1
LUFS Targets for Streaming
Streaming platforms normalize playback loudness: they measure each track's integrated LUFS and turn it down (and on some services, up) toward a reference level, about -14 LUFS for Spotify and YouTube, -16 for Apple Music. A master louder than that simply gets turned down, so the number is a ceiling, not a target.
The consequence is decisive: over-limiting for loudness gains nothing on streaming. A dynamic master at -14 LUFS plays just as loud as a crushed one, and sounds better.
In practice: After Spotify normalization, an album mastered at -14 LUFS plays at the same volume as a pop record crushed to -8, but keeps its dynamics, which the listener hears as quality.
Streaming normalizes to ~-14 LUFS (Apple ~-16): a ceiling, not a target; do not over-crush.
2
Loudness Metering: Integrated, Short-Term, Momentary
Three LUFS measurements work together. Integrated LUFS averages the entire program (the metric streaming platforms normalize by). Short-term LUFS averages a 3-second sliding window, best for chorus-vs-verse contrast checks. Momentary LUFS uses a 400 ms window, best for catching individual loud peaks and dynamics.
The LRA (Loudness Range) metric measures program dynamic range, high LRA means dynamic music, low LRA means crushed.
In practice: An album where the bridge feels weirdly loud or buried compared to verses and choruses: check short-term LUFS during each section, if the bridge is 2-3 dB above the rest, bring it up with automation, not compression.
Loudness metering: integrated for target, short-term for section balance, momentary for peaks, LRA for dynamics.
3
True Peak & Inter-Sample Peaks
Inter-sample peaks (ISPs) are peaks that exceed 0 dBFS between sample points after digital-to-analog conversion or lossy encoding (AAC, MP3, Ogg). Standard sample-peak meters miss them; a True Peak meter (dBTP) reveals them.
Streaming platforms recommend a -1 dBTP ceiling because lossy codecs can add 1+ dB of inter-sample peaks. Use a true-peak limiter (FabFilter Pro-L 2, Sonnox Oxford Limiter) with -1 dBTP ceiling, the audible difference from -0.1 to -1.0 is negligible, but the elimination of codec distortion is immediate.
In practice: A master hitting -0.1 dBTP that sounds distorted on Spotify AAC encoding but clean on the WAV: ISPs from lossy compression, re-master with the limiter at a -1 dBTP true-peak ceiling.
True peak ≠ sample peak, use a -1 dBTP ceiling for streaming masters to avoid lossy-codec distortion.
4
Limiting: How Much Is Too Much
A limiter sets a hard ceiling the signal cannot cross (typically -1 dBTP, to leave room for inter-sample peaks). Pushing more signal into it raises perceived loudness, but every dB of gain reduction also flattens transients and shrinks dynamic range.
So limiting is a trade. For dynamic, acoustic material keep it gentle, a few dB at most, or the picks, chops, and bow attacks that give the music life get crushed.
In practice: A client wanting "pop loud": more than 2-3 dB of limiting flattens the transients, and streaming normalization discards the extra loudness anyway.
A limiter caps loudness, but heavy limiting flattens transients.
5
Dynamic Range for Acoustic Music
Dynamic range, the gap between a record's quiet and loud passages, is a deliberate choice. Pop masters squeeze it tight (4-8 LU LRA) for relentless energy; acoustic and folk music wants it wide (roughly 8-14 LU LRA) so the contrast between an intimate verse and a full chorus survives.
That contrast is the emotional content. Crush it and the music loses its breath, and since streaming normalizes loudness anyway, you gain nothing for the sacrifice.
In practice: A folk album with a quiet verse and energetic chorus: preserving 10-plus LU of loudness range keeps the lift; flattening it to a pop-style 6 LU kills the arc.
Acoustic music wants wide dynamic range, the contrast is the emotion.
6
Loudness Normalization Across Platforms
Streaming platforms normalize playback to a reference level: Spotify about -14 LUFS (Normal mode; Loud is -11, Quiet is -19), Apple Music about -16, YouTube and Tidal about -14. Services set and change these themselves, so treat the figures as approximate. Masters louder than the reference get turned DOWN; quieter masters are left alone or turned UP only as far as their peaks allow, depending on the service (Spotify Loud applies positive gain plus a limiter).
A crushed -8 LUFS pop master loses 3-6 dB on streaming with no loudness advantage and reveals all its over-limiting artifacts. A dynamic -14 LUFS master arrives with full dynamics intact.
In practice: A client insisting on a -8 LUFS master to "compete": demonstrate via Spotify Loud mode, their crushed master loses 3 dB; a -13 LUFS master keeps its dynamics intact and arrives at the same perceived loudness.
Streaming normalizes to about -14 to -16 LUFS: a ceiling, not a target. Loud masters get turned down; dynamics win.
7
Headroom & Gain Staging for Mastering
A mix handed off for mastering should leave room to work: peaks somewhere around -3 to -6 dBFS (a common convention, not a standard), and no limiter or maximizer on the mix bus, loudness is the mastering engineer's job, not the mixer's.
The reason this matters so much: if a mix is delivered clipped, peaks slammed into 0 dBFS with flat-topped waveforms, that distortion is permanent. Turning it down later just gives you quieter distortion.
In practice: A "normalized" mix with flat-topped transient peaks cannot be rescued in mastering, ask for a fresh export with the mix-bus limiter removed and peaks around -3 to -6 dBFS.
Deliver mixes with headroom and no bus limiter, clipping is permanent.
8
Mastering Chain Order
Mastering is the final polishing stage applied to a finished stereo mix, and its tools run in a deliberate order: EQ first (broad tonal shaping), then compression (gentle glue), then limiting (the loudness ceiling). Order matters because each stage feeds the next, EQ before compression means the compressor reacts to the corrected tone, and compression before limiting means the limiter only catches the final stray peaks.
Get the order wrong and the chain works against itself.
In practice: Muddy low-mids: cut them with EQ first in the chain, so the compressor downstream no longer pumps on that muddy energy.
Mastering chain order: EQ → compression → limiting.
9
Using Reference Tracks in Mastering
A reference track is a well-mastered commercial recording you compare your master against, it keeps your ears honest. The non-negotiable first step is level-matching: the louder of two tracks always seems "better", so even a 1 dB difference will bias your judgment.
Choose references in the same genre, then compare three things deliberately: tonal balance, dynamics, and stereo width.
In practice: Comparing a folk master to a brighter pop reference: level-match first, and trust a same-genre reference over the pop one, pop's aggressive top end is the wrong target.
Reference against same-genre tracks, level-match before you judge.
These lessons are the teach side of knowledge cards from the Studio Mix deck. In the daily plan each one is followed by a quiz, spaced over weeks, so it stays known rather than read once.