Myth check: "Streaming loudness normalisation squashes dynamic music"

The claim: streaming loudness normalisation squashes dynamic music, so a quiet, open master gets crushed on Spotify.

Verdict: misleading. On the default setting, normalisation is a volume knob, not a compressor. There is one documented case where a limiter does touch dynamic tracks, and it is a listener setting, not the default.

Where it spread

We could not trace a single origin. It tends to surface where producers worry that streaming "processes" their master, often mixed up with the separate, real problem of loud masters distorting when encoded. Treat the spread as our observation, not a sourced fact.

What the evidence says

Spotify's own artist documentation is the primary source. It says loudness normalisation is applied during playback: loudness is measured at upload, but Spotify does "not process or change your audio tracks before playback".

• Loud masters get negative gain so they play at -14 dB LUFS. Spotify says this "lowers the volume in comparison to the master - no additional distortion occurs".
• Quiet masters get positive gain only as far as their peak headroom allows, leaving 1 dB for lossy encoding. Spotify's example: a track at -20 dB LUFS with a -5 dBFS true peak is lifted only to -16 dB LUFS. That is a dynamic track left quieter rather than limited.
• Albums are normalised as a whole, so "the softer tracks are as soft as you intend them to be". Individual track gain applies on shuffle or in playlists.
• The exception: Premium listeners can pick Loud (-11 dB LUFS) or Quiet (-19 dB LUFS). On Loud, Spotify sets the level "regardless of maximum True Peak" and applies "a limiter to prevent distortion and clipping in soft dynamic tracks", engaging at -1 dB with a 5 ms attack and 100 ms decay.

Normalisation also changes the economics of loud mastering. Wikipedia's loudness war article notes that most major streaming services began normalising by default in the late 2010s and early 2020s, and that mastering engineer Ian Shepherd argued normalisation would eventually make the war irrelevant.

For the mechanics in more depth, see our loudness normalisation explainer and what LUFS is.

What we think

This myth gets the direction backwards. Normalisation is, if anything, the best thing to happen to dynamic music in twenty years: a squashed -8 LUFS master is turned down to sit next to your -14 LUFS one, so it loses its loudness advantage and keeps only the cost. Our inference is that dynamic masters are the winners of default normalisation, not the victims.

Where the worry has a grain of truth is the Loud setting. A very quiet, peaky master can hit Spotify's limiter there, which is a reason not to master far below the target without thinking about peaks. It is not a reason to pre-crush your track: you would be applying, to every listener, the limiting that only Loud-setting listeners might hear.

Also note what this check covers: Spotify is the only platform here with a detailed public document. Other services may behave differently, and we have not verified their processing.