LUFS explained: the only loudness guide producers need
Integrated, short-term, momentary, LRA, true peak: what each number actually measures, what streaming platforms do with them, and the targets that matter in 2026.
- LUFS measures perceived loudness with frequency weighting and gating — not signal level.
- Integrated = the whole track, short-term = 3s window, momentary = 400ms. LRA = dynamics range.
- Streaming platforms turn loud masters down to ~-14 LUFS. They never turn quiet ones meaningfully up.
LUFS is the most misunderstood number in modern music production, and the confusion costs real money: crushed masters, weak drops, endless revision loops. Let me give you the working understanding that actually matters, without the broadcast-standard homework.
What LUFS actually measures
Peak meters read electrical maxima. RMS reads average power. Neither reads what your ear experiences. LUFS (Loudness Units relative to Full Scale) applies a frequency weighting that mirrors human hearing — we're more sensitive to upper mids than to sub — measures energy in overlapping 400 ms blocks, and gates out silence so intros and breakdowns don't drag the number down. One LU equals one dB of change, but the measurement itself is perceptual. That's why every major platform standardized on it.
The four numbers on your meter
Integrated is the loudness of the entire track, gated — the number platforms use to normalize you. Short-term (3-second window) shows how loud each section feels; the short-term of your drop matters more to the listener than the integrated of the whole track. Momentary (400 ms) catches instant peaks of density. LRA (Loudness Range) describes the spread between your quiet and loud sections: electronic music typically lives at 4–8 LU, a dynamic film score might be 15+.
Platforms normalize on integrated. Listeners feel short-term. Great masters manage both.
What platforms actually do
Spotify, YouTube, Amazon and Tidal normalize around -14 LUFS; Apple Music around -16; Deezer -15. Crucial detail: they turn louder tracks down to the target, but (with minor exceptions) do not boost quiet tracks up past their peak ceiling. So a -8 LUFS master plays at -14 on Spotify — same volume as a -14 master, but with all the density and, if done badly, all the distortion of the loudness war baked in. This is why "should I master to -14?" is the wrong question, and I wrote a whole article on the right one.
True peak: the other number that bites
LUFS says how loud; true peak (dBTP) says whether your waveform overshoots full scale between samples. Keep it at or below -1 dBTP and lossy encoders (AAC, Ogg) stay clean; ride it at 0 and you're donating distortion to every stream. The two numbers together — integrated LUFS and dBTP — are 90% of your delivery QC.
Measure it yourself, free
I built a browser-based LUFS meter that runs the full ITU-R BS.1770-4 measurement — gated integrated, short-term timeline, oversampled true peak, LRA — plus key detection and BPM, entirely on your device. Drop your master in before you send it anywhere. Sixty seconds of checking beats a re-delivery every time.
Numbers off target? That's literally my job. Mastering from EUR 60, streaming-perfect on every platform.
Fix my loudness