LUFS (loudness units relative to full scale) is the unit used to express the perceived loudness of an audio program on a scale referenced to digital full scale, where one loudness unit equals one decibel of level change. It is defined by the measurement algorithm in ITU-R BS.1770 and is identical in meaning to LKFS, the label used in the ITU and ATSC documents. In audio mastering, LUFS has replaced peak level and simple RMS as the number that decides whether a track, episode or film mix sounds as loud as the material around it.
Because streaming platforms and broadcasters normalize playback to a loudness target, the LUFS value of a master determines how much gain the playback system will add or remove. Understanding the measurement therefore matters more than chasing a high number: a master that is louder than the target is simply turned down, often with less punch than a more dynamic version.
How a LUFS meter measures loudness
A LUFS meter estimates loudness by filtering, squaring and averaging the signal in a way that approximates human hearing. The process in ITU-R BS.1770 has four stages, and each one exists to correct a specific weakness of older peak and RMS meters.
- K-weighting filter: a high-shelf boost of about 4 dB above roughly 1.5 kHz models the acoustic effect of the head, and a high-pass filter near 100 Hz reduces the influence of deep bass, which the ear perceives as less loud than its energy suggests.
- Mean-square power per channel: the filtered signal is squared and averaged over a measurement window, giving an energy value rather than a peak value.
- Channel weighting and summing: left, right and center channels are weighted at 1.0, while surround channels receive about +1.5 dB of weight; the LFE channel is excluded from the sum.
- Gating: an absolute gate at -70 LUFS removes near-silence, and a relative gate 10 LU below the ungated level removes quiet passages so that pauses do not drag down the integrated reading.
The result is converted to a logarithmic scale with an offset of -0.691 dB, which makes a 1 kHz sine wave at 0 dBFS on one channel read about -3 LUFS in a stereo meter. The offset is a calibration detail, but it explains why sine-wave tests do not read exactly the same in dBFS and LUFS.
Modern meters report several time scales from the same engine. The table below shows the standard measurement windows and what each one is used for in mastering and delivery.
| Measurement | Time window | Typical use |
|---|---|---|
| Momentary loudness | 400 ms sliding window | Watching transients and short peaks of loudness |
| Short-term loudness | 3 s sliding window | Balancing sections, choruses and scene changes |
| Integrated loudness | Whole program, gated | Compliance with delivery and streaming targets |
| Loudness range (LRA) | Distribution of short-term values | Judging macro-dynamics between quiet and loud parts |
| True peak (dBTP) | Oversampled sample peaks | Preventing clipping after conversion and encoding |
LUFS compared with dBFS, RMS and LU
LUFS answers how loud a program sounds, while dBFS answers how close a sample is to clipping. The two scales share the same full-scale reference, but they describe different properties, and confusing them is a frequent source of delivery errors.
LU is the relative counterpart of LUFS: a difference of 1 LU equals 1 dB, and broadcast meters often show loudness relative to the target, so a reading of +2 LU means the program is two units above the specification. The comparison below places LUFS among the related level measures.
| Unit | What it measures | Frequency weighting | Main limitation |
|---|---|---|---|
| dBFS sample peak | Highest sample value | None | Says nothing about perceived loudness |
| dBTP true peak | Reconstructed peak between samples | None | Still a peak, not a loudness value |
| RMS level | Average signal power | None or simple filter | Overrates bass, no gating of silence |
| LUFS | Absolute perceived loudness | K-weighting | Single number hides tonal and dynamic detail |
| LU | Loudness relative to a reference | K-weighting | Meaningless without the stated reference |
Two masters can share the same integrated LUFS and still sound different in density: one may reach its value with a steady, compressed level, the other with quiet verses and loud choruses. That is why engineers read integrated loudness together with LRA and short-term values, and why the neighboring concepts of true peak and dynamic range compression belong in the same conversation.
Loudness targets for streaming and broadcast
A loudness target is the integrated level to which a platform or broadcaster normalizes playback. EBU R128 sets -23 LUFS for European television, the US ATSC A/85 practice uses -24 LKFS, and most music streaming services normalize to roughly -14 to -16 LUFS. These values change occasionally, so delivery specifications should always be confirmed for the current project.
The following table summarizes typical reference values at the time of writing; exact figures depend on the platform, the playback mode and whether album or track normalization is active.
| Context | Integrated target | True peak ceiling | Tolerance or note |
|---|---|---|---|
| European broadcast (EBU R128) | -23 LUFS | -1 dBTP | About ±0.5 LU for most programs |
| US broadcast (ATSC A/85) | -24 LKFS | -2 dBTP | About ±2 LU |
| Music streaming services | About -14 LUFS | -1 dBTP | Louder masters are turned down |
| Some streaming in loud mode | About -11 LUFS | -1 dBTP | Optional user setting with limiting |
| Podcasts and spoken word | About -16 LUFS stereo | -1 dBTP | Often -19 LUFS for mono files |
Normalization changes the logic of mastering. When a platform lowers every loud master to the same target, extra limiting no longer buys loudness, it only reduces transient impact. Many mastering engineers therefore choose a level that serves the music, typically between -14 and -9 LUFS integrated for contemporary genres, and accept that the platform will adjust playback gain.
Using LUFS in a mastering session
In practice, LUFS is a verification tool, not a mixing target. Engineers set tone and dynamics by ear, then read the meter to confirm that the integrated level, loudness range and true peak fit the destination.
A practical routine is to measure the whole song or episode in one pass, because integrated loudness depends on the full program. Checking only the loudest chorus gives a short-term value that can be 3 to 5 LU higher than the final integrated reading. The decision table below maps common delivery situations to a reasonable strategy.
| Situation | Recommended approach | Reason |
|---|---|---|
| Album for streaming | Master by ear, keep true peak at -1 dBTP | Album normalization preserves song-to-song balance |
| Single for club and streaming | Accept -9 to -7 LUFS if the genre demands it | Density is part of the style, not only loudness |
| Broadcast program | Hit the target within tolerance | Non-compliant files are rejected or corrected |
| Podcast episode | Aim near -16 LUFS with controlled LRA | Speech must stay intelligible in noisy places |
| Classical recording | Prioritize wide LRA over a high level | Natural dynamics carry the musical meaning |
Common misconceptions about LUFS
Most errors with LUFS come from treating one number as a complete description of a master. The following points recur in mastering and delivery work.
- Matching the streaming target is not mandatory. Platforms adjust gain automatically; a master at -10 LUFS is not penalized except by being turned down.
- Lower LUFS does not mean more dynamic range. Integrated loudness is an average; dynamics are better described by LRA, crest factor and short-term movement.
- Normalization does not prevent clipping. A file turned up by a platform can exceed full scale, which is why the true peak ceiling still matters.
- Bass-heavy mixes can read quieter than they feel in a club, because K-weighting deliberately reduces low-frequency influence.
- Meters must use the same standard version. Older and newer BS.1770 revisions differ in gating and channel handling, so readings across tools can differ slightly.
The simple explanation
LUFS is a way of measuring how loud something sounds to people. It does not care only about the loudest moment. It listens a bit like an ear does. Deep rumble counts for less, and the middle and upper tones count for more.
Think of a radio station that wants every song and advert to feel equally loud. LUFS gives each piece one loudness score. The station or app then turns each piece up or down to the same score. So making a song louder than the target does not help. The player will simply turn it back down.
How it is used in practice
A mastering engineer prepares a ten-track indie album for streaming and for a regional radio broadcaster. After shaping tone and compression by ear, the engineer measures each finished track from start to end. The loudest single reads -9.5 LUFS integrated with a true peak of -1.2 dBTP, while a quiet acoustic ballad reads -13 LUFS, which preserves the intended contrast when album normalization is active.
The broadcaster requires -23 LUFS and -1 dBTP, so the engineer exports separate broadcast versions by lowering gain rather than adding processing, then confirms each file with an integrated measurement. The result is one set of masters that sound dense and consistent on streaming services and a compliant set that passes the broadcaster’s automatic loudness check on the first submission.