Play a new single on Spotify today and, technically, it no longer matters how hard the mastering engineer slammed the limiter. Spotify measures every track’s loudness and turns it up or down to land at roughly the same level before it reaches your ears — a process called loudness normalization. Apple Music, YouTube, Tidal and most other major platforms do some version of the same thing. In theory, that should have ended the loudness war: the decades-long practice of mastering records as hot as possible to sound bigger than the next song on the radio, the playlist, or the CD rack. It hasn’t. Chart-topping releases are still routinely mastered several decibels louder than the platforms playing them actually call for, and the reasons why have less to do with the technology than with habit, fear, and a music industry that still hasn’t agreed on what "loud" is even for anymore.
What loudness normalization actually does
Normalization doesn’t touch the file an artist uploads. Spotify’s own documentation for artists states that it measures a track’s integrated loudness using the ITU-1770 standard and applies a gain adjustment at playback time, turning louder tracks down and, in some cases, turning quieter ones up, so that everything a listener hears lands close to a target level. Spotify’s default target is −14 LUFS Integrated. Subscribers can override it in three preset tiers — Loud (−11 LUFS), Normal (−14 LUFS), and Quiet (−19 LUFS) — but the default is what almost everyone hears.
Other platforms cluster around similar numbers without matching exactly. Apple Music’s Sound Check feature is widely measured by mastering engineers at roughly −16 LUFS — a figure worth stating carefully, since it comes from independent measurement of Apple’s actual playback behavior rather than a number Apple itself publishes. YouTube normalizes to around −14 LUFS as well, with one meaningful difference: a listener can turn Spotify’s normalization off in settings, but YouTube’s cannot be disabled. Tidal, Amazon Music and SoundCloud normalize too, each landing in the same rough neighborhood with its own quirks in target and enforcement.
The practical consequence is what should have killed the loudness war outright: mastering a track hotter than a platform’s target doesn’t make it play back any louder. It just gets turned down further on the way to the listener’s ears. All the extra limiting an engineer piles on to win that fight buys nothing in perceived volume — it only adds the side effects of aggressive limiting: pumping, flattened transients, and audible distortion on the loudest passages. Chasing loudness on a normalized platform is, by design, a losing trade.
Where "the loudness war" comes from
The term predates streaming by decades — engineers were already competing for jukebox and radio attention with hotter vinyl cuts well before digital audio existed — but the moment that dragged the fight into mainstream conversation was Metallica’s 2008 album Death Magnetic. The CD release was audibly distorted in places from aggressive limiting, and when the same tracks turned up in a different mix for a video game release around the same time, with more headroom left intact, listeners and critics could compare the two directly. The gap was obvious enough that it became a story well outside the usual audio-engineering forums, and it’s the reference point most people still mean when they invoke "the loudness war" today.
Streaming should have ended this years ago — so why hasn’t it?
Spotify’s normalization isn’t new. A louder, ReplayGain-based version of it existed by the mid-2010s, and the platform moved to today’s ITU-1770-based measurement and −14 LUFS target around 2020 and 2021. That gives the industry the better part of a decade to have fully absorbed the fact that mastering loud no longer buys anything on the platforms where most people actually listen.
Ian Shepherd has been watching that adjustment fail to happen for even longer than that. Shepherd is a UK-based mastering engineer, the founder of the annual awareness campaign Dynamic Range Day, and host of The Mastering Show — and he has, on the record, been wrong twice about this exact question. He wrote a piece in 2009 titled "How Spotify Will End the Loudness Wars." In 2015 he wrote another called "YouTube Just Put the Final Nail in the Loudness War’s Coffin." He has since acknowledged, in his own words, that he was "being more than a little over-optimistic" both times, and that normalization "hasn’t" solved the problem — if anything, in his assessment, mastering practice around loudness has become "more polarised," not more converged.
The data backs up his skepticism more than his earlier optimism. In April 2023, Tom Frampton, founder of the mastering-analytics company Mastering The Mix, published a technical breakdown of Spotify’s 25 most-streamed tracks of 2022. The average integrated loudness across those songs came in at −8.4 LUFS — nearly 6 dB hotter than Spotify’s own −14 LUFS target — with an average short-term loudness of −6.1 LUFS and an average dynamic range of just 6.1 DR, a figure low enough that most mastering meters flag it as a dynamics warning. Those are 2022’s biggest songs, analyzed in 2023, mastered years after normalization was already standard on the platform they were topping.
Why the habit outlived its own justification
Shepherd’s own explanation, laid out on the Dynamic Range Day site, isn’t a technical argument at all — it’s a behavioral one. He describes the loudness decision moving through a chain of people on the way to a finished record: artist, producer, mixer, mastering engineer, label, distributor. If any single link in that chain believes, rightly or not, that louder still wins, that belief can push the whole chain toward a hot master regardless of what the streaming platform will actually do to it on playback. Shepherd frames it explicitly as fear, uncertainty and doubt propagating through a production process rather than any real competitive payoff at the end of it.
There are also mechanical reasons the habit hasn’t fully died. Not everything a listener hears goes through a normalizing platform. Bandcamp, notably, does not normalize audio at all. Short-form clips on Instagram and TikTok, where a huge amount of music discovery now happens, don’t uniformly enforce loudness targets either. A physical CD certainly doesn’t. In every one of those contexts, a hotter master genuinely does sound louder than a quieter one sitting next to it — so the incentive to master hot hasn’t vanished, it’s just moved to the parts of a song’s life that streaming’s loudness rules don’t reach. On top of that, in some genres, a dense, high-loudness sound has become part of the sonic identity listeners expect, independent of any competitive advantage it might once have offered.
What the newest chart data actually shows
The most recent version of Frampton’s analysis, published in February 2026 and covering Spotify’s 15 most-streamed tracks of 2025, doesn’t describe a clean resolution in either direction. It describes a split. Bruno Mars and Lady Gaga’s "Die With a Smile" measured a dynamic range of 7.1 DR with a loudness range of 10.4 LU and a short-term loudness of −7.0 LUFS — genuinely dynamic mastering by contemporary chart standards. Alex Warren’s "Ordinary," by the same measurement, came in at 3.3 DR with a short-term loudness of −3.3 LUFS — as hot and dense as anything from the loudness war’s most aggressive years. Both were among 2025’s biggest songs on the same platform, measured with the same tool, a few tracks apart on the same chart.
Frampton’s own framing of the result is the most honest summary available: "There is no single mastering ‘standard’ at the top of streaming in 2025. Successful records range from highly dynamic to heavily controlled." That’s a meaningfully different claim than either "the loudness war is over" or "the loudness war is back." Sixteen years after Death Magnetic and roughly five years after Spotify’s current normalization regime took hold, the platforms have made loud mastering pointless as a competitive strategy — and the industry, at the very top of the charts, still hasn’t converged on what to do about it.