The Loudness Wars

There is a moment in the opening of Brothers in Arms by Dire Straits — a recording often held up as a landmark of early CD-era production — where the guitars fade in from near-silence, and the quiet itself feels like part of the music. That silence has weight. It has tension. It makes the arrival of sound mean something. That kind of silence, that kind of contrast, became unfashionable. For about two decades, the record industry did everything it could to eliminate it.

The story of the Loudness Wars is usually told as a story about technology. It is also a story about fear, competition, and a collective decision made by thousands of individuals — each acting rationally by their own lights — that added up to one of the more self-defeating episodes in recorded music history.

Before the War

Vinyl records imposed hard physical limits on how loud a record could be. Cut a groove too deep and the needle would skip. Push the low end too hard and adjacent grooves would interfere with each other. Engineers worked within these constraints not reluctantly but creatively — the limitations forced decisions, and those decisions shaped the sound of recorded music from the 1950s through the 1980s.

Dynamic range, the difference in level between the quietest and loudest parts of a recording, was not just accepted but exploited. The gap between a whispered verse and a crashing chorus was part of the emotional grammar of a song. When Led Zeppelin dropped into the sparse verse of Whole Lotta Love, the restraint made the explosions more explosive. When Fleetwood Mac pulled back on The Chain before the bass riff hit, that pull-back was a compositional act.

Radio complicated things slightly. Broadcast engineers applied compression to even out levels across different records, and stations competed to sound louder than their rivals on the dial. But the compression happened at the broadcast stage, not on the record itself, and the original dynamics were preserved on the vinyl.

The CD changed everything.

The CD Opens the Door

Compact disc arrived in the early 1980s promising something closer to theoretical perfection: 16-bit audio, a dynamic range capability of around 96 decibels, no surface noise, no degradation over time. Engineers suddenly had headroom they had never had before. There was no needle to skip. There was no groove geometry to negotiate.

For a while, many producers used this freedom well. The early CD era produced some genuinely dynamic recordings — Steely Dan’s remastered catalogue, the audiophile pressings of Graceland, the early Peter Gabriel albums in digital. The format could handle the full emotional range of a performance, and some engineers were excited to let it.

But the CD also introduced something the vinyl era had never quite had: the ability to simply make the whole thing louder. If the limit on a CD was 0 dBFS (decibels relative to full scale — the digital ceiling), you could, in theory, push your average level closer and closer to that ceiling. A record that averaged at -18 dBFS and a record that averaged at -8 dBFS would both play without distortion. But the second one would leap out of speakers relative to the first. On the radio, on a TV advertisement, in a record shop, it would seem to hit harder. It would seem to demand attention.

And so the arms race began. Slowly at first, and then all at once.

The Compression Machine

The tools that made the Loudness Wars possible were limiters and multiband compressors. A limiter is essentially a compressor with an extremely high ratio — it catches peaks and prevents them from exceeding a set ceiling. A multiband compressor divides the audio spectrum into separate frequency bands and compresses each independently, allowing engineers to squeeze low, mid and high frequencies without the whole signal pumping and breathing in obvious ways.

Used carefully, both tools are genuinely useful. Compression can add energy, presence and punch to a recording. The problem came when they were used not creatively but defensively — not to shape a sound but to prevent any moment from being quieter than any other moment.

By the mid-1990s, mastering engineers were under explicit pressure from labels to make records louder than their competitors. The logic was simple and, within its own frame, not unreasonable: louder records seemed more exciting in quick comparisons. An A/B test of two mixes, with one sitting 3 dB higher than the other, will almost always cause the louder one to be preferred — not because it sounds better, but because our ears interpret loudness as quality. Labels knew this. They exploited it.

The result was a slow crushing of dynamic range across commercial music. Metallica’s Death Magnetic, released in 2008, became one of the most frequently cited examples of the damage this process could do. The album was mastered so aggressively that it regularly clipped — not just peaked, but actually distorted — and the Guitar Hero version of the same songs, which was mastered to a lower level for technical reasons, sounded dramatically better. Fans petitioned for a remaster. The band’s own music sounded better on a video game peripheral than on the commercial CD.

But Death Magnetic was not an outlier. It was an extreme point on a curve that included almost every major commercial release of the 1990s and 2000s. The Red Hot Chili Peppers’ Californication — another frequently cited example — is a record of considerable compositional intelligence that is physically unpleasant to listen to at high volumes because of the relentless brick-wall limiting applied at mastering. The dynamics that might have made those songs breathe were simply removed.

What We Actually Lost

It is worth being precise about what dynamic range actually does for a listener, because “dynamics matter” can sound like an audiophile abstraction rather than a real statement about experience.

Dynamics create anticipation. A section of music that is quieter than what surrounds it creates a kind of potential energy — the listener’s attention sharpens, the body prepares for impact. When that impact comes, it is felt physically as well as heard. This is not a metaphor. The startle response, the involuntary physical reaction to a sudden loud sound, is a real neurological event. Music that never varies in level never triggers it.

Dynamics also allow for intimacy. A vocal that sits in a genuinely quiet mix — think of the verses on Nick Cave’s The Boatman’s Call, or the quieter passages on Radiohead’s Kid A — requires the listener to lean in. That leaning-in is a form of engagement. When everything is maximally loud, there is nowhere to lean.

And dynamics are, finally, part of how we perceive emotion in music. The swell, the drop, the held breath before the chorus — these are not decorative. They are structural. They are how music tells time and builds meaning. Strip them out and you may have the correct pitches and the correct rhythms, but something essential has been replaced with a kind of pressurised flatness.

Loudness-processed records also fatigue the ear faster. A session with a heavily limited album at moderate volume will leave a listener more tired than the same duration with a dynamic recording. This is measurable in listening studies, and it is the reason that prolonged exposure to Loudness Wars-era mastering feels somehow exhausting even when it is not especially loud in absolute terms.

Streaming Changes the Equation

The fix, when it began to arrive, came from an unexpected direction. Streaming platforms — Spotify, Apple Music, Tidal, YouTube — all faced the same practical problem: a playlist that moves from a heavily compressed pop track to a quietly mastered jazz recording will jump violently in perceived volume. Users complained. The experience was disjointed.

The solution the platforms landed on was loudness normalisation. Rather than playing every track at its encoded level, the platform measures the integrated loudness of each track (using a standard called LUFS — Loudness Units relative to Full Scale) and adjusts the playback gain to bring it to a target level. Spotify targets around -14 LUFS. Apple Music uses -16 LUFS. Tidal and YouTube have their own targets, clustered in a similar range.

The consequence of this, which took the industry some time to fully absorb, is that submitting a heavily compressed, brick-wall limited master no longer makes your record louder on streaming. It makes it the same volume as everything else — but with all its dynamic range already destroyed. A record mastered with genuine dynamics, sitting perhaps at -18 LUFS, will be turned up by the platform to match the target. A record mastered to -8 LUFS will be turned down. Both arrive at the listener’s ears at roughly the same level. The arms race has been rendered pointless.

The practical shift in mastering practice has been real but gradual. There is now a generation of engineers who actively master for streaming targets, aiming for integrated loudness values that leave room for genuine peaks and genuine quiet. The change is audible in recent releases across genres. Billie Eilish’s recordings — particularly the first album, produced with her brother Finneas in a bedroom — are frequently cited as examples of streaming-era dynamics: intimate, spacious, sometimes startlingly quiet, trusting the listener to follow the music into its quieter passages rather than forcing everything forward.

The Damage That Remains

Not everything is recovered. The catalogue of recordings made between roughly 1995 and 2012 exists in its crushed state, and remastering is an uneven business. Some albums have been sympathetically re-released with restored dynamics — several classic rock catalogues were quietly revisited in the early 2010s, and the difference in the better remasters is striking. Others have simply been re-released with new cover art and the same limiting decisions intact, or in some cases made worse.

There is also the question of what was never recorded with dynamics in the first place — the producers and songwriters who learned their craft in the height of the Loudness Wars, for whom a heavily compressed mix is simply what a record is supposed to sound like. That aesthetic does not disappear just because the commercial incentive to maximise loudness has been removed. It is embedded in how some people hear, and in what a significant portion of listeners have been trained to expect.

And yet. There is something clarifying about what loudness normalisation has done. It has removed the gun from the mastering engineer’s head. The decision about how dynamic a record should be is, once again, a creative one rather than a competitive one. That is not nothing. It is, in fact, the restoration of something fundamental — the idea that how loud the quiet parts are is a musical choice, not a survival strategy.

The silence at the start of Brothers in Arms was always there in the data. It just spent twenty years being considered a problem to be solved.

Explore loudness, dynamic range, and mastering targets for your own releases using the Loudness Guide on Resonillator.