Loudness Normalisation (Streaming Platforms)
How Spotify, Apple Music, YouTube and others normalise loudness — and how it affects mastering decisions.
For most of recording history, "louder is better" was a workable strategy. Streaming services killed that. They normalise loudness automatically — pushing quieter tracks up and pulling louder tracks down so that everyone hears music at roughly the same level.
A track mastered at −6 LUFS isn't "louder" than one at −14 LUFS on Spotify. It's just compressed harder. The dynamic-range cost is the same; the perceived loudness when played back is roughly identical.
What Each Platform Does
| Platform | Target | Notes | |----------|--------|-------| | Spotify | −14 LUFS integrated | Loudness normalisation on by default. Quieter tracks get +6 dB max gain; louder tracks get reduced. | | Apple Music / iTunes | −16 LUFS integrated | "Sound Check" feature. Same principle, slightly more conservative target. | | YouTube | −14 LUFS integrated | Per-video normalisation. Music videos and uploads. | | Tidal | −14 LUFS integrated | Standard normalisation. | | Amazon Music | −14 LUFS integrated | Same target. | | Pandora | −14 LUFS integrated | Same target. | | SoundCloud | No normalisation | Platform plays files as-mastered. Louder mastering still wins here. |
Calculate target loudness per platform →What "Integrated LUFS" Means
LUFS = Loudness Units relative to Full Scale. Integrated LUFS is the loudness measurement averaged across the entire programme.
A song mastered to peak at −1 dBFS but with average loudness of −14 LUFS is "loud enough" for streaming. A song hammered to −6 LUFS will be turned down 8 dB on Spotify.
True Peak
Most streaming services also enforce a true-peak ceiling — typically −1 dBTP (decibels true-peak). True-peak measures what the signal actually does at playback, accounting for inter-sample peaks that exceed sample-level peaks.
A signal sample-peaking at exactly 0 dBFS often has true peaks at +1 to +2 dBTP. After lossy encoding (AAC for streaming), those peaks become audible distortion.
The mastering engineer's standard is: peak no higher than −1 dBTP after limiting. This gives lossy encoders room and avoids audible artefacts on streaming playback.
Should You Master Loud Anyway?
The case against:
- Loudness normalisation cancels the gain advantage. A −6 LUFS track plays at the same volume as a −14 LUFS track on Spotify.
- Smashing dynamics costs musicality. Limited transients sound smaller, less punchy, less detailed.
- Listening fatigue. Listeners A/B test less consciously than engineers think. A dynamic track stands out from over-compressed noise around it.
The case for:
- Some platforms don't normalise (SoundCloud, embedded uploads, podcast playback).
- Quiet listeners might want more apparent volume without changing their device level.
- Genre conventions. EDM, modern hip-hop, and aggressive rock are expected to be loud-mastered. Going against that expectation can sound "wrong" to genre listeners.
Recommended Mastering Targets
For a "neutral" master that travels well to streaming:
- Integrated LUFS: −14 to −10 LUFS. (Closer to −14 for jazz, classical, acoustic; closer to −10 for pop, rock, hip-hop.)
- True peak: −1 dBTP after limiting.
- Short-term LUFS variability: ~3 LU LRA (loudness range). Modern pop typically lands at 4–7 LU LRA.
Ozone 11, Pro-L 2, and the Waves WLM Plus meter all show these values during mastering.
Common Mistakes
Mastering for SoundCloud, releasing on Spotify. SoundCloud doesn't normalise; Spotify does. A track mastered at −6 LUFS sounds "right" on SoundCloud and gets crushed back down on Spotify.
Skipping the true-peak check. A master peaking at −0.1 dBFS might true-peak at +1 dBTP. After streaming-service AAC encoding, those peaks become audible distortion.
Believing "the loudness war is over." It isn't. Major releases still arrive at −8 to −7 LUFS routinely. Streaming normalisation reduces the advantage of being louder, not the prevalence.
Ignoring platform context. A YouTube music video has different normalisation behaviour to a Spotify single. A streaming radio station may apply additional processing on top of the platform's normalisation.
Workflow
- Mix at sensible levels — peaks at −6 dBFS, plenty of headroom.
- Master targeting −14 to −10 LUFS integrated.
- Check on multiple devices — phone speaker, car, monitors.
- Verify true-peak — use a true-peak limiter (Pro-L 2 with TP enabled).
- Reference against tracks in the same genre at the same loudness target.