The Loudness War: A History of LUFS—and Why We Still Master Loud

Christos Tsantilis

5 hours ago

4 min read

Put two songs next to each other. If one is louder, it often wins the first ten seconds. Turn them up to the same perceived level and the conversation changes: which one has punch, depth, and a chorus that actually lifts? That little experiment explains decades of the loudness war—and why the fight is still going in 2026.

Before the wall of sound, there were physical walls

Vinyl and tape could be pushed, but neither let you ignore physics. A hot vinyl cut might create tracking problems or force compromises in running time and low end. Tape gave us a different set of tradeoffs: saturation can be glorious, but it can also become distortion you did not ask for. Mastering engineers were already making records feel exciting; the medium simply made them negotiate with the limits.

The CD changed the negotiation. Digital audio still had a hard ceiling at 0 dBFS, but engineers could use compression and, later, brickwall limiting to keep more of a song near that ceiling. Peaks got smaller, average level climbed, and a record could jump out in a quick comparison. Then its neighbor got louder. You can see where this is going.

The arms race

By the late 1990s and 2000s, loudness had become a competitive request. A powerful mix and a dense master are not automatically bad—some music thrives on intensity. The problem was treating every remaining bit of dynamic range as an obstacle. Drums lost their initial crack. Choruses had nowhere to grow. You could make the waveform look impressively solid while making the record feel surprisingly small.

Peak and RMS meters told part of the story, but neither was a great stand-in for perceived loudness. A peak meter shows the tallest instant. RMS gives a broad average of signal energy. Neither can tell you whether the vocal feels forward, the kick has impact, or the listener wants to hear the song twice. The meter was useful; it just was not the whole audience.

Enter LUFS

Broadcast engineers needed a better way to keep a program and its commercial breaks from lurching in level. ITU-R BS.1770 gave us a standardized loudness measurement using frequency weighting and, for integrated measurements, gating of very quiet material. LUFS—Loudness Units relative to Full Scale—became a far more useful way to compare program loudness than staring at peaks alone.

Useful is the key word. Equal LUFS readings do not guarantee two songs will feel identical. Arrangement, spectrum, transients, and the way a mix develops over time still matter. The ears remain on the payroll.

Streaming changed the incentive

Playback normalization turns a loud master down on services where it is enabled. Spotify, for example, describes a normal playback setting around −14 LUFS. That number is a playback reference, not a law requiring every stereo master to arrive at −14. Apple’s Sound Check likewise uses loudness information to reduce jarring jumps between songs. Settings, devices, and listening contexts vary.

Here is the practical consequence: if I destroy a snare transient merely to gain another decibel on the master, a normalized stream can take away the level advantage while leaving the damage. That is a lousy trade. On the other hand, turning a dense, exciting mix down to meet an arbitrary number does not magically restore dynamics either. I master for the song, then check how it behaves at matched playback level.

So why are some 2026 masters still around −7 to −6 LUFS?

Because the war did not end with a peace treaty and matching T-shirts. Some productions are dense by design. Some listeners turn normalization off. DJs, downloads, clubs, and other playback paths do not all behave like a normalized playlist. And sometimes the controlled aggression of a loud master is genuinely part of the artistic sound. There is no virtue in making an energetic record timid just to satisfy a chart.

But there is a difference between choosing density and chasing a number because somebody else’s record measured louder. At −7 or −6 LUFS, I want to know what the mix is gaining—and what the kick, vocal, and chorus are giving up. If the song gets more exciting, great. If it only gets flatter, the meter is winning an argument the music never started.

Atmos asks a different question

Dolby Atmos Music has its own delivery framework. Dolby specifies an integrated loudness target of −18 LKFS and a maximum true peak of −1 dBTP for Atmos music mixes. That is a format delivery requirement, not a stereo mastering target. Spatial contrast, movement, and depth deserve room to work; squeezing everything for raw level can reduce the very experience the format is meant to create.

My rule in the mastering room

Level-match before judging. Listen to the transient, not just the meter. Check true peak and the actual delivery requirements. Let a chorus earn its lift. If a loud master serves the record, I will make a loud master. If backing off half a decibel lets the groove breathe, I will take that half decibel every time.

LUFS helped us measure the problem, and normalization changed the economics of it. Neither can make the artistic decision for us. That is still our job—and, happily, the interesting part.

— Christos Tsantilis

MixByStos / StosLabs · Brooklyn, New York

MixByStos

CHRISTOS “STOS” TSANTILIS