So, I want to make a death metal vocal library - any advice?

Sampler and Sampling discussion (techniques, tips and tricks, etc.)
RELATED
PRODUCTS

Post

Yeah, really, I've been thinking about this for months now. I know a girl who can do the vocals, and she's got exams until June 23. So, I have three weeks to make a plan. I know male death metal vocals would be more marketable, so if this works I'll definitely look for a guy to do a male version too.

Right now I'm trying to figure out how many different sounds we'll need to record. I've asked another girl who's majoring in linguistics to ask her professors, and I've found a list of what needs to be recorded to make a voicebank for Utau, but that's Japanese. So, anybody got a list for English? Or at least Texan or Geordie?

I know that vocal sample libraries are much bigger and more complicated than anything I've done with basses, but death metal should make it easier. I figure that not having to worry about pitch will simplify things and greatly cut down on the number of samples (so maybe we can do some round robins or different articulations instead, like whisper, growl and scream), and also death metal pronunciation isn't perfectly clear, so it will be easier to get away with words not quite sounding right.

I know there is a metal/hardcore vocal sample library out there, but that one has a dictionary of about 100 words, so apparently they did entire words? http://soundiron.com/products/voices-of-rage

Anything else I should know? At this point I have an idea, a vocalist, and not much of a clue.

Post

Just hire someone to do them. Look for an appropriate forum at Ultimate metal for example, and hire someone.

I love Soundiron, but with the metal vocal library, they really didn't know what's up. I mean, it could work for you if you think it sounds good, but to me (actually producing metal music) it sounds like horse shit. Hell, even I could do a better job at growls (done it before) - I guess that pack would work for metalcore or such well enough though. It absolutely depends on your context - if you just want something with that "death metal feel"; sure, I guess that library works *well enough*.

To be honest though, if your motivation is to get something that doesn't quite need to hit a note, there's always a 'perceived' note in death metal vocals as well that you'll need to hit to sound good. You don't just go at it and start screaming. You'll need an ear to hear which perceived note produced by the throat fits the musical piece. Granted, it isn't as hard as actually singing clean notes, but it's something to think about and it's definitely something the pro death metal vocalists know by heart.

Post

My motivation is mostly that this would be real interesting. More complicated than anything I've done before with samples, but not so complicated that I can't handle it. Plus, I've worked with this girl before, though if this works I'll definitely find a guy to do a male version, too.

Pitch is a good question. There's no need to really sing melodies and no need for a wide range, but like you say, it has to fit. She did tell me something once about recording vocals over a drum track, before any guitars were done, but that seemed weird even to me. So, like a master pitch control to match the tonal center of the song, but no need to deal with the pitch of each syllable individually, maybe. Maybe.

I don't have the Soundiron library, I pretty much just know it exists.

Post

I know jack about the singing technique, other than what I've heard over the years... Be that as it may, a couple of things come to mind to get started:

1.) Start with a dictionary of common words used in such songs... it would probably be the same old ones you find in any old song... "waiting", "my", "in", "for", "the", "dying", "trying", etc. etc. ... the list is going to be much smaller than a real dictionary, but maybe 200 words?

2.) I wouldn't sample the words without context, but instead maybe write out some fake lyrics that can actually be sung in context -- you want the singer you're recording to be "in the moment", as it were. The lyrics don't have to make sense, per se, but each individual line should be cohesive... "waiting for my life to end"... or whatever. Come up with dozens of such fake lyrical lines, and mix in the same words many times to get a variety of inflections etc. of the same words across different lines. Do your fake song recording against an actual backing track.

3.) Include some special growl, scream, etc. techniques (obvisouly!), and make sure you capture many many variations of each.

4.) Hire a professional singer, unless you know a hack that's good with the techniques etc. and can give you a couple of good recording sessions without blowing out his or her voice box. :) I'm sure you can find somebody for 2 x 4 hr sessions @ $100 per... that's good beer money!

5.) Pitch -- luckily you don't have to worry about being too melodic... you need some control over pitch, but what little variation you need you should be able to get just with sample-level pitch/formant control. I think. THis simplifies your recording and post processing requirements considerably.

6.) Word length -- this is where it gets tricky... You may want to perform the same word in different ways ... like very staccato/quick "Don't make me <whatever>" vs. "Dooon't maaake meeeee <whatever>"... know what I mean? Does this mean you get fancy and chop each word into begin/mid/end, crossfade and envelope them together? Or do you use the platform's support for timestretch and get fancy with that? For example, Kontakt has timestretch capability... are you even targeting kontakt?

7.) Production quality end results, etc. -- really take a deep look at how these bands produce their vocals, and try to provide those production techniques out of the box. It's one thing to record a bunch of dry samples of somebody screaming and growling, but the producers must also apply other techniques to get things to sound the way they do. Track doubling/tripling (again, having multiple of the same words will support doubling, tripling without phasing)... layering of pitched (e.g., octave down) tracks... etc. Your patches and/or samples could provide those pitched layers out of the box.

Anyway... some things to think about...
You need to limit that rez, bro.

Post

Thanks, that's lots more to think about. Especially number 6. Might need short/long vowels at the very least, I don't think doing just long vowels and cutting them short will sound good.

7 is also something I've been thinking about, especially the layering part. Hopefully we don't need too many samples total, so recording everything two or three times which can then be used either as layers or as round robins will be an option.

This is definitely gonna have to be Kontakt. Eventually I would like to have a wordbuilder like something in Voices Of Prague (which is what got me thinking about this) or Realivox Blue, but that's a huge challenge in itself and something I'd have to hire a serious software developer for later. For now I just want to make the samples a wordbuilder would need.

Here's the vocalist with her old band. I wish I had video of her doing a verse in the middle of a Shakira song with my pop band, heh. Anyway, this should give everyone some idea of how many different techniques or articulations we'll need.

http://www.youtube.com/watch?v=1pwBQhsQHl4

Post

There is already a product like this on the market. It has been recently released. It's Voices of Rage from Soundiron.

Post

I will take the Lord's name in vain, whenever I want. Hail Satan! And his little goblins too. :lol:

Post

Hey bro,
go look at the thread for Blue. I included 3 lists of English vowel/consonant/diphthongs. Also look at anything written about blue, cause you are basically doing a (smaller) balls to the wall (sort of) version of blue.
Image

Post

Yup, I've definitely been paying attention to that thread. Very useful.

Post

Girl voice for death metal? Isn't it quite rare choice, I'm not really a connoisseur, but those growls are usually out of a woman's scale. Unless you can get someone like Yma Sumac or Violetta Villas and soak her with pure alcohol for a couple of weeks before the session.

Post

I think both male and female would be great. Would give lots of options for... uh... harmonies... Death Metal Love Ballads... that sorta thing.
You need to limit that rez, bro.

Post

RE: wordbuilder... keep it simple, IMO. Like I said, how many actual sampled words do you need? Maaaybe 150-200 to start, maybe not even that many... If I were you I would find the 80-90% mark... say you need short, med, long versions of each word, then you provide a simple timestretch for the med and long (perhaps stretch via mod wheel). Really, the limits of simple "sung phrases" (not including long screams, growls, etc.) must be within a typical range, right? Like the longest single word might be 4 seconds max? So maybe you do .5 sec, 1.5-2.0 sec, and 3.0-4.0 sec with timestretch on the latter two to fill in the gaps. Wordbuilding within the context of very strict tonal characteristics is very difficult.. I would think adding a bunch of white noise, harmonics, distortion, etc. would make the problem an order of magnitude moreso... :shrug: For example, imagine wordbuilding a breathy voice -- you have the tone, formants, whatever to matricize, then this whole other breath dimension complicating thing. The main thing with wordbuilding (for me) is that it's too easy for it to sound unnatural. For a general voice system it may be the least complexity compared to a more simplistic/brute force method, but death metal voice is not a general voice system, it's very specific.
Last edited by kbaccki on Sun Jun 01, 2014 7:07 pm, edited 1 time in total.
You need to limit that rez, bro.

Post

<ignore>
You need to limit that rez, bro.

Post

Zombie Queen wrote:Girl voice for death metal? Isn't it quite rare choice, I'm not really a connoisseur, but those growls are usually out of a woman's scale. Unless you can get someone like Yma Sumac or Violetta Villas and soak her with pure alcohol for a couple of weeks before the session.
Are we saying rare because you arent a connoisseur, or because we are going to put on full black, some face paint, and be anal about the terms metal, death, and death metal? :)

Cripper, arch enemy, iwrestledabearonce, in this moment, winds of plague, the agonist, cadaveria, astarte, holy moses, save, astarte, otep, holy moses, save.... wait someone already took time to compile a list;

http://youtu.be/fF_p3Ywerqg

There you go, 250 females that would make justine bieber wet her panties.

I think simple wordbuilder would be good, check that video, or some of the artists on the list, can you make our every word?
Image

KVR Audio Forum Signature is currently unavailable for your web browser.

Post

Wordbuilding is definitely gonna be the biggest challenge. I honestly have no idea how well it will work. In the worst case I'll just end up with a set of growled vowels useful for sound FX, death metal guitarists who are making demos to show the vocalist that "I wrote a new song and want the vocals to go here", and resynthesis in Harmor. I've made some nice synth sounds by resynthesizing vowels from samples of Daleks from "Doctor Who", so...

That's the worst case scenario. But hopefully I'll end up with more than just that.

Post Reply

Return to “Samplers, Sampling & Sample Libraries”