip from there into Soundforge for 15 minutes and you can iron out all the wrinkles. Useful for some things, and the best I've heard.

Doesn't sing yet, though.

 

[Prexus]

http://www.youtube.com/watch?v=MJmSO2E7QX0&mode=related&search=
Awesome.

 

[hima]

@arcthemoneky
Wow that's neat! They should have expression option in the future XD

@Prexus
Nice found O_O. This is so old that I totally forgot about it. Sounds pretty good indeed lol.