LUFS (Loudness Units Full Scale)
The modern loudness measurement standard — what it is, how it works, and what every platform targets.
LUFS is the measurement standard that finally brought sanity to loudness. If you're releasing music in 2026, it's the single most important number on your meter — more important than peak level, more important than RMS, more important than whatever your limiter's gain reduction meter says.
Here's what you need to know.
Key Information
What LUFS Actually Measures
LUFS measures perceived loudness — not peak level, not average signal level, but how loud something actually sounds to a human being.
It does this through K-weighting, a frequency curve that models the way we hear. Our ears are most sensitive to the mid-range (roughly 1–6 kHz) and less sensitive to low frequencies. K-weighting applies more weight to the frequencies we hear most prominently and less to the ones we don't. The result is a loudness figure that correlates far more closely with subjective experience than any simple level measurement.
When people refer to "the LUFS of a track," they almost always mean integrated LUFS — the average perceived loudness measured across the entire duration. This is the number streaming platforms use for normalisation. Short-term (measured over a 3-second window) and momentary (400ms window) readings are useful for monitoring dynamics in real time, but integrated is the figure that matters for delivery.
LUFS vs dBFS vs RMS
These three measurements tell you different things, and confusing them is one of the most common mistakes in mastering.
dBFS (decibels relative to Full Scale) measures the peak sample value in a digital signal. 0 dBFS is the absolute ceiling — the loudest a digital sample can be before clipping. It tells you nothing about perceived loudness. A snare transient can hit 0 dBFS while the track sounds quiet overall.
RMS (Root Mean Square) measures the average signal level over time. It's closer to perceived loudness than peak measurement, but it applies no psychoacoustic weighting. A bass-heavy track and a mid-heavy track with the same RMS will sound like they have very different loudness levels.
LUFS combines time-averaged measurement with K-weighting to give you a figure that actually reflects what your ears hear. It's the only one of the three that streaming platforms use to normalise playback volume.
Check your track against platform targets →Platform Targets
Every major streaming platform normalises loudness. They each pick a target and turn tracks up or down to match it. If your master doesn't hit the target, the platform adjusts it for you — sometimes gracefully, sometimes not.
| Platform | Target | Normalisation | |---|---|---| | Spotify | −14 LUFS | Up and down | | Apple Music | −16 LUFS | Down only (Sound Check) | | YouTube | −14 LUFS | Down only | | Tidal | −14 LUFS | Down only | | Broadcast (EBU R128) | −23 LUFS | Strict compliance | | Broadcast (ATSC A/85) | −24 LKFS | Strict compliance |
Spotify is the outlier — it normalises both ways, meaning quiet masters get turned up and loud masters get turned down. Apple Music only turns things down (Sound Check must be enabled by the listener). YouTube only turns down.
The practical takeaway: if you master to −14 LUFS integrated, you'll be at or near unity on most platforms. Master to −16 LUFS if you want to preserve more dynamics and are comfortable with Spotify raising the volume slightly.
Calculate your loudness for every platform →True Peak
True peak is the actual peak level of your audio when the digital signal is reconstructed to analogue — and it can exceed the highest sample value in your file.
Here's why. Digital audio is a series of discrete samples. Between those samples, the reconstructed analogue waveform can overshoot — these are called inter-sample peaks. Your DAW meter, which reads sample peaks, won't catch them. But when a lossy codec (MP3, AAC, Ogg Vorbis) re-encodes your audio, those inter-sample peaks can cause audible clipping and distortion.
This is why true peak metering exists and why delivery specs always include a true peak ceiling alongside the LUFS target.
Typical true peak ceilings:
- Streaming (Spotify, Apple Music, YouTube): −1 dBTP
- Broadcast (EBU R128): −1 dBTP
- Broadcast (ATSC A/85): −2 dBTP
Use a true peak limiter on your master bus. Most modern limiters offer a true peak mode — use it. A standard brickwall limiter set to −1 dBFS is not the same as a true peak limiter set to −1 dBTP.
The Loudness War (And Why It's Over)
For two decades, the music industry engaged in a race to make records louder. More limiting, more compression, more gain — because louder sounded better in a direct comparison, and there was no normalisation to level the playing field.
The cost was dynamics. Records from the mid-2000s are noticeably more fatiguing than their counterparts from the 1970s and 80s, not because the music is worse, but because every last decibel of dynamic range was sacrificed at the altar of loudness.
LUFS-based normalisation killed that race. If Spotify turns your −8 LUFS master down to −14 LUFS for playback, you haven't gained any loudness advantage — you've just lost 6 dB of dynamic range for nothing. Your track will sound flatter and more fatiguing than a version mastered to −14 LUFS with its dynamics intact.
The loudness war is over. Dynamics won. Master for the target, not for the meter.
Common Mistakes
Confusing LUFS with dBFS. They measure different things entirely. A track can peak at −1 dBFS and sit at −20 LUFS. The numbers are not interchangeable.
Mastering louder than the platform target. If you deliver at −8 LUFS integrated and Spotify's target is −14 LUFS, your track gets turned down by 6 dB. You've thrown away dynamics for no benefit. The track won't sound louder than one mastered to −14 — it'll sound the same volume but with less dynamic range.
Ignoring true peak. Your master might sit neatly at −14 LUFS with peaks at −0.3 dBFS, but if inter-sample peaks hit +1.5 dBTP, every streaming codec will clip on those peaks. Always check true peak, not just sample peak.
Measuring momentary instead of integrated. Momentary LUFS (400ms window) fluctuates wildly — a loud chorus might read −10 LUFS momentary while the track's integrated loudness is −14. Platform normalisation uses integrated loudness. That's the number you need to hit.
Applying the same target to every genre. A classical recording mastered to −14 LUFS would be absurdly compressed. A hip-hop track at −23 LUFS would feel lifeless on a playlist. Use the platform target as a guideline, not a commandment, and let the music dictate the dynamics.