working with vocals

[Sharmaji]

yo dave we really need to do a sticky how to work with vocals honestly
seems to be the new topic

i'd be down to do this. any interest?

[dubesteppe]

YES

[Mad_EP]

I concur

[Cubicle]

Yes please

[outbound]

I recorded a single note vocal bit earlier, after I'd finished processing and whacking it in a sampler though it ended up sounding like a flute

[theedman]

yes x1,000,000

[Sharmaji]

ok, cool-- let's talk about working with vocals!

Vocals occupy THE most important space in your mix. If the human voice is there, everything is second place. Even if they're super-processed samples-- the vox are still the most important thing Burial's "Archangel" and Skrillex's "Scary Monsters." They create the single-most important emotional connection between the listener and the song; the words, the vocal melody, the phrasing, the texture of the singers' voice-- EVERYTHING about a vocal is important, and deserves to be looked at.

If you're working with original vocals, specifically working with a singer, then we can break this down into 3 elements: preproduction, including writing and editing; recording; and mixing. There's a good bit of overlap between the 3, but if you can delineate them from each other, you'll have a much easier time staying on target. Your target being, of course, getting the absolute best vocal peformance for the song.

STAGE ONE: PREPRODUCTION

Chances are in this case, you've written a track and are looking to get vocals cut on top of it. This can happen in any number of ways; i've sat in my apartment with a pen and a pad and written lyrics with singers before; i've had fully-produced vocal sessions emailed to me recorded by professional and organied (and sometimes not-so-professional-and-organized) engineer) that need to be edited for a song; i've sat in the studio for hours on end with singers, sculpting a vocal phrase, editing the rhyme schemes, word choice, melody, etc until it's the best it can be.

In this stage, the vocals work as another element of the track-- you get a melody, you find the right instrument (in this case, the voice) for it, you move along. Some things to really consider:

-- Key of the song. Every voice, and every singer, has a range in which the voice is the most powerful or most beautiful; in opera they call this the Tessitura, and it's worth keeping track of. It's hard for any singer to keep solid pitch at the bottom or top of his or her range; Thus, you want to work in a key where they can sound their best. If i'm writing a tune that i think is going to be good for a female vocalist, i'll start in A or Ab; for men, C# or D is usually good. Well-trained singers can move all around their range, and it's worth you doing some extra work to go back and re-pitch your song if the singer sounds much better in a different key. In other words-- save your midi!

--phrasing and diction. Great songs are easy to sing along to, and the words are easy to hear. Whether or not they're easy to understand, or mean something at all, is an artistic choice, but you've got to be able to hear them, and deciper them, to work. bunching up words into weird rhythms, putting alot of sibilants (c's and s's) or plosives (p's and b's) will make a song sound like a tongue twister. it can work if you're Twista; it rarely does otherwise.

--melodic ideas: again, if you're writing something mean to be catchy, having melodies that you can sing along with and get stuck in your head are an absolute goal. You can do this with any number of methods: have a slow-paced chorus with fast paced verses (or vice versa)-- a chorus that jumps across the whole octave, while the verses stay around a few notes; or the good ol' "let just make this oohs/aahs/yeahs here." You may dislike the song, but the David guetta/Usher tune "Without You" does each of these fantastically. and that song's been stuck in the collective unconscious for the better part of a year.

--production/arrangement ideas: You should be keeping notes for yourself as to what parts you want double-tracked, harmonies, whispers, adlibs, etc. Note that an n+1 rule is always good for doubles; if you want something doubled, get a keeper triple. if you want 2 stacks of harmonies, get 4. etc, etc.

ok great, so you've worked with your singer, or at least have a song to record that they'll sound good singing. let's record it!

[Sharmaji]

STAGE TWO: RECORDING:

again, our goal is to get the best performance possible, and no detail is too small. we should have an idea of where to spend our time and effort though, so my list of elements in a vocal session, from most important to least:

song/singer/producer/engineer/studio environment/mic/pre.

there are endless conversations on the interwebz about the last 2, but they are the 2 least important elements to the vocal as a whole.Song & Singer you've hopefully figured out in part 1; as far as producer, engineer, environment-- there's a couple of ways to do this. If you're homies with the singer, have a good rapport, can push them to really nail a note, then by all means-- produce the session. If you're not, or if you're inexperienced, know that how you run a session and where you run it can make a HUGE difference.

the most important part of this post: producing a singer, working with them in the recording process to really get an amazing performance, and engineering it so that it's captured well, is an art and a skill. The engineer gets the best possible sound out of the player; the producer gets the best possible performance. If you're recording a talented guitarist, ideally he or she just comes in, listens to the tune, chooses his instrument, you guys talk about the part, they record it and the check is in the mail. Recording the voice comes with a whole other world of concerns. You're not just recording a performance, you're recording an actual person. There's no separation between person and performance, and that commands a ton of respect from all involved. Coupled with any singer is emotion, self-worth, connection to the song, etc. As a producer you need to take all of this into consideration and make it happen.

This isn't easy; especially in the extremely competetive world of songwriting, we call this process "beating up the singer;" you're sitting in the control room asking this person to really push their abilities, to hit that high note that they can't probably actually hit, but their voice sounds amazing. after the 17th take where they can't get it, imagine how their head is doing-- repeated failure at something you're good at is NOT an easy thing to live with. You need to conversely support them, challenge them, and maintain a positive environment in which they feel safe to work at this level. It's not easy.

finally, mic's and gear-- use a mic. a good one. use it with a pop filter. use a good pre track with some compression if you like the sound of it.

seriously though, i've cut keeper vox on everything from a tube47 that was used to record Duke Ellington, to a plastic mic i bought on canal street. Learn what tools are available, learn how to mic a singer, and go from there. If you don't know, you'd be better off spending $200 in a good studio for 4 hours, rather than $200 on shoddy gear. absolutely phenomenal records have been cut on sm58's:



the choice of mic' isn't what's going to make your record awesome. Using a studio will also get you a most-likely-experienced engineer, who can create vocal comps of takes for you. Otherwise, this is time-consuming and not-very-fun work. If you do it as you go, you run the risk of depressing the singer as they hear back all the non-keeper takes; if you do this on your own, you need to be meticulous in terms of labeling and organizing your comps. Very necessary and not much fun for the average producer.

[Sharmaji]

STAGE THREE: MIXING VOCALS:

so at this point, you're probably ahead of the curve; you've got a great song written so that the singer can sound great singing it; you worked with him or her in a performance-conducive environment and got an excellent, and excellently-recorded performance (the recording of which suits the song)... all that's left is to fit the performance into your song.

There are probably as many methods and theories on how to mix vocals as there have been songs recorded. Michael Brauer and David Kahne mix by sending the lead to no less than 5 (!) different compressors, each doing something different-- bringing up tone, squashing transients, etc. mix them all together and you have their sound. There are far more straight-ahead processes but some things pop up more often than not in modern recorded vocals:

--there's too much dynamic range
-- the vox get "honky" at 2-4k when the singer gets loud and their throat really opens up
--the low end gets muddy because of proximity effect
--the sibilant consonants get waaaaaay too present post-compression
--the vox aren't bright enough.

there's plenty all-in-one, or nearly-all-in-one solutions for these, both in hardware and software land, but generally some balance of dynamic control, de-essing, eq, and limiting will get you to a good place. As an extremely broad plan, I usually route my vox as such:

--lead vocal, single or double, to its own bus. quite often there are overlaps of words between sections, so you'll have the verse, prechorus, chorus, and bridge sections all tracked separately. rather than deal with 8 separate channels all with the same treatment, i'll just bus them together. gives me a good balance of separation, for automation's sake, and workflow.

--main background vocals, hard-panned L &R and sent to another bus. If these are consistent, i may just bounce them all down to stereo and work like that. no sense in keeping things busy in the arrange window if they don't have to be.

--harmonies, same thing, their own bus and maybe bounced to stereo.

--i'll then route all those busses to a final vocal bus, where i can compress more if i need to, and adjust the levels of the vocals as a group if i need to-- if something else gets out of whack, i don't have 43 channels of vox i need to change. just the one bus.

i'll spend a LOT of time on a lead vocal. Early on, i'm going for tone-- what balance of compression, eq, and maybe limting will get it feel the right way for the tune? I'm usually looking for it to feel like it lives in the center of the song, tonally fits it right-- if there's a lot of distorted guitars/synths going on, i don't want a clean celine dion-sounding vocal-- and i don't hear atifacts of overcompression, de-essing, eq.

FX then come into play. So much of mixing is about creating the right vibe in a song, and reverbs and delays can help to add density-- which, in turn, adds excitement and energy. i like to use a lowpassed 1/8th-note delay alot on lead vocals; i'll route it to the main vocal bus as well. you might not even be able to discern it as a delay in a final mix, but it adds density and thickness to the sound without getting washed out w/ verb.

you also shouldn't rule out completely fucking up your vocal sound. guitar amps can do wonders, as can bandpass filters, etc. sometimes a really effected, stylized vocal is the way to go, even if just for sections. If you look at the song "nectar" that i've linked, I ran kendra's vocal through a vox guitar amp and an amp modeler for the duration of the tune-- she was singing about "saturation" and it helped to get the vibe across. A studio i work at has the tiny peavey amp that Julian Casablancas recorded the 1st strokes record's vocals through-- it's the right sound for the record. Similar for Meetu's vox in "Pyar Baile," I recorded her in my hallway with a TON of compression, distorting the preamp and tracking with spring reverb on the lead-- we needed a blown-out bollywood sound, and that worked. be fearless in getting your sounds!

i keep 3 reverbs on sends (long plate, short hall, very long hall) and will write those sends in on the lead with automation. I generally don't like a really washed out reverb sound as a main vocal sound, but it's a great effect.

at that point once i've got my sound via inserts and sends, i'm going to do my vocal rides, and probably do them on small speakers so that i can hear that the vocal always sits at a good level. macbook speakers are great for this; so are the $15 speakers i bought at staples. if the vocal sounds loud and clear above everything else in those, you're good to go. Put your bus channel in "latch" or "write" mode, and take a pass at the whole tune, moving your fader up and down as it needs to be.

for backgrounds and harmonies, i'll be more aggressive in the processing, as i don't want them to distract from the lead-- just to augment them. i'll often remove all breath sounds, de-ess like crazy, boost a bunch at 12k, compress and limit so that the meter doesn't move at all, and cut them at about 12db/octave at 200 or 300hz. they can be surprisingly thin and still add an element of hugeness.

after all that, bring your master vocal fader all the way down, bring it back up to where it sounds right... and add 1dB to the level. voila!

hit me back here w/ questions...


finally, some credentials: recently-released tunes i've co-written and/or recorded/mixed/produced etc:

[mitchAUS]

Really appreciate what you have done here.

[Electric_Head]

Awesome.

Thanks Sharmaji

[Ongelegen]

Thanks

[lloydy]

Brilliant post sharm thankyou.

[3za]

Great post

A studio i work at has the tiny peavey amp that Julian Casablancas recorded the 1st strokes record's vocals through-- it's the right sound for the record

I have always liked the sound of his vocals on that album, always thought it had something to do with a guitar amp, but have never been able to get that sound. If you know anything else about the recording of the vocals/album I would love to know?

[deadly_habit]

ah noted to come back and have fun with, great job so far dave

[Sharmaji]

cheers ya'll-- fire away!

@3za-- i know that they made Fabrizio move his hihat to the right of the kit, as they were tracking the snare live with distortion and it was making a bloody mess of the hat sound.

for more fun w/ vox and guitar amps, get yrself one of those shure green bullet mic's. you'll get grounding noise for days, but that thing is badass!

also the copperphone mic can be a fun tool for bandpass sounds-- again, run it into an amp, distort, and get a funky high end.

[drake89]

dude thank you so much for the little nuggets of wisdom. especially the bit about 1/8 delay. I remember you said you put 1/4 reverb on your default, but I've only ever seen it in ms (ableton stock). What verb do you use for this?

[Gurnumsbug]

sharm

[narcissus]

this is all good.... but i must disagree with your point that the vocals are the most important part of the mix... there are many many recordings in garage, house, and trance (other styles as well but these leap to mind) where vocals are used primarily as a rhythmic and tonal element purely... the drums or bass taking main stage. vocals can provide background ambience like no other... in fact i believe delegating them to such a task can give a track an unusual subtlety. yes voice is the most ear-catching sound there is.. but that means that when you put it exclusively as an echoey ambience, the ear is attracted most to the 'environment' which a verbed out vocal creates, and the mind is free to frolic as it will among spacious landscapes of echoes, while not being distracted from groove that the drums/bass are creating. you can use the 'tone' of vocals, without the melody, or even rhythm.
by using this attractive, yet indefinite/blurred quality, one can create some truly strange and wonderful effects

[JTMMusicuk]

Big thread, thanks for those posts Sharmanji; will be very helpful as i am recording vocals for my EP soon