High Quality Alternative To Vocaloid

[braincell]

I was not aware of this but EastWest has a product that sings and it sounds much better than Vocaloid. It's classical style but that is what I like. The choice is clear now:

http://youtu.be/aI5Gg2-mhmU

Update:

In a review I read, the word editor is very difficult and time consuming to use so that's something to consider if you are writing an opera.

The sound quality over Vocaloid probably trumps the difficulty in programming it. Even the newest Avanna Vocaloid is going to sound more synthetic than this does.

[Nestor]

If you try it out, I think you're going to be disappointed, it does not work very well, and the sound is somehow sort of compressed or something.

[braincell]

Perhaps but Mike Oldfield uses Vocaloid and that sounds much worse. I'm sure with the right effects I will make good use of it.

[dante]

If its better than vocaloid Ill check it out, depending on price.

[braincell]

They lowered the price $250 now. It seems times are hard. I think I better wait and see what Avanna sounds like. Probably neither program is ideal. Symphonic Choirs has way more gigabytes though at 30 gigabytes. I think it's more professional but neither is professional. Avanna is a celtic themed pop singer. Probably in the same price range.

[Neil B]

The difference with EW choirs from Vocaloid though is that you cannot have a solo vocalist singing words or phrases.
If I remember correctly there are solo voices but they can only sing aahs. Only the choirs SATB can be used with wordbuilder.
So if you're after a solo vocalist, you are still limited to Vocaloid.
If I get time, I'll drop a track using both onto my website for comparison.

Can you tell me more about the Celtic side of Avanna (a link?).
My wife is a huge celtic music fan (Clannad, Capercaillie, Iona etc)

[Neil B]

The following uses Lola, EWQL Symphonic Choirs & Spectrasonics Symphony of Voices for a few bits too.

It's part latin, part English (at the same time)

Start at the 2 minute mark to save yourself a lot of ear-pain

www.tonaleffects.co.uk/agnusdeinb.mp3

[braincell]

Symphonic Choirs sounds great to me in that recording. The Vocaloid is nice but obviously not a human sounding singer. It's too bad that Symphonic Choirs doesn't include any solo voices with singing.

[braincell]

Jerry Goldsmith:

http://youtu.be/o8DCmBnmvjE


Could someone actually think this is not real singing? It probably sounds a bit muffled because of YouTube data compression.

[astroman]

no, it sounds like a real choir...
but I'd sack the engineer in no time if I were in charge...
the 'sss' drove me nuts within 20 seconds - not sure if that's just YT
it has no atmosphere either (the kind you associate with a real environment)

cheers, Tom

[braincell]

Isn't atmosphere something you can add with reverb? Better no atmosphere than a bad atmosphere. I like to record dry when possible.

[astroman]

sometimes it's better to just use the location... if available
like here... at 0.40 you see a 2nd ribbon 90 degree crossed for ambience... look at the room, listen
try it with a reverb

cheers, Tom

[braincell]

I like the clarity of her voice but it is not blending with the music at all. It sounds stuck on top of it rather than a part of it.

Maybe I should consider Voices of Passion.

[kensuguro]

hey, this is pretty cool! I like how easy it is to set up.

[braincell]

What is cool Kensuguro?

[kensuguro]

The plugin you just posted.. sry, I didn't follow the topic of discussion all the way down. But ya, the plugin sounds pretty cool. I wonder how well you can control dynamics. It does have that "compressed" sound you talk about. Or, more like it sounds unnatural because the demo notes are held straight with no dynamic changes. Also the demo sounds pretty aggressive, like f~ff intensity. I'm going to look for more demos. It seems usable, I just doubt I'll have any context to use them in.

just checked soundsonline demos..
Criteria: I think the plugin is presented too clear in this one. Really shows the bad phoneme transitions. Maybe it's a timing thing, but the gaps between each vowel makes is sound strange. The "s" particularly sound strange to me. But with beefier chords and held notes there are moments where it sounds good. I'd say this tune is how not to use this plugin

O Grosse Lieb: Well, the piece is well written by Bach, who obviously doesn't suck. He doesn't suck by quite a bit. The reverb on this one works well to smear the unnatural phoneme transitions. And ya, a good composition will make things sound good. Good subtle dynamics in this one. I wonder how long it took to put together. You can really hear each voice in this one, and I think the voices remain consistent with the part. (tenor is always tenor) but that could be because of the composition.

The Secret:
This would probably be the more typical use case for games and such.. The slow held notes definitely bring out the best in the plugin. The composer is smart to use the voices as padding, and all other instruments to build up tension.

Siberia:
This one's pretty bad.. unnatural transitions stand out again in this one.

Kira:
bad section strings.. anyway the chorus.. Again, long legato phrases sound good. When note movement is busier, the transition problem strikes again.

Itineris Spiritus:
Beautiful composition.. More reverb on this one helps too. Great subtleties. Ohh man, nice cello rise. I just like the song. I think this is one of those cases where I probably wouldn't care too much what the tune was executed on because I like the writing style. Great blend of chorus and strings, man, this guy is good. This tune is more technique than plugin....

Obviously, sampling is showing advantages over the synthesis method used by vocaloid. I still think Vocaloid uses some sort of klatt synth based method since it has that signature sound... well, definitely not klatt straight up.. But I think this is where concatenative synthesis and spending lots of time on phoneme analysis would help, kind of like how Nuance's tts voices sound so much better than other ones. Synful takes this approach to symphonic samples and I think it's awesome.
(ftp://ftp.scansoft.com/pub/wav/rvoice/usf012.wav)

Chorus is definitely a place where research from tts engines can be applied. I think Vocaloid's technology is outdated.

[braincell]

I think you must mean the Vocaloid 1 demo. That is old. Vocaloid is more popular in Japan. There are many more Japanese voices.

Maybe all these drastic price drops in Native Instruments Quantum Leap means they are getting ready to release new products to replace them.

[kensuguro]

ya, I know all the anime voice vocaloids.. those sound.. well, you can't use them for anything because they're anime voices, and they all sound like autotuned robot voices.

[braincell]

People like fake autotuned voices. 808 drums do not sound like real drums. The problem with Vocaloid so far isn't that it sounds fake but that it just does not cut through a mix very well the way real voices do. That's my perception of it. There is some lack of character that I can not define. The Avanna demo uses an Enya song and if you compare it to the original, Enya sounds so much better but so does the real orchestra in her piece versus the sampled orchestra in the Avanna demo. Maybe it's not the best genre to use that software in.

[kensuguro]

oh, I was under the impression you were looking for classic voices... but anyway, autotuned anime voices are fine for specific cases.. just tough to use as a production tool because it's so niche. And for chorus work.. well, if I had a full chorus of robot anime voices I'd go insane. BTW, does vocaloid even work on "real" voices anymore? I think I saw some for vocaloid 2, but for 3 it seems like they're focusing only on anime voices.

I checked out the Avanna demo.. at least this one's not in your face anime voice. I guess they just make these voices into anime characters so it's easier for people to conceptualize. But still, the sound isn't very attractive. This one seems usable for mockups.. but I still wouldn't use it as chorus. As a vocal recording, like you say, it's just not too good. Not that much definition and fidelity. I also think there's something up with the pronunciation that screws up Vocaloid. The synthesis engine may be okay, but the phonemes just don't sound natural.