Every track is mastered in the context of other tracks, and the context pushes in a specific direction. When listeners compare a new track against what they have already heard, the louder one feels better, and the incentive to be louder than the last one is difficult to resist.

How the incentive works

Loudness perception is dominated by level. Two otherwise identical masters, one three decibels louder, and most listeners will prefer the louder one and some will not be able to identify a difference. This is a robust perceptual fact and it is what makes the competition self-reinforcing.

Nobody deliberately sets out to produce a worse master. The process is incremental. Each track is mixed louder than the last because that seemed marginally better in comparison, and the reference point drifts. After a decade the whole catalogue has moved, and nothing sounds wrong with any individual track.

What it costs

The currency is dynamic range. Making a track louder requires compression and limiting, and both reduce the difference between quiet and loud parts. There is a point where a mix stops communicating its own structure, where the quiet passage before a drop is no longer quieter than the drop itself.

The audible cost is not distortion, because modern limiters are transparent at moderate settings. It is a loss of impact, since everything sounds like the loud part. A listener may not be able to say what is wrong, but the track fatigues faster and communicates less.

Streaming changed the arithmetic

Loudness normalisation changed this materially. Most platforms measure a track's integrated loudness and turn it down to a target before playback, so a track mastered dramatically louder than the platform target is attenuated before anyone hears it.

This means the loudness war is largely self-defeating on streaming. The compression spent getting a track 6dB louder is removed before playback, and the track ends up quieter than a competently mastered one while having lost all of its dynamics in the attempt.

Defensible targets

For most streaming destinations, integrated loudness somewhere between minus 9 and minus 11 LUFS is a reasonable range, with the lower end giving the platform more headroom to work with. True peak around minus 1dBTP avoids clipping in the reconstructed waveform.

Within that range, the correct approach is to use the reference only to establish a level, never as a target to exceed. Match the loudness, then judge the mix on its own terms.

Mastering to a target rather than to a rival

1

Choose your destination and set a target in advance. Streaming and club are different targets, and pretending otherwise produces a compromise that suits neither.

2

Mix to the target, not past it. If the mix reaches minus 10 LUFS integrated naturally, stop. A limiter is a tool for catching peaks, not for adding loudness.

3

Check the reference at matched loudness, then stop looking. The purpose of a reference is to catch an error in your own mix, not to provide a level to chase.

4

Preserve the quiet parts. If a section is too quiet, fix the arrangement or the mix. Raising it with a compressor removes the contrast that made the section work.

Loudness targets by destination

DestinationIntegrated targetReason
Streaming, generalMinus 9 to minus 11 LUFSNormalisation headroom
Club / festivalMinus 8 to minus 6 LUFSHigh SPL systems, no normalisation
YouTubeMinus 14 LUFSPlatform normalises more aggressively
VinylMinus 16 to minus 18 LUFSPlayback level constraints
BroadcastMinus 23 LUFSStandard, with its own processing

Retuning and loudness decisions

A retune can change the loudness of a mix slightly, because denser low end interacts differently with a limiter. After shifting with 432Hz MASTER, it is worth re-measuring integrated loudness rather than assuming the master is still on target.

This matters because the temptation, having already done the work, is to correct the difference by limiting. Adding limiting to fix a two-decibel discrepancy is the wrong trade.

The bottom line

Set a target from the destination, mix to it, and use the reference to catch mistakes rather than to compete. The loudness war is largely over on streaming platforms, and continuing to play it costs dynamics for no gain.

Frequently asked questions

Is the loudness war still relevant?

Largely not on streaming. Platforms normalise to a target before playback, so a track mastered much louder than that target is attenuated and the extra compression is removed before anyone hears it. It still matters for club and festival playback, where no normalisation happens.

What loudness should I master to?

For most streaming destinations, between minus 9 and minus 11 LUFS integrated, with the lower end giving normalisation more headroom. For club systems, minus 8 to minus 6. YouTube normalises more aggressively and sits around minus 14. The target should come from the destination, not from rivals.

Does making a track louder actually sound better?

It sounds bigger and more present, which listeners often prefer. The cost is dynamic range, because louder requires compression and limiting, and both reduce the contrast between quiet and loud sections. The result typically fatigues faster and communicates less, even if individual defects are harder to identify.

Should I match a reference track's loudness?

Match it once to establish a level, then judge your mix on its own terms. A reference is for catching an error in your own mix, not for providing a level to chase. Using it as a target to exceed is how the loudness war continues.

More from the atixUniverse blog