Why −14 LUFS is not the right target for every song

Spotify’s normal playback setting is −14 LUFS. That fact is useful. Turning it into a universal mastering law confuses playback gain with artistic loudness.

If a finished master measures −9 LUFS, Spotify normally turns it down for playback. It does not reach into the file and undo the limiting, saturation or transient shaping used to get there. The song remains dense; it is simply played quieter.

What Spotify actually does

Spotify’s current artist documentation says normal playback adjusts tracks to −14 LUFS under ITU 1770. Album playback preserves relative track-to-track levels, while shuffle and mixed playlists use track normalization. Premium listeners can choose Loud (−11), Normal (−14) or Quiet (−19); Loud can engage Spotify’s limiter on softer dynamic tracks. The web player and some third-party devices do not use normalization.

Spotify recommends −1 dBTP for a master around −14 LUFS, and below −2 dBTP when a master is louder than −14 because encoding distortion becomes more likely. Crucially, it asks for one high-quality stereo master, not a separate file for each playback mode.

Loudness still changes the sound

Two masters normalized to the same playback level can sound very different. A louder master may have denser sustain and a more aggressive texture. A more open master may keep bigger transients and movement. Neither choice is automatically correct. The arrangement, genre, emotional direction and release context decide whether that trade is worthwhile.

My eleven purchased reference masters span −11.4 to −7.3 LUFS. Their LRA spans 3.5 to 12.3 LU. That range contains sparse pop, hip-hop, acoustic music and modern band production. Averaging them would erase exactly the differences that make them useful references.

What I target instead

PolishMyTrack treats loudness as a paired decision: a musical target and a true-peak ceiling. For the five unfinished sources in our current gate, adaptive Auto selected roughly the low −12s to high −11s, while PCM stayed at or below −2 dBTP. Already-hot commercial masters took a constant downward-gain path—no limiter, no tonal processing—and kept their displayed LRA unchanged.

That approach is deliberately conservative. If the peak ceiling prevents the song reaching its loudness target without changing the envelope, the ceiling wins and the receipt says so. A delivery-safe result is better than pretending every number can be satisfied simultaneously.

Use −14 LUFS as a forecast, not a commandment

I still want to know how a master will move under normal playback. That forecast helps expose whether extra limiting buys anything once gain is normalized. But I do not flatten an acoustic track or undercook a dense electronic track just to make a meter print −14. The right question is: does this processing help after the loudness advantage is removed?