--- title: "K-Weighting in LUFS: The BS.1770 Filter Explained | Gawergy Audio" description: "Understand why BS.1770 filters audio before loudness calculation, how the two-stage K-weighting concept works, and what it does not model about human hearing." canonical_url: "https://gawergy.com/learn/k-weighting-lufs" md_url: "https://gawergy.com/learn/k-weighting-lufs.md" last_updated: "2026-09-23" date_published: "2026-09-23" --- # What Is K-Weighting in LUFS Measurement? K-weighting is the frequency-shaping part of the ITU-R BS.1770 programme-loudness algorithm. It filters audio before the meter combines energy across time and channels. Its purpose is to make a useful standardized loudness measure rather than treating all sample amplitudes as equally informative about listening level. It is a defined engineering approximation, not a complete model of hearing. ## Key takeaways - BS.1770 K-weighting combines a first-stage shelving response with a second-stage high-pass response. - The filter changes how different frequency regions contribute to the loudness calculation. - Gating, time windows, and channel weighting are additional parts of a LUFS measurement, not synonyms for K-weighting. - Equal LUFS does not guarantee equal subjective loudness for every spectrum or playback condition. ## Why a raw amplitude average is insufficient An unweighted RMS calculation averages squared sample amplitudes over a chosen window. It is mathematically well defined, but it treats the waveform according to that operation rather than according to a standardized account of perceptual frequency sensitivity. Two signals can have similar raw effective amplitudes while their spectral distributions make them feel different in level. Broadcast and programme-loudness work therefore needed a reproducible method that follows perceived level more closely than a peak meter or arbitrary unweighted average. ITU-R BS.1770 specifies such a method, with K-weighting as its frequency-shaping stage. The claim is modest: frequency weighting improves the relevance of an energy calculation for programme loudness under the standard's intended conditions. It does not imply that the filter knows the music's genre, listener's hearing, playback level, room, masking, or emotional impact. Even before other limitations, a single static weighting curve cannot reproduce all context-dependent aspects of perception. The [RMS versus LUFS guide](/learn/rms-vs-lufs) compares the mathematical effective-amplitude measure with the standardized loudness result. This article isolates the filter that helps create that difference. A peak reading has an even narrower purpose. It identifies a maximum, not weighted energy over time. A single high sample can set the peak while contributing little to a programme loudness value. K-weighting does not 'fix' the peak reading; it belongs to a separate loudness path. The [peak versus RMS article](/learn/peak-vs-rms) explains the maximum-versus-average distinction, while the [LUFS/dBFS/dBTP guide](/learn/lufs-vs-dbfs-vs-dbtp) keeps the units separate. ## The two-stage filter specified by BS.1770 The current ITU-R BS.1770-5 text describes a two-stage pre-filter. Its first stage is a shelving response associated with modelling acoustic effects of the head. Its second stage is a high-pass response, also described in the standard as RLB weighting. The concatenation of those stages is designated K-weighting. This is a precise technical definition: the meter applies the prescribed response before it accumulates mean-square energy. A producer does not need to memorize the coefficients to understand that different frequencies do not contribute exactly as they would to an unweighted RMS average. The standard publishes filter coefficients for a reference sampling rate and instructs implementations at other rates to provide the same intended frequency response. That is important because the algorithm is a specification, not an arbitrary EQ curve that each meter designer is free to improvise. A compliant meter should have comparable behavior across supported sample rates within the standard's tolerances. It also means that a sample-rate change by itself should not be assumed to change a song's LUFS for artistic reasons; if the audio content and measurement remain equivalent, the weighting response should remain equivalent in intent. A common shorthand says K-weighting is 'like an ear curve.' That can be useful as a first intuition but is imprecise. The standard's two stages have specific design motivations and measured responses. Human hearing is nonlinear and context-dependent; K-weighting is a fixed filter used within a larger programme-loudness algorithm. Calling it a perfect hearing simulation would overstate the evidence and lead to poor interpretation when two equal-LUFS tracks still feel different. ## Filtering comes before energy and channel combination After the frequency-shaping stages, the algorithm calculates mean-square power over the relevant measurement interval. It then combines channel contributions under specified weighting rules. This is why the word *weighted* has more than one possible role in a LUFS discussion: K-weighting shapes the frequency response of each channel's audio, while channel weights govern the contribution of channels to the programme result. They are related parts of the standard but not the same operation. The exact multichannel rules matter when comparing stereo with surround or object-based content. A simplistic formula that measures a mono signal's raw RMS and adds an offset cannot recreate all of this. The offset would depend on the spectrum and channel configuration, and later gating may also change an integrated measurement. For a steady signal at a known reference frequency, one can derive a particular relationship from the standard's calibration. That special case does not generalize to arbitrary music. The goal of a LUFS meter is consistent treatment under the defined method, not a universal conversion table from every other level metric. BS.1770 also defines true-peak measurement, but that is another branch of the recommendation. K-weighting is not a filter placed on the delivered audio and is not a true-peak limiter. It is part of how the loudness meter analyzes the signal. A file does not become spectrally changed because a LUFS meter reads it. This distinction matters when separating measurement from processing. The [true-peak limiting guide](/learn/true-peak-limiting) addresses the processor that may act on a reconstructed peak; K-weighting itself does not process the audience's playback signal. ## K-weighting is not the whole LUFS algorithm Integrated loudness needs rules for very quiet passages and silence. BS.1770 specifies a gated measurement process with overlapping blocks and thresholds. EBU Mode specifies meter views such as momentary, short-term, and integrated loudness. Those time and gating rules operate in addition to frequency weighting. A statement such as 'LUFS is just K-weighted RMS' catches part of the energy basis but leaves out meaningful standard behavior, particularly over an entire programme. The [Loudness Range article](/learn/loudness-range-lra) discusses a separate EBU descriptor that uses its own statistical treatment of short-term loudness. This separation helps diagnose apparent meter disagreements. Two meters with equivalent K-weighting but different analysis windows or reset positions can report different values for a changing passage. A gated integrated reading can differ from a raw whole-file weighted average if silence or quiet material is present. A true-peak display can move independently because it addresses a maximum. Before attributing a mismatch to a filter error, check the meter mode and interval. EBU Tech 3341 documents the expected EBU Mode views, and BS.1770 specifies the underlying programme-loudness algorithm. The standard's numeric gates and block durations are public measurement definitions, not recommended mastering settings. They tell a meter how to calculate. They do not tell a producer how much to compress, clip, or raise a track. Keeping this boundary clear prevents a technical explanation of LUFS from becoming a prescriptive loudness recipe. ## Why the result still differs from listening The ITU document itself cautions that its algorithm is not generally suitable for estimating subjective loudness of pure tones. That is a useful reminder that a standardized programme meter has a designed domain and limits. Real music adds additional complications: masking between sounds, transients, arrangement changes, playback level, and listeners' environments. Two songs can match in integrated LUFS and still feel different, and one song can feel different in a quiet room versus a noisy commute. The [same-LUFS-different-loudness article](/learn/same-lufs-different-loudness) examines those perceptual gaps without dismissing the utility of the standard. K-weighting also should not be confused with tonal advice. A meter's frequency response does not imply that a mix should be EQ'd to maximize its reading. Raising a band merely to change LUFS can worsen balance and may cause other peaks or harshness. Conversely, a tonal change made for the music may alter the LUFS reading even when a raw peak remains stable. That is expected because the method is intentionally frequency dependent. Use the meter to describe the result; let the mix's musical needs guide processing decisions. Platform playback adds yet another layer. Services can normalize according to current policies and settings, which are not identical across platforms and may evolve. K-weighting helps make a standardized measurement available to them, but it does not set a universal production target. The [streaming loudness targets article](/learn/streaming-loudness-minus-14-lufs) addresses that misconception. The defensible conclusion is that K-weighting is a specified part of a useful measurement method, not a magic curve that makes every equal-LUFS signal subjectively identical. ## How to interpret a K-weighted result When a LUFS reading changes after processing, consider whether the audio's level, spectrum, time structure, or measured interval changed. A compressor may alter the amount of energy across a phrase; a clipper may add harmonics and change peaks; an EQ may change frequency balance; a fade may alter the programme interval. A K-weighted meter responds according to its defined method. Its movement does not by itself reveal which of those mechanisms acted. Compare with peak, spectrum, and listening observations if the cause matters. A careful report might say that the integrated loudness of a particular exported file under BS.1770 analysis changed, while its true peak and short-term distribution were checked separately. It need not claim that every listener will hear the same numerical difference as the meter. Nor should it claim that raw RMS or sample peak can substitute for K-weighted loudness. The value of a standardized unit is comparability under known conditions, not complete perceptual certainty. The core model is simple: a prescribed two-stage filter shapes frequency contribution before the loudness calculation, and the rest of the standard supplies time, gating, and channel rules. K-weighting matters because human listening level is not captured well by a single raw maximum or unweighted average. Its limits matter because even a good standardized measure cannot replace listening to the music. ## About G-Clipper Pro Clipping can change both peak shape and spectrum. A LUFS meter then analyzes the resulting audio with K-weighting; that measurement is separate from G-Clipper Pro's processing curve. ## Sources & References - [BS.1770: Algorithms to measure audio programme loudness and true-peak audio level](https://www.itu.int/rec/R-REC-BS.1770-5-202311-I) - ['EBU Mode' metering to supplement EBU R 128 loudness normalisation](https://tech.ebu.ch/publications/tech3341) - [Loudness — EBU Technology & Innovation](https://tech.ebu.ch/loudness) - [FabFilter Pro-L 2 Help - Loudness metering](https://www.fabfilter.com/help/pro-l/using/loudnessmetering) ## Continue Reading - [RMS vs LUFS: What’s the Difference?](https://gawergy.com/learn/rms-vs-lufs) - [LUFS, dBFS, and dBTP Explained](https://gawergy.com/learn/lufs-vs-dbfs-vs-dbtp) - [Why Two Songs at the Same LUFS Can Sound Different in Loudness](https://gawergy.com/learn/same-lufs-different-loudness) ## Sitemap See the full [Gawergy.com sitemap](https://gawergy.com/sitemap.md).