Kind of like every human being has a height value, every audio clip has a peak level value. So what can we do with that information, beyond knowing that going above 0dB usually isn’t advised? In our eternal quest for ultimate audio quality, the inclination can be inherent to record loudly—as close to 0dB—as possible, and thereafter maintain that peak level. We fear that by mixing with track levels that are too quiet, we might be losing fidelity, some harmonic detail in the saturation floor or something.
While it is true that recording analog signals as loudly as possible without signal clipping to begin with will indeed minimize noise floor, the benefits reaped by maximizing peak level for individual tracks tapers off as you get closer to zero. At what peak level are audio sources louder than they really need to be? At what level are they too quiet, that they might need to be boosted excessively later? Using an algorithm based on the energy of how sounds stack together, i devised a set of go-to ranges for peak levels based on track counts. Since peak levels can vary fairly wildly based on content, you are given minimums and maximums (instead of single target values). Juicy details ahead…
Pretty much all current digital production strategies work with relatively excellent signal-to-noise ratio. These go-to peak levels will provide a fulcrum around which to balance your mix to taste by pushing or pulling individual elements, providing you with creative freedom but keeping things from getting way out of whack.
Lowest Safe Peak Level—Probably Lower Than You Think
Minimum Safe Peak Level (Arbitrary # of Tracks)
Based on my custom calculations, it’s generally safe to have the peak level for a single track be as low as -26dB. While it’s instinctual that a louder initial peak level reduces the potential for low-level noise increasing later, the returns diminish as you get closer to 0dBFS, providing less benefit (to S/N) than risk (of clipping).
Recommended Max Peak Level (Arbitrary # of Tracks)
Pushing the peak level for a track in a mix (assuming it has at least 2 tracks) to greater than -13dB or so… is basically pointless.
Modest Levels for Creative Flexibility
Keeping track peak levels at modest levels prevents your mixbuss from getting overloaded too soon, giving you plenty of headroom for playing with the levels of elements in a mix, adding parts, applying effects, and layering in auxiliary return tracks. Of course, you can always plan to avoid downstream overshoots using various toys like limiters, clippers, maximizers, etc, but using the strategy of working with relatively low peak levels in the first place ensures ample headroom to not need to, instead keeping those things as optional creative choices rather than as irrevocable necessities.
…So without even thinking about track counts, the general recommended target for the max peak value of a single audio track (or consolidated stem) is between -26 and -13dB.
Too Loud? Too Quiet?
If most of the channels in a mix you’re working on are peaking above -13dB, you’re likely to benefit from increasing headroom by reducing all track levels equally. Likewise, if every channel is peaking lower than -26dB, it’s probably a good idea to bring everything up a bit (depending on the track count, of course).
It is expected for some tracks in a mix to stray from these ranges… for example, an auxiliary room reverb meant to only inject a touch of subliminal space is probably not meant to peak near as loudly as the instrumental tracks it’s sourced from. Likewise, several instruments summed together in a group may intended to be the same relative level as another single instrument (in those cases, treat the group as a single track). If certain tracks here and there stray from the ranges, don’t worry about it, as long as they’re balanced as you like in the mix.
Recommended Peak Levels Based On Track Count
The following spreadsheet gives values for my custom calculations of Safe Minimum Peak Levels and Recommended Maximum Peak Levels based on specific track counts (with values rounded to 0.5dB). Keep in mind that these are loose guidelines; individual tracks can and should vary based on their prominence in an arrangement.
When to check peak levels? I’d recommend checking two main times during the process of mixing. Initially, right at the beginning during “faders up” pre-mixing to get the levels at a decent starting point. Secondly, later on during mixing, especially after having applied compression, transient shaping, etc to tracks, as these can greatly affect the resultant peak levels. It may also be useful to check peaks after any heavy usage of effects. Handily, Ableton Live provides peak level meters which show the max peak values of each track for the most recent playthrough.
If you aim for each track’s max peak value to land somewhere between its safe minimum and recommended maximum, you’ll be within a usable, flexible ballpark. Remember, these are not rules to be followed strictly, but rather guidelines to help out if you’re having trouble sorting out your levels. If a track is peaking within its target range but still sounds too quiet in a mix compared to everything else, that indicates it likely could use some saturation, compression, and/or limiting to bring down its peak level compared to its body… or it might be being frequency-masked by another element.
Check out the table below for peak value ranges based on song track count. If your exact track count is not listed, extrapolate an estimation based on adjacent values.
Track Count: Number of tracks in the song. The value “1” is useful for assessing the master channel or pre-master mixdowns.
Safe Minimum Peak Level: As long as individual tracks hit this mark you shouldn’t have any concern about any signals being “too weak”, freeing you to focus on the overall balance of the piece.
Recommended Maximum Peak Level: Try to avoid individual tracks from peaking louder than this value. Don’t worry too much if an important track (like a lead vocal) sometimes breaks the rule.