Vocal Recording/Mixing Tips

[Navi]

I know this isn't the best forum to ask (I usually just check the FutureProducers forums) but I figured I'd toss it out here anyways.

The past few years I've been recording based on hands on experience and tips from GrayLightning and other people I've talked to. I'd like to get a wider range of opinions and maybe learn some new tricks in the process.

So for anyone who records vocals or has experience mixing them into a song, what tips would you recommend to get the best sounding vocals? I'm talking anything from compression and reverb to EQ cuts and boosts to mic placement and input levels.

Thanks.

edit: and just to clarify, i'm not specifically talking about hiphop vocals. I'd just like to hear some general tips and personal experiences.

[zircon]

A few general words of advice..

1. When recording the male voice you will inevitably have a lot of bass content. So, you really need to consider the "proximity effect" (with non-omnidirectional mics, bass response increases exponentially as you get closer to the mic). I usually prefer recording 8-14 inches away from my SM-58. It's sensitive enough that it is still very detailed, and while the recording is a bit on the quiet side, I can always normalize it. This avoids most of the proximity effect so you don't have to spend time hacking at your recording w/ EQ. In addition you won't have to worry about sibilance and plosives as much.

2. Compression is really really useful. Most times I don't use extreme compression on tracks because it can produce unnatural pumping. However, by virtue of how people normally talk and sing, you're going to have a wide dynamic range. Compressing it is really the only efficient way to level everything out into something more manageable. Otherwise you're basically going to have to manually automate the track volume for louder and softer sections, which is pointless.

That's about all I have.

[Compyfox]

In the last couple of years I learned and used a couple of tricks, that're simple, yet very powerful sounding in the end. I mean I'm pretty sure that others might disagree with me, it also depends on what kind of style you want to produce - cause there's no set rule which applies to everything - but there're some small hints that I can give you on the way.

HINT 1 - RECORDING IN A PROPER ENVIRONMENT:

Always try to record in an as neutral environment as possible. The less noise in the background, the less you have to edit out (which destroys the sound quality). I had this problem for example with tracks from pixietricks while working on Chrono Symphonic. Even though the recordings from "To Far Away Times" (known on OCR as "The Place We Knew") were fairly clean, there was still noise in the background.

You "can" edit it out (noise reduction tools - i.e. "X-Noise" from Waves, "Redunoise" from Voxengo, "SoundSoap" from Bias or "Studio Clean" from AconDigital), however you will also loose some of the upper harmonics, which "can" be kinda reproduced with exciters - but you will have it a lot easier with clean environments.

HINT 2 - DON'T PUSH THE RECORDING RIGHT FROM THE START:

The most known problem while recording is trying to find the right level. Most recordings are overpowered (especially from beginners) and tend to clip. That results from very high set input levels. If you tend to have a huge dynamic range while singing, chances are that you set your recording level fairly high, and as soon as you get loud, the recording clips.

You usually prevent that with softlimiting or slight compression (ratio 2:1 to 4:1 for example, fast attack, medium release) in the input chain, but then you have problems while engineering later. A hint from my side is to try to have a loudest peak (test it with "almost" screaming while singing) of about -1dB and only use a softlimiter at -0.5dB for uncontrolled peaks - no compressors while recording (unless you want that tube-compression sound). This way you have enough headroom without noticable pumping or squashing. You can always make it louder in the end, but you can't get rid of clipping. Alternately you can of course record at -2dB (loudest peak), so that you don't even need a limiter at all. If you have a friend who knows some stuff about mixing, this might help you in a lot of ways, as he/she can handle the recording, while you handle the singing.

HINT 3 - GOOD RECORDING EQUIPMENT IS MY FRIEND:

Sad but true. You can't really expect a normal Shure 57 (dynamic) plugged into a Creative Live! to sound good. Forget that, whoever tells you different - DON'T EVER LISTEN! Same counts to using a built in Macintosh iBook microphone or gamer headsets (like the PS2 USB microphone). That doesn't mean that you can't produce "some" stuff with it after massive editing. But this equipment is usually crap in terms of frequency ranges - so don't even think of getting good results with them.

While recording vocals, the better equipment you have, the better sound you get. Of course you don't need ultra highend equipment for thousands of bucks - you can get away with decent equipment, but here I recommend to invest into a proper preamp at least (tube or not, doesn't matter - tube just adds some warmth or better said "natural distortion" in a subtile range unless you overpower it on purpose) and best if you have a large membran microphone. Here a B1 (Behringer) for example can be more than enough. Small membranes like AKG or Samson can work too, but they tend to be "too crisp" and are a bit flat in the lower frequencies (that's why they're mostly used as overheads for drums for example).

If you don't have the money (though large membrans like AKG, Samson, Behringer and Marshall are fairly cheap nowadays), dynamics are a start, but here it depends "how close" you're to the microphone. The closer, the louder, the more "boomy" due to the volume boost. Condenser are a bit better in terms of control, the negative side on them though: they tend to record even the slightest noise. So a fairly clean room is essential. You can also shape the sound right from the start with the microphone and your preamp. While "tube preamps" tend to add some natural harmonics in the lower section (aka soft distortion and bass enhancement), large membrans can also add some more bass and upper frequencies than small membrans. If you know your equipment, you can work against that. But less effect usage is sometimes more.

Also essential should be a plosive shield. May it be one out of a metal mesh, a regular one, or one "self built" out of nylons from your girlfriend - doesn't matter. It destroys a lot of plosives and hissing right from the start. Set it up around 3-5 inches away from your microphone, and you can get as close to the shield as needed. Result are still clean recordings.

HINT 4 - TIME TO CLEAN AND SPICE ME UP:

Depending on your recording environment, you can now clean your vocals or not (noise removal tools, de-esser). But the cleaner you recorded, the better. I guess the topic would become too large if we add a part "how you should sing", but one thing whould be mentioned though: while using headphones (half-open or closed), take into considerration that you can produce some overtalking if the phones are too loud - which makes it harder to edit out, though in a loud mix, the "subtile noise" can drown in the arrangement (which is still not a good thing). If your singer says "OMG make the volume louder, I don't hear anything", be persistent and say "no" and stick to a certain level. Encouraging can give similar results than turning up the volume, so that your singer has to "adjust" and sing louder, too. But this is a different issue.

The next part is adding effects. Well... here there is no set rule to apply, only some small hints. Every track needs a different approach. Rap for example tends to have fairly dry vocals and strong compression, while ballads can have a long reverb trail and slight compression. It really depends.

Your main weapons of choice however are EQ (or if your recordings are too muddy or lack upper harmonics, even an exciter), compression, delay and reverb. Sometimes even a subtile chorus effect. But the main ingredients are always the same. Don't forget automation or ducking (sidechain compression) either, as vocals "always evened out" tend to overpower certain sections, or you need to push instruments a bit while the vocals need some more backing, etc.

HINT 4 - EXAMPLE 1: THAT TRANCE IN MY HEAD:

Again, please don't see this as guide, as every producer does it different - more like hints. Trance productions are known for their airy productions. Airy as if form of crisp and long reverb trails. Delay can be used for echo too, but usually reverb is where it's at.

The vocals are usually slightly compressed (ratio 2:1 to 4:1 depending on the effect, fast attack, slow to medium release). Now the layering is important: the reverb is usually not used as insert effect. Stuff like Delay and Reverb is in Send mode. This way you have not only better control with automation, but it also sounds more natural and you can use the very same effects for other instruments too.

Reason is simple:

Insert effects use the original signal and run it through the effect section. the sound will be morphed into the "wet" signal. Send effects however "add" the FX signal to the original one, which makes it louder, too, but you still hear the "clean" track (which is mostly the desired effect). Not only is it less CPU intensive, you can also "check" the clean, or the FX signal seperately. In short: way better control. But back on topic.

You have the clean recording. Now use a send bus and route it to the setup reverb (usually large halls with fairly quick damping, maybe even a gate at a certain level). In Cubase SX for example you can set up how strong the levels are. Depending on your track, it can be between -20dB (subtile) to -10dB (more present) - pre- or post-fader (depending on what FX you want to have). But don't make the reverb even with the vocals, that only kills the mood.

"Echo" can be created with delays automated to certain words. Best if you can synch your delay to your host. Then every 1/4th beat should be okay for an echo. Even here: some early reflections/quick damping and maybe even a gate at a certain level. Don't overdo it. The rest can be automated. In Cubase again: catch the word/phrase with automation at -20dB or something, then turn down the reverb around 10dB at the same time (so that the delay is louder) and use the delay for the echo, but still have a slight trail from the reverb for the lush sound.

HINT 4 - EXAMPLE 2: THIS IS SOME SERIOUS RAP SH*T:

In hiphop and reggae music, the game is a bit different. Here strong compression is used (ratio 4:1 at least) and the reverb is only on a short and subtile level - just that the vocals fit into the track. The "FX" comes from backup tracks like "stabs" and "chorus", but the main vocals remain fairly clean. Pop or Soul ballads use the same technique, but here the reverb is a bit more noticable and has a longer trail.

I can't give you much hints there, as every hiphop track needs an own spice to bite, but don't center everything. Pan out the chorus or the stabs (for example 10% to the left, while some "slapback delay" panned 10% to the right from center creates the proper width), or even record "whispered" parts to your raps to add some more spice. Playing around is the best hint I can give you.

HINT 4 - EXAMPLE 3: OMG POP!:

Most pop tunes use fairly decent compression (ratio 2:1 to 3:1, fast to medium attack). The rest is playing around with delay (slapback) and reverb (plate). Sometimes even layered a couple of times to create a vast environment. In modern pop productions, the vocals are also doubled, you have vocal licks (that "ooohoooo!" and "yeaheaaaheah" stuff) and a chorus that was recorded 10 times at least, then properly cut and reduced to 4 channels slightly moved apart from each other to the left, and the same again to the right, to create a rich and "spread" environment.

Even here, the main vocals are fairly clean, only the backing vocals use FX (chorus, massive reverb, I even saw and heard tracks that used bandpass filtering). It's always a matter of taste, how you want your track to sound like (Justin Timberlake in "Cry me a River" moved the vocals to the left and right, each around 45% to create that "beyond the head" sound). But usually the center (main) vocals are widened with very subtile slapback delay and some decend reverb trails depending from pop and rmb ballads to more electronic stuff. The possibilities are endless.

GENERAL HINTS 1 - SLAP THAT SH*T BACK!

This effect is as old as the Beatles. Yeah you heard right - the Beatles! Paul McCartney used this FX on his vocals to create that "widened" sound. Even I used one of it's forms in a subtile way in the "Chrono Symphonic" remix project (again: "To Far Away Times"). But how to create it?

In the old times, a mono delay unit was used to create a very fast "echo" in the ms section. This way the vocals sounded as if they were "doubled". However it could be used live, too. Therefore the vocals sounded more rich. Nowadays it is reproduced with stereo delays. Whatever you use, is your thing. Kjaerhus Audio "Classic Delay" produces some good results, OhmForce and PSP however can use 4 sections to 8 sections in their plugins. Else you only have 2.

Anyway... you can either synch your delay to your host, use very fast delay times (1/8, 1/16 or very fast with 1/32) and create that "natural echo" in an arrangement, that can sometimes sound better than reverb alone. Even here, reflections should become dull fairly quick. If you want to use gating at a certain point is up to you. Important is the "panning" however. The Kjaerhus Classic Delay has a so called "ping pong" mode. With this you can set the panning (hint: start with L15 up to L45 - don't pan it out to the max, which will sound better). Depending on the slider, it starts with left (if turned to left) or right (if turned to right) first and jumps around from side to side.

Now add it subtile (pre-fader, -25dB to -20dB send effect) to your mono or centered stereo track (vocals for example) and see how your stuff will be "spaced out" from the center. Simple trick, large impact -mono tracks are suddenly not "dead center" anymore. The faster the delay, the more it sounds like a chorus or "doubling" effect. Alternately you can simply copy your vocal part, pan it how strong you want to the left/right and slightly move the files a bit backwards (2ms to 8ms). Same trick, different approach.

GENERAL HINTS 2 - LET'S GET DOWN TO MOTOWN

Old as hell, too - but still used: The so called "Motown Effect". This is an effect, where the original signal is being doubled by a slightly or heavily compressed signal. Both signals are being mixed and you get a unique sound that's either "naturally balanced" or "way stronger sounding". It's mostly used for drums and vocals. This trick was and is always gambled high, yet it is fairly simple. All you need to do is this:

First use your recording and make a 1:1 copy (exactly lined up) so you have 2 tracks. Now you should hear a sudden boost of your signal, which is normal (both signals being added twice, give twice the boost). To fix this problem, reduce both tracks by 6dB. For example: Both tracks are at 0dB, not reduce them to -6dB.

Here's the trick: Now head over to your doubled channel and apply compression. The trick hereby is not to squash the shit outta it, but only catch the transients (example: threshold -5dB, ratio 2.5:1, attack 10ms, release autio, gain: the same as your reducing while compression - so that there`s a more controlled limiting). The compressor should only go on from time to time. The resulting effect should now be a more "natural evened out" track. Good for example for fluctuating drums, or if your vocal automation doesn't really work. Due to the fact that both signals are added to each other, you still get the loudness you want to have, but also get some unique effect that's fairly subtile, but give you the "WTF hot shit" effect. If you push the compressor from the copied track a bit more (still not that it's pumping like hell), your tracks sound even more "in your face".

After that, the "clean" and the "compressed" channels are routed to a group channel. From there you can apply other effects as needed. This is all there is to the motown effect. Some are even repeating this technique more than just once.

GENERAL HINTS 3 - EQ?! Eh... Q!

EQ is very important in shaping a sound, but also in fitting your vocals into a track. The cleaner your recording, the less EQ you need and no harmonic exciter either, unless you want that sound and the original recording was fairly dull from being encoded into mp3 or something like that.

However what I found out that in the last years, the "uber-crisp" sound doesn't come from "highend hardware". Hell no, it comes from "overusing" the effects. Which means... adding 1dB to 2dB extra in the treble section (10kHz and up). This way you overdo it, but this is unfortunately what seems to be the trick to get that razorsharp pop sound lately. One hint from my side... if you want to do it, do it subtile. If you have a track that's rolled off at 250Hz for example, and has a boost of like +1dB from 10kHz ongoing (cause the recording was fairly okay), then try to reduce the bass in your arrangement, rather than pushing the vocals more. And if you want to push, then only subtile (0,2dB/0,3dB - something like that).

Nowaday pop-productions are so squashed to death that it's no fun anymore listening to the stuff without getting headaches. Don't listen to others in terms of overusing cause it's "in", develop your own sound.

GENERAL HINTS 4 - JUST OVER THE TOP

This is a fairly simple, yet hard to archieve tip: Keep your vocals just over the top of your arrangement. Even here, it depends on the track. Orchestra and solis can use a more balanced "blending" of the vocals into the arrangement, while pop productions need to have vocals that "sit" on the music, rock uses a technique of both, and hiphop is like "ah fuck it, as long as the beat kicks some ass, I'm game".

A standard since the early 40ies and 50ies was to make the vocals at least 2-3dB louder than the band. This technique is still used today, but doesn't always work. Automation, careful listening to the music parts and waging the balance of the track compared to the vocals is important. Finding the right "spot" is not easy, this is why most of the tracks are taking so long in terms of engineering.

A general hint from my side: Take your time, play with the automation (volume and even the send FX ammount) and you can create interesting soundscapes rather than boring, compressed and mudded up vocals. Finding the balance is the most difficult thing while engineering.

I hope I could offer you some interesting material/hints/aspects. I also recommend to get the last three issues of Future Musik (UK Version, issues 172 to 174) with it's "50 tips and tricks for recording/engineering" ("Make better music" section), as they give even more general tips and tricks than I just did.

[Nicole Adams]

Not a tip to do with EQ or compression, but doubling (and sometimes even more) your voice to fill the stereo field is something to do. The additional vocal tracks should be sung without--shall I say--accenting certain letters that you normally would in your main vocal track. These letters being "t", "s" and "b", for example. Also, vocal effects should not be applied to the main vocal track.

[Tyler Heath]

in working with FL, my personal experience (which is probably wrong and/or juvenile when compared to "real" recording and producing) has shown that a chorus module turned down to about 20-30% in the mixer can do wonders for making the track less "I recorded this in my room at my parents' house :x"

[k u n g f u c h i c k e n]

I love you compy.

[Navi]

Also, vocal effects should not be applied to the main vocal track.

When you say this, do you mean that they should rather be applied to send tracks?

Also, thanks so far to everyone who has contributed, especially Compyfox.

[Nicole Adams]

Also, vocal effects should not be applied to the main vocal track.

When you say this, do you mean that they should rather be applied to send tracks?

No, I mean apply effects to a second vocal track (a completely new vocal file). For example, you have your main dry vocal track, your second vocal track (which is almost identical to your first) that has the effects applied to it and then you can have a third track that harmonizes with the first two vocal tracks.

[Compyfox]

Also known as "ad lip's" and "vocal stabs".

[Yoozer]

ad lib

[Compyfox]

Okay... one d less (just fixed that), and here's the copy from Wikipedia:

In music, the term ad lib is used in a similar way to mean an improvised passage. It is also an instruction found in sheet music; see ad libitum.

see ad libidum

[Navi]

one question. sometimes, in rap songs, there's a slight distortion in the vocals, particularily if the rapper is raising his voice. I assume this is distortion caused by specific compression or maybe the tube?

http://www.mythrilnazgul.com/odbsample.mp3

for example

[Compyfox]

In this example, there were either the vocals recorded totally overpowered (micamp too loud) or distortion used on purpose.

[Navi]

It's a common thing i notice in a lot of the alternative hiphop/underground sound - not heard much of it in mainstream rap (production is too slick, i guess).

[Compyfox]

Well underground stuff isn't usually produced in "big studios", but more like living rooms. Especially newcomers like to overuse certain things, or just don't have the equipment.

That recording for example sounds like a multitrack recording from a Live Gig (i.e. battle). In the studio, you usually wouldn't go for that sound as labels would be like "no - we don't release that". Unless it's really done on purpose. Then however the distortion is more subtile (saturation).

[Navi]

Well underground stuff isn't usually produced in "big studios", but more like living rooms. Especially newcomers like to overuse certain things, or just don't have the equipment.

That recording for example sounds like a multitrack recording from a Live Gig (i.e. battle). In the studio, you usually wouldn't go for that sound as labels would be like "no - we don't release that". Unless it's really done on purpose. Then however the distortion is more subtile (saturation).

Well, to refer back to the example (ODB's rap from the latest Ghostface album), I hear it a lot specifically in Wu Tang albums - and I wouldn't say the Clan are basement recorders or newcomers to the game. I'm leaning towards it being more intentional, albeit slightly more noticable than usual. I'll try and pull up some more examples tonight.

[Compyfox]

Then it is to push the "agressiveness" of the track on purpose. Overdoing the saturation, etc. Like I said in the post with the "general hints", as long as the beat kicks, the vocals aren't that important.

[Flashbulbs]

True Wu Tang Clan are vets in the industry, but have you ever a\b compared their albums to other major-label releases? Wu Tang Clan's recordings can be very similar to garbage on this scale.

[Navi]

I wouldn't call it garbage. Rather, it's more raw and (intentionally) lofi than the shiny pop-rap recordings.

As you can tell, I'm a wu tang nerd.

[k u n g f u c h i c k e n]

Wu Tang started off with a low quality recording setup. As they moved on they kept that raw sound despite more expensive equipment. Intentional.

[Flashbulbs]

The being intentional part was obvious... of course they continued that way puposefully. That is their sound, but their aren't tons of major-label rap releases that sound "raw" like that intentionally. Not a lot of artists that are wanting major label attention and come to me and say "hey can you make me sound raw and sometimes on the side of lo fi?" (beside the random wanted effect in a few cases such on the production sid e of things) Anyway, I don't believe there is much need to further explain. I wasn't trying to offend anyone. Cause WU TANG CLAN AINT NUTTIN TO F&*K WIT!

[Navi]

You produce?

Got some samples?

[Flashbulbs]

no clips online and I wouldn't use an mp3 as a reference.