Rules for mixing

Submitted by mic on Mon, 11/20/2023 - 11:20

Here are my most important mixing rules.

Take mixing one step at a time

Understand what you want to do and why you are doing it. If you a solving a problem in the mix, understand what that problem is. Take care of one thing at a time.

I just spent a day with a guitar track that had too much sustain and was dry and muddy. I got rid of the sustain, checked the mid to low end, added sizable reverb, and tried different panning. None of that worked. I realized that the only thing bothering me was a sustained ringing around 3.5 kHz. I added a slight EQ (/wiki/equalizer) notch around that frequency. This allowed a more interesting reverb. The sustain no longer bothered me, so compressors were not necessary. With the new reverb, the low end was not as much of an issue. I tried many things, until I understood that I only needed a slight EQ.

In one track, the only change I made to a vocal was to drop 1.5 dB from the first two syllables. In another, I tuned one syllable. In another, I simply moved a guitar to the middle. If you know what you want to accomplish, drastic changes to the mix are often not necessary.

To be sure, your song will always be very "produced". You will use compressors, equalizers, reverbs, automation, and so on. Most tracks will go through some significant production. Your DAW will struggle. But unless you have a very clear idea of where you are going, it would be a good idea to get there slowly.

Target many devices with your mix

My mix must sound good on all devices. I only mix on one setup, but I test on many. I mix on headphones first (Senheiser HD 400 Pro) and move to monitors later (Thiel speakers through a Krell amplifier). The monitors are properly set up in a room with some acoustic treatment. I have a good setup, but the mix is not just for that setup. It is checked exhaustively, everywhere, because I never know what my listeners are up to.

To check the mix, I use different headphones, including simple earbuds. Those accentuate sibilance. My car speakers are bad but help me hear problems with the low end. My phone speakers are even worse, but they accentuate vocals and help identify compression and gain staging problems. You have options and you might as well use them. If I hear something I don't like, I revert back to my good setup and decide if something should be done.

Mix with reference tracks

You need a reference song. But you don't always need a very similar song. A song in the same genre that you like and that sounds good on many speakers is good enough.

A reference song helps to take a break from listening to your own song and to imagine what your song could be. Then, there are obvious easy comparisons to make, such as gain levels and frequency content. There may also be new ideas: perhaps the drums and vocals do not need to be so pronounced, more reverb is OK, and sending a crash to the far left is acceptable.

Not everything in the mix needs to be heard

Some tracks are in the mix simply to color other tracks – some vocal harmonies, some guitar parts. These don’t need to be plainly audible. You will notice the difference in the mix with and without them, but the listener does not need to know exactly what they do.

Monitor at low levels

The ear works in mysterious ways. Things that sound good loud may not sound good quiet. In a quiet listen, vocal parts may come out, as our ears are tuned to pay attention to them. Other parts may fade away.

Take breaks from mixing

I don’t have a strict plan, but I take a break when the mix sounds OK or when I’ve spent an hour with the same song. It takes a while for the ear to become unprepared for the mix and notice problems that it would otherwise not notice.

Work with the song changes

Songs go through changes. Even if one track and one pattern repeat throughout the whole song, other tracks may make it sound different in different parts. Use your mixing tools to make that track and every other track work with the rest of the mix. You can adjust its gain, pan, and practically anything else throughout. You don’t need the same gain, pan, reverb, or EQ on different parts.

Don’t follow all rules

Some rules work most of the time, but nothing works all the time. Here are some examples.

  • "Your songs need headroom". This is a good idea, but a lot of 90s heavy metal had no headroom. Nobody noticed.
  • "Put a limiter on the master channel". That depends. A rock song, for example, will always have a couple of very short peaks that stick out above everything else, usually a kick or a snare. I would be surprised if a listener notices those two clips. Most importantly, DAWs have a workflow. They may mix to the master channel before applying a limiter. Clipping in the mix would have already occurred before the limiter. You have to set gains appropriately low, then limit, then add gain, which often makes sense only if you are looking for a level of loudness or dynamic consistency.
  • "Use narrow bands for subtractive eq, wide bands for additive eq". This generally makes sense. But a high pass is a subtractive EQ and some high passes can be very wide. Some master channel EQ, especially to reduce low mid frequencies, can also be wide. If you are looking for a sizable effect, like a "radio voice", you are also stuck with wide subtractive EQ.
  • "De-es your vocals". Don’t de-ess your vocals unless they need it. De-essers, if they provide any flexibility, are hard to tune. You are more likely to end up with vocals with a lisp. If there are only a couple "s" that bother you, adjust their gain. It is a painful process of trial and error but produces good results.
  • "Compress everything or compress the drums". Compressing the drums does make them snappy, which could be good. It could also be bad if you are looking for a raw sound. If you are using drum samples, then compressing is probably not necessary. Those samples may have already been compressed and otherwise processed. Compressing everything else is a choice. Vocal style compression (e.g., fast attack) on heavily distorted guitars, for example, is probably unwise. There isn’t much amplitude movement in such guitars anyway. You’ll just end up ruining the initial note accent.
  • "Pan a certain way". There are no rules for panning. I am partial to clean LCR mixes, but I have tried other mixes too. Panning separates instruments, creates a left-right balance, and creates a more interesting sound picture. That left-right balance however depends not only on the pan, but also, for example, on frequency content. Thus, your panning depends on your EQ and your EQ depends on your panning. You will not get a good balance by simply following a predetermined panning structure.
  • "High pass everything". If you high pass everything, you lose your low end. Sure, I’ve used a high pass on many things, including bass, guitar, vocals, kicks, and snares. But my high passes have been gently sloping and often with cutoff not large enough to completely separate instruments. If you high pass your guitars at or above 300 Hz, you may hear them cleanly separated from the bass. But if you high pass them at 200 Hz, they may give your mix and the bass guitar exchange more power.
  • "Side chain the reverb". This is certainly a great way to ensure that the reverb does not interfere with the dry vocals. Of course, this makes sense with longer, more pronounced reverbs and with instruments that need clarity. But you probably don’t need that for short plate reverbs on snares.
  • "Double the guitars, one in each channel, with some delay in one of the channels". That is a great effect for thick, full guitars and for separating the guitar from the bass without simply sending the guitar left or right. I do this very often, but not always. This is problematic where you need sizable low end and probably anywhere where you get bell-like overtones, as with some clean guitars.

Watch for the following

The following are not rules. These are the things to which I always pay attention. They are the typical mistakes I made in the past.

  • Drum levels. I like drums, but there is rarely a need for the drums to exceed the rest of the mix by more than 2 dB. In many cases, drums can even stay below the mix. A reference song will show this quickly. The usual culprit is often the kick, sometimes the snare. The high hats and other cymbals can also stay quiet. They will carry the rhythm either way, even if barely audible.
  • Vocal levels. Similarly, there is no reason for vocals to carry the whole song. This is especially true for rock songs from a couple of decades ago. Vocals, in fact, can be quite recessed. Since we, as listeners, are naturally inclined to pay attention to vocals, we will recognize them either way.
  • "Air". The one great thing about good monitors is how easy they help us understand whether the song sounds dark and closed up or open and breathing. You can recognize whether your song is missing anything above 10 kHz.
  • Low end. I make sure the low end is there. In the past, my high pass filters were sharp. Today, they are not. Often, I replace high pass filters with gentle equalizers.
  • Gain staging. It is not enough to compress the vocals. Even with even dynamics, they may still pop in and out depending on the song. It is OK, for example, to drop a 2 dB from a forceful chorus vocal that comes in after a calm verse, even if it is just at the beginning of that chorus and even if that drops the chorus below the verse. We don't always perceive sounds as flat, even with serious compression.
  • Vocal tuning. Suppose that you compress your vocals, reverberate them, and get the gain right, but the vocals still do not fit well in the mix. My first assumption is that they are out of tune. Often, this means only syllable here and there, but it is worth a look.
  • Unfortunately, this problem should have been solved much earlier. First, you must work out your vocal melodies. Put them in MuseScore or play them on a guitar or piano. An impromptu vocal line could be great, but often will contain a wrong note here and there or the right note in the wrong place. Second, if you are recording a song, help the vocalist. Add a throwaway guitar line that replicates the vocals as well as chord accompaniment. Singing well without accompaniment is tough. I see this in hard rock and metal songs, where a singer has to follow power chords that are often distorted and sometimes missing the root, and generally don't provide a good tonal center.
  • Third, tune your vocals carefully. Unless the singer is well trained or super talented, there will be mistakes. But those mistakes are not throughout. Tune only what is necessary. And do not simply move syllables to the center closest note in your software. What if the closest note is not the right one? Not common, but it happens. What if there is a slight inflection in the middle of the syllable? This happens all the time with untrained singers and tuning software is not very good at figuring out what this means. What if the instruments themselves are off? 15-20 cents would make a huge difference. Tune what you need and tune it to what sounds good. Often, this simply means tuning the start of vocal lines, as it is difficult for many singers to pick up the right note straight for the beginning. I often suggest a throwaway accompaniment that starts earlier than the verse. Sometimes, this means tuning the ends of vocal lines, as that is what listeners notice.
  • Dynamics. Songs without dynamics are boring. My first surprise when recording was just how big the difference can be between song parts, especially in drum tracks. The drummer, and the song, moved up and down by about 5 dB, between the verse and chorus. If you want to mix for specific LUFS, do so. But that doesn't mean that the whole song should become one even blob. If the song has changes, dynamics are a great tool in your toolbelt. This is typically where engineers get lazy. The song already sounds good and who knows what streaming platforms adjust the volume to. But, if you can make the song even better with dynamic changes, you should do so.

authors: mic

Add new comment

Filtered HTML

  • Freelinking helps you easily create HTML links. Links take the form of [[indicator:target|Title]]. By default (no indicator): Click to view a local node.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
Enter the characters shown in the image.