Vocal Synthesis

Founded and At Last are two of the most beautiful songs I've heard in years, owing mostly to the incredible interplay between the 'vocals' and the rest of the instruments.

I was shocked to read in an interview that these aren't manipulated recordings, they are purely synthesized.

In some of the interviews they talk about using 'physical modeling algorithms' in a lot of their new stuff, which I suspect is just Logic Pro's Sculpture(judging from the sounds on Scintilli).

Is there a similarly simple explanation for the synthesized vocals, or did they cook up something fancy in Reaktor/Max/etc.?

Sounds like Vocaloid to me.

Wasn't sure what to make of the vocals on Scintilli until i read about the way Plaid achieved them using purely synthesizers and physical modeling algorithms... to think these vocals aren't real or made using samples amazes me.

I was actually disappointed upon hearing the album that the guitars and vocals were synthesized. The way you guys were talking about them, I was expecting them to be real!

  On 9/21/2011 at 5:32 AM, modey said:

I was actually disappointed upon hearing the album that the guitars and vocals were synthesized.

that's the beauty of it, for me!

Yeah.. it might have "sounded better" with actual instruments & voices... but them being synthesized & reconstructed makes it much more interesting to listen to for me.

hmm. skeptical of this "'physical modeling algorithms". physical modeling is a bit of a dark art, and to achieve any kind of realism is extremely complicated. not saying it can't be done.. but when you look at the complexity of a program such as Praat, and how it does vocal synthesis.. I find it hard to believe that this was programmed by two music producers. Can someone post an example of the tracks in question? I haven't heard the new Plaid, and am curious to hear the vocal synthesis in question.

Got the Japanese CD yesterday, for what it's worth Benet Walsh is credited with guitar on Unbank. I've not a fucking clue what all this technical geekspeak is about how the vocals etc. were done, but it is similar to those on Faster, which came out about four years ago?

  On 9/21/2011 at 10:33 PM, Caretstik said:

Got the Japanese CD yesterday, for what it's worth Benet Walsh is credited with guitar on Unbank. I've not a fucking clue what all this technical geekspeak is about how the vocals etc. were done, but it is similar to those on Faster, which came out about four years ago?

i always assumed that was Ed's voice on Faster. sounds like him singing, to me.

  On 9/22/2011 at 3:01 AM, NewSchoolScience said:

http://magneticmag.c...ing-your-heart/

jazz

pardon?

  On 9/21/2011 at 10:33 PM, Caretstik said:

for what it's worth Benet Walsh is credited with guitar on Unbank.

Interesting. I can believe that, it sounds less forced than the other guitar parts on the album. But it's also the simplest guitar part on the album - you'd think they would have gone for live playing on the tracks that use acoustic guitar sounds instead/as well.. The sampled guitar just sounds dinky to me, then again I don't believe acoustic guitar can ever be correctly emulated. Distorted electric guitar, yes, but not acoustic! Guess it's just my guitar elitism showing.

  On 9/21/2011 at 5:02 PM, thanks robert moses said:

hmm. skeptical of this "'physical modeling algorithms". physical modeling is a bit of a dark art, and to achieve any kind of realism is extremely complicated. not saying it can't be done.. but when you look at the complexity of a program such as Praat, and how it does vocal synthesis.. I find it hard to believe that this was programmed by two music producers. Can someone post an example of the tracks in question? I haven't heard the new Plaid, and am curious to hear the vocal synthesis in question.

http://www.youtube.com/watch?v=CqoXlr1s0ZA

http://www.youtube.com/watch?v=jUZ16X9RmQ4

  On 9/21/2011 at 5:02 PM, thanks robert moses said:

hmm. skeptical of this "'physical modeling algorithms". physical modeling is a bit of a dark art, and to achieve any kind of realism is extremely complicated. not saying it can't be done.. but when you look at the complexity of a program such as Praat, and how it does vocal synthesis.. I find it hard to believe that this was programmed by two music producers. Can someone post an example of the tracks in question? I haven't heard the new Plaid, and am curious to hear the vocal synthesis in question.

For what it's worth in the interviews I've read physical modeling wasn't a term applied to the vocals...

i.e.

  Quote

But ghosting the human voice is just one of many cool tricks on Scintilli. Plaid also used physics algorithms in real-time to generate various sounds in a shared virtual space. It’s a computer-intensive process that removed all the cages in their instrumental zoo.

“It works very well for bell-like sounds or percussive struck sounds,” Ed explains. “It sounds acoustic-y. Effectively you have all these things resonating together inside the computer and affecting each other. So that was another road we went down. Just things to keep us interested, to keep the sparks going.”

For those I really just think it's Sculpture or something similar. Calling fiddling around in Sculpture "[using] physics algorithms" is like saying a dog is doing calculus when it's catching a frisbee - a bit overblown even if it's technically 'true'.

Which makes me think that an easier to use commercial software was put to use for the vocals.

No matter the method though the results are unlike anything I've heard... very, very beautiful!

..about the voices, they said it in another interview, a more recent one... check the interviews thread.

from what i've read it sounds like they actually have created the voices (from scratch) aswell. - awesome + lovely.

  On 9/22/2011 at 1:30 PM, sirch said:

..about the voices, they said it in another interview, a more recent one... check the interviews thread.

from what i've read it sounds like they actually have created the voices (from scratch) aswell. - awesome + lovely.

well yes they said in an interview that , they studied what makes human voice and they just produced it (from scratch)

i mean , no they didnt use microsoft sam/narrator, or whatever...

which is,, amazing..

does this sound familiar?

yes, they've used virsyn's Cantor for the vocals.

  On 1/11/2012 at 5:06 PM, mish said:

does this sound familiar?

http://www.virsyn.de/Demo/CANTOR2/WithoutLove.mp3

yes, they've used virsyn's Cantor for the vocals.

mystery solved, thanks komshu!

Woah, learning this amazes me alot ! I already found this album beautiful but now it's on another level.

I thought the guitar were real, but post-processed - I like the way they sound anyway, especially on Missing and Founded. But I'm flabbergasted about the voices !!!

Feel like Scintilli has many more surprises yet to be discovered

  On 1/11/2012 at 5:06 PM, mish said:

does this sound familiar?

http://www.virsyn.de/Demo/CANTOR2/WithoutLove.mp3

yes, they've used virsyn's Cantor for the vocals.

not really!

  On 1/18/2012 at 10:47 PM, theplaided said:

  On 1/11/2012 at 5:06 PM, mish said:

does this sound familiar?

http://www.virsyn.de/Demo/CANTOR2/WithoutLove.mp3

yes, they've used virsyn's Cantor for the vocals.

 

mystery solved, thanks komshu!

you're welcome, arkadashim! :)

  On 1/19/2012 at 12:25 PM, StocKo said:

..But I'm flabbergasted about the voices !!!..

I fell in love in this lady.. until I understood she's not real.. :D

i don't think the 2 sound the same. similar, tho, i'll give you that. but quite different also.

  On 1/19/2012 at 9:52 PM, sirch said:

i don't think the 2 sound the same. similar, tho, i'll give you that. but quite different also.

well, they are not completely the same, but you can hear the algorithm looks quite similar.

the mystery for me here is rather if it's a real language, or just random phonemes..

i don't know what software/hardware they used to make the vocal samples, but in response to mish: I swear every time i hear "at last" I hear the vocals say "We've been waiting for so long." Perhaps some of the vocals are lyrics and others are just phonemes (doubt they are ever "random" in the case of plaid though).