How to Get that Perfect Mix

November 17, 2008 8:18 AM

So here's the thing: Finally over the Veteran's Day holiday, my wife and I wrapped up the recording phase of the new project we've been working on for Tangemeenie. Now comes the hard part: Mixing. All along, I've been haphazardly churning out rough mixes, a few of which I've posted here in the past. But now, on closer listen, it's clear to me that I abused limiters, compressors and other signal processors and generally made a mess of things in so many minor and major ways in these initial mixes that I now need to do almost a complete reset. So now, I'm eating humble pie. MeFiMu mixing gurus--here's your chance to share your favorite lessons learned or any other pieces of practical advice you might have about the art and science of mixing.

I'm mostly interested in nuts and bolts technical tips and any rules of thumb you've found helpful in the past. The focus is meant to be on getting clean, problem-free mixes that are present but not over-compressed.

After putting my ignorance on display in this previous thread a while back, that discussion got me started thinking a little more about just what I actually do know and understand about mixing, and it's become more and more apparent to me the answer is not much. It seems I don't even know my own recording tools very well. For example, contrary to what I used to think, of course the mixer in Nuendo has effects sends and returns like any other mixer. All this time, embarrassingly, I've been using only insert effects because I just didn't know how to access the effects sends (Nuendo is a relatively new tool to me, and my habit is always just to kind of dive in and learn things as I go, which usually works out fine--until I start getting cocky enough to think I actually know what I'm doing).
posted by saulgoodman (16 comments total) 8 users marked this as a favorite

It all depends on what sort of music you're making. I really do have an awful lot of compression on my vocals on songs where the guitar is intended to be this massive wall of noise -- otherwise the vocals would swim in and out of the mix.

But if I were doing something like a folky acoustic track, I'd be very, very easy on the compression in the vocals but perhaps drive the compressor input from the guitar track a little bit to even out the attack. In general, mix all your tracks dry and uncompressed and get the best sound out of it, and then apply "expansive" efx like compressors/limiters/reverb/delay as a surgical tool. Also, do you use any autotune or pitch correction on the vocals? It sounds like overcorrecting may also be contributing to the problem.

For example, on a track like "The Book of Love" you've got WAY too much compression on the vocals. In a song like this, I would work on the mixing with no compression at all, and if there are portions where one track really overwhelms the other parts, I'd compress it a bit with the threshold around where your other peaks are. In general you would be much happier on this song to try a very soft knee on the compressor and a medium-long release. It sounds like the release is very short so there's almost a weird tremolo-ey sound on the vox?

As for a song like the gilded cage, you've just smooshed all of the wonderful dynamics out of the keyboard arpeggio they don't really "breathe" in comparison to the mellow and steady guitar phasey sounds, which will simply due to the distortion have a limited dynamic range.

With a song that has a lot of instrumentation like this, it's not too bad to have enough compression on the vox (in this song, you may want to pull the vox drive back a little bit but it's not bad IMO) to keep it from being lost in the mix in some words. I don't mean to be harsh, but you really killed the drums here, you can tell especially in the transition to "ain't it stray-ee-ange". In this part of the song you want to hear a bit of a crescendo, a buildup of power, but since the whole mix is pushed so hard it just sounds like another part (with a weird overcompressed drum mixed over on that right channel. This is the sort of mix output you would hear from a radio station with "squash everything into the last 0.1db" setting. In this song the bass is mixed too high.

Anyway, just a couple tracks' suggestions. Take it for what you will, as I have a definite tendency to use a lot of compression.

For what it's worth, each song will "tell you" how it needs to sound. Let a quiet mellow track be quiet and mellow. Worry about driving it louder only when it's time to put it onto an album. But take my advice with a grain of salt as I've been accused of being a compressor-fiend.
posted by chimaera at 9:26 AM on November 17, 2008 [1 favorite]

Read this, all twelve sections. Then read it again. Even though you don't use Pro Tools (neither do I). I still make sloppy, unbalanced mixes, but there's a lot more focus now than there used to be, even on one-off roughs. Dye's walkthrough of his process really made a huge difference in mine, even on non-rock tracks. Every plugin, every tweak, has a purpose.
posted by uncleozzy at 9:27 AM on November 17, 2008 [5 favorites]

Also, do you use any autotune or pitch correction on the vocals? It sounds like overcorrecting may also be contributing to the problem.

We don't ever use pitch correction (which I'm flattered you couldn't tell), but we do sometimes double the vocals, and when I tracked the "Book of Love" originally, I think I applied a pretty hefty dose of compression to the signal before it went to tape. Back then, I had no practical experience beyond recording/mixing lo-fi to portable 4-track or 8-track.

For example, on a track like "The Book of Love" you've got WAY too much compression on the vocals.

Yep--am starting to realize I've been doing that kind of thing all over the place. Thanks for the notes on Gilded Cage; I agree. That one's meant to be on the record, so it's important to eventually get it right, and I already started giving it a serious reworking last night. Keep 'em coming...

Thanks for the link, uncleozzy. A lot of good (non-Pro Tools specific) stuff to chew on there. One article mentions the Renaissance Compressor, which is quickly becoming one of my new best friends.

More generally, I'm especially interested in signal flow. For example, where in a signal path compression or other processing seems to work best for a particular purpose; when to use sends vs. inserts, pre-fader/post-fader, etc. A concise list of common mistakes to avoid would also help.
posted by saulgoodman at 10:39 AM on November 17, 2008

The best advice I ever got was along the lines of:

The first step in mixing is to do nothing.

Set all the faders to unity. Leave the pan pots where they are. Turn off all your signal processing, and just listen to the mix.

One of the hardest parts about mixing is not being overwhelmed at the number of choices and techniques at your disposal. The best approach is to methodically work your mix from the ground up. I generally follow this generic method (which may vary depending on the music you’ve recorded):

Start with the faders, and don’t touch any other knobs until you get a good mix with volume alone. Next, move to the pan pots, place your sounds in the stereo field, and adjust levels if necessary. Third, add as little reverb as you can to eliminate any sense of claustrophobia.

Now listen to your mix again. The best mixes cater to the sounds already present in the tracks. This is frustrating but true - if the sound you want is not there, on a fundamental level, you cannot fix it in the mix. Aside from a featured “cool effect” on a guitar track or two, all additional processing should just tweak the mix (in a perfect world, anyway!). Less is more, and small changes (just a shade of EQ, just a touch of compression) can really make big differences.

In my limited experience, the question is not where to use compression (always in-line, inserts, in the way of the signal), but rather when to use compression (as infrequently as possible!). A step-by-step approach will prevent you from saying at the outset, "Well, my guitar sound isn't snappy enough, so let me start by compressing it." And don't forget that (a little) compression on mixdown to two tracks can really bring a track alive and help the instruments to gel.
posted by abc123xyzinfinity at 11:40 AM on November 17, 2008 [1 favorite]

Yeah, less is more. If, when thinking about mixing, the first thing you consider is which plugins to use, you're doing it wrong. Step one, faders to unity (actually, I usually set them to -6 or -9dB), step two, balance, step three, fader automation.

That said, before I start mixing, I usually set up my submixes (drums, guitars, lead vox, BG vox, sometimes one or two more) and effects (short delay, long delay, short reverb, long reverb, some kind of modulation effect, sometimes a stereo detune-spread). I don't commit to any sounds, but it gives me a place to start so I'm not thinking about it while I mix.

One thing I do use judiciously, though, is subtractive EQ. Make holes in your guitars for your vocals, in your pads for your guitars, in your kick drum for your bass guitar, and so on.
posted by uncleozzy at 12:48 PM on November 17, 2008

uncleozzy, how do you decide how big of a hole you make while tweaking the subtractive EQ? Tell us more about the process.
posted by umbú at 2:07 PM on November 17, 2008

I can't speak for uncleozzy but when I use subtractive EQ for giving instruments their "place", it's using the parametric EQ that's in SONAR and I have a wide Q around, say, 200Hz and decrease the level there by maybe 2dB. Doing this on a guitar and snare/kick drums gives the bass a bit of room in that range. Think of where each instrument really "lives" -- a metal guitar will be in the 2k range a lot (and very scooped mids), a kick drum you'll apply a low pass filter and roll off anything above 150Hz unless you want that "quarter on the drum beater" sound, in which case you'll boost a bit above 4k. I'm sure uncleozzy can be more thorough, here.

The other reason i use selective EQ is to make up for unpleasant peaks in vocals. I sometimes roll off a little bit in the 200Hz or 500Hz range, but I tend to use a very sharp Q and be very selective. Use a sharp Q and boost that frequency and if the thing that bugs you gets louder, cut that frequency by a lot. Sometime I have a very, very tight Q and cut as much as 10dB in extreme cases.
posted by chimaera at 5:21 PM on November 17, 2008

Don't go nuts with effects/compression/EQ. Stuff should sound pretty good flat at first. If it doesn't, you have shit coming into your inputs... I.E. crummy sounding instruments/voices.

I can't stand it when people try and EQ the hell out of my expensive guitar (with its expensive on-board valve pre-amp). It sounds best FLAT. Warm it up after you get it in, if need be.

Study everything Brian Eno has to say on the subject, especially his oblique strategies. If you have stumbled over a mixing problem, he has already:

1. solved it
2. articulated how to solve it
3. published the solution
4. written a zen koan
5. made everyone else look foolish
posted by chuckdarwin at 2:10 AM on November 18, 2008 [3 favorites]

I make pretty drastic EQ cuts, actually. Logic's Channel EQ is pretty clean, so you don't get too terribly punished for it. I high-pass lots of stuff (more than I ought), and I usually low-pass electric guitars too (mostly because amp sims are fizzy up top). Other than that, it's a game of sweeping little peaks and valleys around. Sometimes, I'll admit, I use my eyes instead of my ears to find fundamentals and big harmonics, by turning on the real-time EQ display, but more often than not, I just pull stuff out until suddenly something else sounds sharper. I mean, like chimaera says, you get a sense of which instruments live where, but more than not, every mix is different.
posted by uncleozzy at 5:33 AM on November 18, 2008

I think it really matters what you actually want the final product to sound like. do you want a 'superclean/super perfect' psuedo pop type sound or do you want something a bit dirtier..

one of the most interesting albums i've heard this year was MGMT - Oracular Spectacular. and well it kinda sounds pretty shit. dirty distorted and over-compressed. but it gives it character in a way.
posted by mary8nne at 7:58 AM on November 18, 2008

and well it kinda sounds pretty shit. dirty distorted and over-compressed. but it gives it character in a way.

That's a good point. Growing up, a lot of the earliest music I liked was punk--stuff like the Cramps, old DRI, Circle Jerks, Minor Threat, Misfits. If that stuff had been well-mixed, by conventional standards, it probably wouldn't have been half as memorable or interesting. Same for a band like My Bloody Valentine (although I guess the difference there is that the over-compression and other normally undesirable features of those mixes contribute to the overall mood in a very deliberate, controlled way). I guess that's part of the art of mixing, as opposed to the science.
posted by saulgoodman at 12:49 PM on November 18, 2008

Use high quality "channel strip" plugins and play with the presets. I very much recommend the Waves SSl bundle as a cover-all solution.
posted by dagosto at 3:13 AM on November 20, 2008

My advice is to be realistic about how much change you can, or should, introduce in the mixing process.

While some people view mixing as 50% or more of the process of taking a song from take 1 to a mastered final version, I think mixing should aim to be an exercise in tweaking, rather than building.

The song should be able to stand on its own even as a rough and dirty mix; if not, then maybe you need to go back and start from scratch. Fixing it in the mix is the wrong way to approach recording, eats up enourmous amounts of time, and is often futile.

Having said that, for practical tips, such as typical EQ cut and boost points, I recommend the Mixing Engineer's Handbook, which is evenly split between useful advice and interviews with respected and wise engineers who provide a lot of insight.
posted by Paid In Full at 1:28 PM on November 20, 2008

I've been producing music for less than year, so my advice is probably not the best, but here's what I've learned so far.
I think the best way to approach a mix is to totally disregard all that extra signal processing stuff and first work only with the faders. When everything sounds balanced to you, you can start fixing all the little things that require the signal processors and other controls such as pan. You're probably going to have to constantly readjust faders to compensate for the volume changes that will result from these processors.
I agree that before you do anything you should listen through the song a few times and take notes (mental or physical) about what sounds wrong and what sounds right so you have some idea of where you want to go with the mix. (I actually don't use this technique much...because most of my music is produced on my laptop, I generally mix as I go)

I also agree that to make a good mix you have to start out with good parts. I think the phrase that is thrown around a lot is "polishing a turd". In the end though, what's most important is how it sounds, and not what you've done to it.

And almost every track (excluding bass & kick) should have a low cut filter!

And personally, I'm not a fan of music that sounds really clean and well produced...I think some dirt and distortion makes a track sound more organic and human.
posted by god particle at 5:22 PM on November 22, 2008

Mix early and often.
posted by dagosto at 2:34 AM on November 25, 2008

chimaera: here's a new mix of gilded cage in which i tried to take your suggestions into account. still a work in progress, of course, but i think it's already shaping up some, at least in terms of overall clarity.
posted by saulgoodman at 7:10 AM on November 25, 2008

« Older Re-ordering MeMu playlists?   |   Live Sound Perfection Newer »

You are not logged in, either login or create an account to post comments