/ʃɔːrt tɜːm ˈlaʊdnəs/
Short-Term Loudness is a loudness measurement averaging audio energy over a sliding 3-second window, expressed in LUFS. It gives producers a real-time view of section-level dynamics without the long-term averaging of integrated loudness.
Every streaming platform is quietly judging your mix right now — and the 3-second window is where they make their decision.
Short-Term Loudness is a psychoacoustically weighted measurement of audio signal energy calculated over a continuously sliding 3-second integration window, expressed in Loudness Units relative to Full Scale (LUFS), sometimes equivalently written as LKFS. Defined within the ITU-R BS.1770 standard and refined by the EBU R 128 specification, it occupies a crucial middle position in the loudness measurement hierarchy: more responsive than Integrated Loudness (which averages an entire program from start to finish) and far more stable and programme-representative than Momentary Loudness, which uses only a 400-millisecond window. That 3-second integration time was deliberately chosen to correspond with how the human auditory system perceives sustained sound passages — long enough to catch a full verse phrase, a drum break, or a sustained chord bloom, yet short enough to track meaningful dynamic shifts between sections of a track.
The measurement applies a K-weighting filter before integration. K-weighting combines a high-shelf pre-filter that boosts sensitivity to high frequencies (mimicking the acoustic effect of the head and ear canal) with an RLB (revised low-frequency B-weighting) filter that attenuates sub-bass energy below roughly 60 Hz. This means Short-Term Loudness is not a simple peak or RMS average: a sine wave at 100 Hz and a sine wave at 3 kHz at identical digital amplitudes will read differently. The perceptual weighting ensures the number you see on your meter corresponds much more closely to what a listener actually experiences as loudness — which is exactly why streaming platforms adopted LUFS-based normalization in the first place.
In a production workflow, Short-Term Loudness serves three overlapping functions. First, it provides a section-level diagnostic: as you loop a chorus, a drop, or a breakdown, the Short-Term readout tells you the perceptual loudness of that passage in isolation, independent of the quieter or louder sections surrounding it. Second, it functions as a dynamics-consistency tool — if your verses read −18 LUFS Short-Term and your choruses spike to −8 LUFS Short-Term, you have a 10 LU macro-dynamic swing that will translate very differently across streaming platforms that normalize to targets between −14 and −16 LUFS Integrated. Third, it serves as a real-time mastering reference: running your master through a calibrated loudness meter and watching the Short-Term readout while skipping through the track is one of the fastest ways to identify sections that are under-compressed, over-compressed, or dynamically inconsistent with the rest of the mix.
It is important to distinguish Short-Term Loudness from the other metrics in the BS.1770 / EBU R 128 family. Momentary Loudness (400 ms) reacts quickly enough to catch individual kicks and snare hits and is useful for gain-staging individual elements in real time, but it is too volatile to represent a musical passage. Integrated Loudness (whole-program average with a gating algorithm that ignores silence and very quiet passages) is what streaming platforms actually measure when they apply normalization — Spotify targets −14 LUFS Integrated, Apple Music −16 LUFS Integrated, YouTube −14 LUFS Integrated. Short-Term sits between these two: it responds fast enough to be musically meaningful but is stable enough to represent a listener's experience of a sustained passage. True Peak, the fourth metric in the family, operates independently — it catches inter-sample peaks that reconstructed DACs can clip on, and it is typically evaluated alongside rather than as part of the Short-Term measurement.
For producers working in genres where loudness wars still exert creative pressure — commercial EDM, hip-hop, pop — Short-Term Loudness is the metric that most directly predicts how a particular section of a track will feel after platform normalization. A chorus peaking at −6 LUFS Short-Term will be turned down to match the platform target, potentially losing the visceral impact the producer intended. A producer who understands Short-Term Loudness can design that chorus to hit −14 LUFS Short-Term with carefully sculpted transients and harmonic density rather than pure brickwall limiting — achieving perceived punch through dynamics rather than raw amplitude. This is the modern craft of loudness-aware mixing and mastering, and Short-Term Loudness is its central readout.
The technical architecture behind Short-Term Loudness follows the ITU-R BS.1770-4 algorithm precisely. The incoming audio signal — whether mono, stereo, or multichannel — first passes through a two-stage K-weighting filter applied independently to each channel. Stage one is a high-shelf filter with a gain of approximately +4 dB at frequencies above 1.5 kHz, implemented as a pre-filter to account for the acoustic effect of the listener's head in a free-field listening environment. Stage two is the RLB filter, a second-order high-pass filter with a −3 dB point at roughly 38 Hz, which reduces the disproportionate energy contribution of sub-bass content. After filtering, each channel's signal is squared (converting amplitude to power), and the mean square value is computed over the 3-second sliding window. For multichannel content, channel-specific weights are applied before summing: front left and right channels carry a weight of 1.0, centre carries 1.0, surround channels carry 1.5 (the +1.5 dB elevation accounts for their closer proximity in standard surround configurations), and the LFE channel is excluded entirely from the loudness sum.
The 3-second window updates continuously — it is not a block average that resets every three seconds but a true sliding window that advances sample by sample (or in practice, in short overlapping blocks of 100 ms per the EBU R 128 implementation). This means the reading on your meter at any given moment represents the weighted mean square energy of the previous 3 seconds of audio, recalculated with every new block of data. The result is then converted to a logarithmic LUFS scale using the formula: L = −0.691 + 10 × log₁₀(∑ weighted mean squares). The constant −0.691 is a calibration offset ensuring that a 1 kHz sine wave at 0 dBFS reads approximately −3.01 LUFS, aligning with traditional broadcast calibration references. In practice, this means a fully saturated 0 dBFS sine wave at 1 kHz held for more than 3 seconds will read approximately −3 LUFS Short-Term — a useful mental anchor when calibrating your monitoring chain.
The gating algorithm that defines Integrated Loudness does not apply to Short-Term Loudness. Integrated uses an absolute gate (ignoring blocks below −70 LUFS) and a relative gate (ignoring blocks more than 10 LU below the absolute-gated mean) to exclude silence and room noise from the long-term average. Short-Term has no gating: it measures whatever is present in the 3-second window without qualification. This makes it honest about silence — a 3-second fade-out ending in near-silence will drag the Short-Term reading downward in real time, which is the correct and expected behavior. Producers relying on Short-Term for section comparisons should therefore always evaluate it during the sustained body of each section rather than during transitions, fades, or sparse breakdowns where the trailing energy of the window may still include the previous denser passage.
Metering implementations vary slightly across tools. The EBU R 128 specification recommends that Short-Term Loudness meters display a range of at least −18 to 0 LU relative to the target (typically −23 LUFS for broadcast), update at a minimum refresh rate of 10 Hz, and offer a numerical readout alongside the bar graph. Professional tools including Nugen Audio VisLM, TC Electronic LM6, Waves WLM Plus, and iZotope Insight 2 all conform to this specification. Some DAW stock meters display Short-Term as the primary or secondary readout by default; others require configuration. The critical practical point is that Short-Term Loudness is the number to watch while mixing in real time — it integrates long enough to be meaningful but responds quickly enough that you can hear the passage it is measuring while the meter is reading it.
Understanding the relationship between Short-Term Loudness and the physical mix parameters that drive it is what separates producers who use LUFS meters as decorative widgets from those who use them as precision diagnostic instruments. Short-Term rises when sustained energy increases — denser arrangements, heavier compression reducing dynamic range, saturation adding harmonic energy, or simply more simultaneous elements. It falls when elements drop out, when compressors open up and allow transient peaks to dominate over sustained energy, or when heavy low-cut filtering removes bass content that the RLB filter would otherwise attenuate anyway. A skilled mastering engineer can read a Short-Term Loudness trace over time like a score, identifying every section, every fill, every buildup and drop from the shape of the curve alone.
Diagram — Short-Term Loudness: Short-Term Loudness 3-second sliding window shown alongside Momentary Loudness and Integrated Loudness across a typical track section, illustrating measurement responsiveness differences.
Every short-term loudness — hardware or plugin — operates on the same core parameters. Know these and you can work with any implementation.
The 3-second sliding window is not user-adjustable in compliant meters — it is a defined constant of the standard. It was chosen to correspond with the human auditory system's sustained loudness perception, roughly matching the duration of a short musical phrase. Understanding this fixed window helps producers know which transitions and passages the meter will and won't catch in real time.
K-weighting applies a high-shelf boost (approximately +4 dB above 1.5 kHz) and an RLB high-pass filter (–3 dB at ~38 Hz) to each channel before integration. This means sub-bass heavy genres (trap, bass music) will read lower on Short-Term Loudness than their raw RMS suggests, while bright, high-frequency-dense mixes (pop, acoustic) will read relatively higher. Producers should account for K-weighting when comparing genres.
In stereo, both channels carry equal weight of 1.0. In 5.1 surround, surrounds are weighted at 1.5 (adding approximately +1.76 dB of contribution) and the LFE is excluded. For stereo music production, channel weighting is transparent, but producers working in Dolby Atmos or surround formats will see Short-Term Loudness behave differently from their stereo references due to surround channel contributions.
EBU R 128 specifies a minimum refresh rate of 10 Hz (every 100 ms) for Short-Term Loudness meters. Most professional meters refresh at 10–25 Hz. A slower update rate means the visual feedback lags behind the audio — critical when using Short-Term as a real-time mixing aid. Always verify your meter's update rate in its settings; some DAW stock meters default to slower refresh rates that can mislead during dynamic passages.
Most EBU R 128 compliant meters display Short-Term Loudness on a scale from approximately –36 to 0 LUFS, though some broadcast-focused tools center the display around the –23 LUFS broadcast target. For music production, centering the display around –14 to –18 LUFS better suits streaming normalization targets. Choosing the right display range prevents the false confidence of a meter that compresses the relevant working range into a small portion of its scale.
Professional loudness meters present both a numerical LUFS value and a bar graph simultaneously. The numerical readout gives precise session-archivable data; the bar graph conveys trend and rate of change at a glance. During active mixing, the bar graph is more actionable — a rising bar through a chorus buildup is immediately readable. For documentation (deliverable verification, mastering notes), the numerical readout is what matters.
Session-ready starting points. Values are practical music production targets for streaming delivery; broadcast contexts use –23 LUFS Integrated as the reference anchor.
| Parameter | General | Drums | Vocals | Bass / Keys | Bus / Master |
|---|---|---|---|---|---|
| Typical Short-Term LUFS (verse) | −18 to −16 | −20 to −17 | −22 to −18 | −20 to −16 | −20 to −16 |
| Typical Short-Term LUFS (chorus/drop) | −14 to −10 | −13 to −10 | −16 to −12 | −15 to −11 | −12 to −9 |
| Macro-dynamic range (chorus vs. verse) | 4–8 LU | 5–9 LU | 4–7 LU | 4–7 LU | 3–7 LU |
| Streaming target (Integrated, reference) | −14 LUFS | −14 LUFS | −14 LUFS | −14 LUFS | −14 LUFS |
| Short-Term budget for loudest section | −10 to −8 | −10 to −8 | −12 to −9 | −11 to −8 | −9 to −7 |
| Acceptable Short-Term variance within a section | ±2 LU | ±3 LU | ±2 LU | ±2 LU | ±1.5 LU |
| True Peak ceiling (alongside Short-Term work) | −1 dBTP | −1 dBTP | −1.5 dBTP | −1 dBTP | −1 dBTP |
Values are practical music production targets for streaming delivery; broadcast contexts use –23 LUFS Integrated as the reference anchor.
The conceptual lineage of Short-Term Loudness begins in broadcast engineering, not music production. Through the 1990s and early 2000s, television broadcasters faced escalating viewer complaints about jarring loudness jumps between programs and commercial breaks — a problem that peak-based metering (the VU meter and its descendants) had never solved, because perceived loudness and peak amplitude are only loosely correlated. Research at the Communications Research Centre Canada, led by engineers including Glenn Soulodre and Michel Lavoie, culminated in a landmark 2003 paper proposing a loudness measurement algorithm built on mean square energy, frequency weighting, and multichannel summation. This work directly informed the development of ITU-R BS.1770, published in 2006, which formalized the K-weighting filter and the mean square loudness algorithm that underlies all LUFS measurements — including Short-Term Loudness — to this day.
The 3-second Short-Term integration window was not an arbitrary choice. Psychoacoustic research through the 1990s, including work by Zwicker and Fastl published in their seminal text Psychoacoustics: Facts and Models (1990, revised 1999), had characterized the human auditory system's sustained loudness perception as settling into a steady-state evaluation over roughly 1–3 seconds of continuous stimulation. The European Broadcasting Union's PLOUD group — the technical working group responsible for drafting EBU R 128, published in 2010 — adopted the 3-second window for Short-Term Loudness as the measurement most likely to correspond with a listener's moment-to-moment experience of a broadcast programme. The 400-millisecond Momentary window, by contrast, was retained for instantaneous monitoring. EBU R 128, authored with contributions from engineers at BBC, IRT Munich, and NRK Norway, became the definitive European broadcast loudness standard and established the –23 LUFS Integrated program loudness target that remains standard in European broadcasting.
The transition of Short-Term Loudness from broadcast to music production was catalyzed by the platform-level adoption of loudness normalization between 2013 and 2017. ReplayGain, introduced by David Robinson in 2001 and later refined by Frank Klemm, was the first widely implemented loudness normalization system for digital audio files — it targeted –89 dB RMS with a modified Fletcher-Munson A-weighting and had limited adoption outside audiophile software. Spotify introduced loudness normalization using an integrated LUFS target in 2013, initially at –11 LUFS before adjusting to –14 LUFS Integrated. Apple Music adopted –16 LUFS Integrated normalization in 2016. YouTube applied –14 LUFS Integrated normalization as a default in 2017. Tidal and Amazon Music followed with similar targets. These platform decisions effectively made Short-Term Loudness a production-critical metric: for the first time, mastering engineers needed to understand not just Integrated Loudness (the normalization anchor) but also Short-Term Loudness as the moment-to-moment loudness signature their masters would present after normalization gain was applied.
Hardware metering evolved in parallel. TC Electronic's LM2 Broadcast loudness meter (released 2010) was among the first hardware units to implement EBU R 128 Short-Term Loudness alongside Momentary and Integrated readouts, and its rackmount successors including the LM6 Radar became mastering studio standards. Nugen Audio's VisLM, released in 2010, brought full EBU R 128 and ITU-R BS.1770 metering to the software plugin domain at a time when DAW stock meters were still overwhelmingly peak-based. Waves Audio released the WLM Plus loudness meter in 2013, bringing affordable LUFS metering to a broader production audience. iZotope integrated loudness metering into Insight (first released 2012, significantly updated in Insight 2 in 2019) and eventually into Ozone, making Short-Term Loudness monitoring a native part of the mastering workflow for a generation of producers who had never worked in broadcast. By 2020, Short-Term Loudness had migrated from a specialist broadcast concept to a term found in bedroom producer tutorials, and every major DAW had incorporated at least basic LUFS metering into its stock plugin set.
In a practical mixing session, Short-Term Loudness functions primarily as a section-level diagnostic tool. The most effective technique is simple: place a calibrated LUFS meter on the master bus, loop each major section of your track individually — intro, verse, pre-chorus, chorus, bridge, drop, outro — and note the Short-Term LUFS readout for each. You are building a loudness map of your arrangement. A well-balanced pop or hip-hop track will typically show verses reading –18 to –16 LUFS Short-Term, pre-choruses rising to –16 to –14, and choruses landing at –13 to –10. If your chorus reads –8 LUFS Short-Term and your verse reads –20, the 12 LU gap may feel dramatic in your studio but will translate to a pumping, over-limited chorus and an extremely quiet verse after platform normalization adjusts the integrated level. The Short-Term map tells you where to apply gain automation, additional compression, or arrangement density changes before you touch the limiter.
For mastering engineers, Short-Term Loudness is a precision delivery verification tool. After the master is finalized, running a real-time loudness analysis (using tools like Nugen VisLM, iZotope Insight 2, or Youlean Loudness Meter 2 in offline analysis mode) reveals the Short-Term trace over the full duration of the track. A well-mastered commercial release will show a Short-Term trace that moves within a controlled range — perhaps 6–9 LU of variance across the full track — with predictable peaks on the loudest sections and a smooth trajectory through buildups and drops. Outliers — single-section Short-Term spikes more than 4 LU above the next loudest section, or a Short-Term plateau that barely moves across the entire track suggesting over-compression — are flagged for revision. The Short-Term trace is often included in mastering session notes delivered to clients as documentation of the dynamic profile of the master.
Vocals present a particularly instructive use case. A lead vocal bus, especially after compression and limiting, can read anywhere from –24 LUFS Short-Term (during sparse verses with natural breaths and phrases) to –12 LUFS Short-Term during dense, belted choruses. Applying a gentle vocal bus compressor with a slow attack (20–40 ms) and medium release (100–200 ms) to target a consistent Short-Term range of –18 to –14 LUFS across all vocal sections gives you a perceptually stable vocal that sits forward in the mix regardless of arrangement density. This is more effective than simply raising or lowering the vocal fader, because the Short-Term measurement reflects sustained energy rather than individual syllable peaks — it catches the sustained vowel-and-breath energy that defines intelligibility and presence in a full mix.
Electronic music producers working with dense arrangement automation — drops, tension builds, breakdown silences — use Short-Term Loudness to verify that their intentional dynamic contrasts are reading as designed rather than being masked by limiter behavior. A common target in festival-ready EDM is a drop that reads –10 to –8 LUFS Short-Term and a breakdown that drops to –24 to –20 LUFS Short-Term, creating a 14–16 LU perceived contrast that delivers maximum impact even after streaming normalization brings the integrated level to –14 LUFS. Without Short-Term monitoring, producers often discover in mastering that their brickwall limiter has compressed the dynamic contrast between drop and breakdown to 4–6 LU — a critical loss of impact that Short-Term metering during the mix stage would have flagged immediately.
One email a week. The techniques behind the terms — curated by working producers, not algorithms.
Abstract knowledge becomes practical when you can hear it in music you know. These tracks demonstrate short-term loudness used intentionally, at specific moments, for specific purposes.
The verse of 'bad guy' is a textbook study in controlled Short-Term Loudness: the sparse instrumentation (sub-bass, minimal percussion, close vocal) produces a verse reading of approximately –18 to –16 LUFS Short-Term, while the chorus — adding only a slightly denser kick pattern and doubled vocal — rises to only –14 to –13 LUFS Short-Term. This deliberately narrow 2–4 LU short-term contrast is what gives the track its unsettling intimacy rather than a conventional loud/quiet dynamic. Finneas's production maximizes K-weighted density through sub-bass concentration rather than mid-frequency density, which reads as punchy in the room but preserves the intimate Short-Term profile that survived Apple Music's –16 LUFS normalization without gain reduction artifacts.
HUMBLE. demonstrates aggressive Short-Term density management. The track opens with a brief piano motif reading approximately –22 LUFS Short-Term before the 808 kick and hi-hat grid hit at 0:18, immediately pushing the Short-Term reading to approximately –10 to –9 LUFS — a 12 LU jump in under a second of integration window overlap. Mike Will Made-It's production uses the 808's sustained sub energy (which K-weighting partially attenuates) combined with the crisp, bright snare and hi-hat energy (which K-weighting amplifies) to maintain a consistently high Short-Term reading throughout the main groove. The track's Integrated Loudness sits near –9 LUFS before Spotify's normalization pulls it to –14 LUFS — meaning listeners receive the track approximately 5 dB quieter than mastered, yet the Short-Term dynamics within the track remain fully intact.
Get Lucky was mastered during the apex of the loudness war yet managed to deliver a dynamically coherent Short-Term profile that has aged well in the streaming era. The main groove section reads approximately –11 to –10 LUFS Short-Term across most of its sustained duration — high by today's standards but relatively consistent, with less than 3 LU of variance between the verse and chorus due to the constant-energy Nile Rodgers guitar and live drum groove anchoring both sections. After Spotify's normalization, the track loses approximately 3 dB of level but retains its dynamic integrity, illustrating that Short-Term consistency within a track is as important as absolute level when it comes to surviving normalization. The guitar's sustained mid-frequency energy (amplified by K-weighting's high-shelf component) is the primary driver of the track's persistently high Short-Term reading.
Flume's production on Never Be Like You is an excellent study in Short-Term contrast architecture in electronic music. The verse, built around sparse piano chords, Kai's airy vocal, and minimal percussion, reads approximately –20 to –18 LUFS Short-Term — remarkably quiet for a mainstream electronic release. The drop at 1:30, introducing dense synth layers and a heavy kick pattern, pushes the Short-Term reading to approximately –10 to –9 LUFS, creating a genuine 9–10 LU short-term contrast. This contrast was preserved intact through Spotify's normalization because the Integrated Loudness of the whole track was already close to –14 LUFS — the dynamic range had been designed in rather than compressed out. Listen specifically at the 1:28 moment where the Short-Term meter will visibly climb over the 3-second window as the drop elements enter.
In European broadcast contexts, Short-Term Loudness is measured relative to a –23 LUFS Integrated target, with Short-Term readings typically staying within a –23 ±8 LU range for mixed-content programming. Hardware loudness controllers like the TC Electronic LM6 and Jünger Audio Level Magic implement real-time Short-Term monitoring with loudness history logging for compliance verification. Producers delivering audio for broadcast must verify that Short-Term peaks do not exceed +8 LU above the –23 LUFS Integrated reference for sustained periods, as defined in EBU R 128 Supplement 2.
For music production targeting Spotify, Apple Music, and YouTube, Short-Term Loudness is evaluated relative to the –14 LUFS Integrated streaming norm. In this context, producers use Short-Term to verify that their loudest sections (typically –10 to –8 LUFS Short-Term) sit at a controlled ceiling that allows the Integrated average to land near –14 LUFS without requiring aggressive limiting. Nugen VisLM and iZotope Insight 2 are the preferred tools for this workflow, offering section-by-section Short-Term annotation and offline analysis modes for post-render verification.
Some mixing engineers use Short-Term Loudness as a primary level reference in place of traditional VU or peak metering, calibrating their monitor gain so that a Short-Term reading of –18 LUFS corresponds to a comfortable, well-calibrated listening level. This approach, popularized by Bob Katz's K-System metering philosophy, uses Short-Term's 3-second integration as a de facto programme loudness meter that prevents ear fatigue and level creep during long sessions. Waves WLM Plus and Youlean Loudness Meter 2 are commonly used in this role due to their clean visual design and configurable target markers.
For podcast production and dialogue audio targeting Apple Podcasts (–16 LUFS Integrated) and Spotify Podcasts (–14 LUFS Integrated), Short-Term Loudness is used to verify that sustained speech passages — which have very different Short-Term profiles from music due to the stop-start nature of conversational speech — read consistently in the –19 to –16 LUFS Short-Term range during active speaking passages. Auphonic's cloud leveler uses Short-Term analysis internally to apply adaptive gain that maintains perceptual consistency across variable microphone distances and dynamic speakers.
These MPW articles put short-term loudness into practice — specific techniques, real tools, and applied workflows.