Learn
What Is Loudness Normalisation (LUFS)?
If you have ever noticed that some songs on a streaming playlist sound quieter than others, or that a podcast episode suddenly blasts after a mellow track, you have experienced what loudness normalisation is designed to prevent. Every major streaming platform now adjusts the playback volume of audio so that listeners hear a consistent level from one piece of content to the next. This is a significant shift from how audio was distributed for decades, and it has real implications for how music, podcasts, and other audio content should be produced and mastered.
What loudness normalisation is
Loudness normalisation is the practice of adjusting the playback volume of audio content so that different tracks, episodes, or videos play at a roughly consistent perceived loudness. The key word is "perceived." Loudness normalisation does not simply look at peak levels or average signal amplitude; it uses a measurement that correlates with how the human ear perceives volume, accounting for the fact that we are more sensitive to midrange frequencies than to very low or very high ones.
When you play a playlist on Spotify, for example, the platform analyses each track and adjusts its playback gain so that quiet, dynamic recordings are not drowned out by heavily compressed, maximised masters. The listener hears a consistent volume without having to reach for the volume knob between songs.
This happens entirely on the playback side. The audio files themselves are not modified. The platform applies a gain offset, turning tracks up or down as needed, based on its measurement of each track's integrated loudness.
The loudness war and why normalisation exists
To understand why platforms adopted loudness normalisation, it helps to understand the problem it solves.
For decades, the music industry engaged in what is commonly called the loudness war: a progressive escalation of mastering levels driven by the belief that louder recordings sounded better and stood out more on radio, in record shops, and in playlists. Mastering engineers were pressured to make each release louder than the last, which they achieved by using heavy limiting and compression to push the average level of the audio as close to the digital ceiling as possible.
The cost was dynamic range. A loudness-war master has very little difference between its quietest and loudest moments. The quiet verse is nearly as loud as the explosive chorus. Transients are flattened, the sense of space and depth is reduced, and the music loses the natural ebb and flow that makes it feel alive. Technically, the audio is louder; perceptually, it is fatiguing and often sounds worse than a more moderate master.
This arms race made some sense in the context of radio, where louder songs did grab attention against the ambient noise of a car or a room. But streaming changed the equation. When every track in a playlist is normalised to the same loudness, the loudness-war master gains nothing. It just sounds flat and over-compressed compared to a well-mastered track with preserved dynamics. The platform turns it down to the same level as everything else, and now its squashed dynamic range is a liability, not an advantage.
Streaming normalisation effectively ended the loudness war for any producer or mastering engineer willing to accept the evidence. The incentive to crush dynamics is gone. The incentive to preserve them is stronger than it has been in years.
What LUFS means
LUFS stands for Loudness Units relative to Full Scale. It is the standard unit for measuring perceived loudness, defined by the ITU-R BS.1770 standard and refined by the EBU R128 recommendation in Europe. You may also encounter the term LKFS (Loudness, K-weighted, relative to Full Scale), which is numerically identical to LUFS; the two terms come from different standardisation bodies but describe the same measurement.
LUFS measures loudness as humans perceive it, not as a simple average of signal amplitude. The measurement applies a K-weighting curve that boosts midrange frequencies (where human hearing is most sensitive) and attenuates low frequencies (where we are less sensitive to loudness). It then integrates this weighted measurement over time to produce a single number representing the overall perceived loudness of the audio.
There are several types of LUFS measurement. Integrated LUFS is measured over the entire duration of the audio and is the figure that streaming platforms use for normalisation. Short-term LUFS is measured over a sliding three-second window and is useful for monitoring loudness variation during playback. Momentary LUFS uses a 400-millisecond window and captures the loudest moments.
LUFS values are negative, with 0 LUFS representing full scale. A track mastered at -14 LUFS is quieter (in terms of integrated loudness) than a track mastered at -8 LUFS. The lower the number, the quieter the perceived loudness.
It is important to distinguish LUFS from peak levels. Peak level measures the loudest instantaneous point in the audio signal, while LUFS measures perceived loudness over time. A track can have a high peak level but a low integrated LUFS (a dynamic recording with occasional loud moments) or a low peak level with a high integrated LUFS (a heavily compressed recording where everything is consistently loud). The two measurements tell you different things, and both matter in mastering.
Platform loudness targets
Each major streaming platform has adopted a specific loudness target, though the exact numbers and behaviours differ.
Spotify normalises to -14 LUFS integrated. If your track is louder than -14 LUFS, Spotify turns it down. If it is quieter, Spotify's behaviour depends on the user's settings: with normalisation enabled (the default), quieter tracks may be turned up, though Spotify applies a limiter to prevent clipping when doing so. The practical effect is that tracks mastered to -14 LUFS play back at their native level without any adjustment.
Apple Music uses Sound Check, which targets approximately -16 LUFS. Sound Check has been part of Apple's ecosystem since the iTunes era, predating most other platforms' normalisation efforts. Because Apple's target is lower than Spotify's, a track mastered at -14 LUFS will be turned down slightly on Apple Music.
YouTube normalises to approximately -14 LUFS for music content, though its behaviour is less precisely documented than Spotify's. YouTube does turn down loud content but has historically been less consistent about turning up quiet content.
Broadcast standards differ from streaming. The EBU R128 standard, used across Europe, specifies -23 LUFS for broadcast content. The ATSC A/85 standard in the United States specifies -24 LKFS (equivalent to -24 LUFS). These broadcast targets are significantly quieter than streaming targets, reflecting the wider dynamic range available in broadcast environments and the different listening contexts (television typically plays in quieter settings than mobile streaming).
Podcast platforms generally target -16 to -19 LUFS, though specific targets vary and are less rigidly enforced than music targets. Apple Podcasts and Spotify both apply normalisation to podcast content, and producing at around -16 LUFS is a widely accepted standard for podcast loudness.
How normalisation works in practice
The mechanics are worth understanding because they explain why mastering decisions matter more, not less, in the streaming era.
When a track louder than the platform's target is uploaded, the platform calculates how many decibels of gain reduction are needed to bring the integrated loudness down to the target. A track mastered at -8 LUFS on a platform targeting -14 LUFS will be turned down by 6 dB. The platform does not re-master or re-compress the audio; it simply reduces the playback volume. The track's internal dynamics, its crest factor, its tonal balance, all remain exactly as the mastering engineer set them. It just plays back quieter.
This is the crucial insight. A heavily limited, loudness-war master at -8 LUFS will be turned down to play at the same perceived loudness as a dynamic, well-crafted master at -14 LUFS. But the loudness-war master has sacrificed its dynamics to achieve that -8 LUFS level. Once the platform turns it down, those dynamics are still gone, and the track sounds flat and lifeless compared to the more dynamic master playing at the same volume.
In other words, the loud master is turned down and sounds worse. The dynamic master plays at its native level and sounds better. The loudness war, on streaming platforms, is a losing strategy.
When a track is quieter than the target, behaviour varies by platform. Some platforms turn it up; some leave it alone. Spotify's normalisation can apply positive gain to quiet tracks, but this can introduce clipping if the track's peaks are already close to 0 dBFS. This is one reason why leaving headroom (mastering with a true peak ceiling below 0 dBFS) is important even when you are not mastering loud.
Practical implications for mastering
Knowing how normalisation works leads to concrete mastering decisions.
Do not master as loud as possible. The old instinct to push the limiter until the waveform is a solid block serves no purpose on streaming platforms. It costs you dynamics and gains you nothing.
Aim for the target of your primary platform. If your music will be streamed primarily on Spotify, mastering to around -14 LUFS integrated is sensible. If your primary audience is on Apple Music, -16 LUFS gives you a touch more dynamic range. If you are producing for broadcast, the targets are considerably lower.
There is no single correct loudness for all music. A dense, energetic pop or electronic track might sit comfortably at -10 to -12 LUFS, which means the platform will turn it down slightly but its dense character is preserved. A sparse acoustic recording might naturally sit at -16 to -18 LUFS. Forcing either to an arbitrary target for its own sake misses the point. The goal is to serve the music, not hit a number.
Preserve dynamics. This is the biggest practical shift. With loudness normalisation in place, dynamic masters sound better on streaming platforms than squashed ones. A well-mastered track with 8 to 12 dB of dynamic range, where the quiet moments are quiet and the loud moments are loud, creates more impact and engagement than a track compressed to 4 dB of range.
When preparing masters for distribution, keep your source files in lossless formats and at the highest sample rate and bit depth your session supports. If you are delivering stems for mastering, include notes about the intended loudness target so the mastering engineer can optimise accordingly.
True peak and inter-sample peaks
Loudness targets are only half of the mastering equation for streaming. The other half is true peak level.
A true peak measurement accounts for what happens to the audio signal between the digital samples. When a DAC (digital-to-analogue converter) or a lossy encoder reconstructs the audio from its digital samples, the reconstructed waveform can exceed the level of any individual sample. These inter-sample peaks are invisible to a standard peak meter but can cause distortion in DACs and clipping in lossy encoders.
The widely recommended true peak ceiling for streaming is -1 dBTP (one decibel below true peak full scale). This provides enough headroom to prevent inter-sample peaks from clipping when the audio is converted to lossy formats like AAC or Ogg Vorbis for streaming playback. Some engineers use -1.5 or -2 dBTP for additional safety, particularly if the audio will undergo multiple encoding passes.
Most modern mastering limiters include a true peak mode that prevents the output from exceeding a specified true peak ceiling. Enabling this mode and setting it to -1 dBTP is a simple, reliable way to ensure your masters are streaming-safe.
How to measure LUFS
Measuring loudness in LUFS requires a meter that implements the ITU-R BS.1770 algorithm. Fortunately, these are widely available.
Most modern DAWs include a built-in loudness meter or can display LUFS measurements in their metering options. Logic Pro, Cubase, Reaper, and Pro Tools all have some form of integrated loudness metering, though the implementation and ease of access varies.
Dedicated loudness metering plugins provide more detailed information. Youlean Loudness Meter is a popular free option that displays integrated, short-term, and momentary LUFS along with a histogram of loudness over time. It also shows whether your audio meets the targets for various streaming platforms and broadcast standards. Commercial options like iZotope Insight, Nugen Audio VisLM, and TC Electronic's loudness meters offer additional features for professional workflows.
The key number to watch is integrated LUFS, measured over the full duration of the track or programme. Short-term and momentary readings are useful for understanding how loudness varies through the piece, but the integrated figure is what streaming platforms use for normalisation.
When metering, play the entire track from start to finish. Integrated LUFS is a full-programme measurement; checking only the chorus or only the verse will give you an inaccurate picture of the track's overall loudness.
Loudness for podcasts
Podcast loudness normalisation follows the same principles as music but with different targets and considerations.
Speech benefits from consistent loudness more than music does. A listener following a conversation does not want to strain to hear a quiet speaker and then be startled by a loud one. Podcast loudness targets are generally lower than music targets: -16 LUFS is a common aim, with -19 LUFS sometimes used for speech-heavy content. Apple's podcast specifications recommend -16 LUFS integrated with a -1 dBTP true peak ceiling.
Podcast processing typically involves compression to even out the dynamic range of speech (reducing the gap between the softest and loudest syllables), followed by a limiter to catch peaks. This is distinct from the heavy limiting used in loudness-war music mastering; podcast compression aims for clarity and consistency, not maximum loudness.
If you are producing podcasts, transcription tools work most reliably when the source audio has consistent, well-managed loudness. Extreme dynamic range or very quiet passages can reduce transcription accuracy. Normalising your podcast audio to around -16 LUFS before export serves both the listener and any downstream processing.
Loudness for video
Video content faces its own loudness considerations, particularly when it includes both dialogue and music. The dialogue must be intelligible, the music must support the mood without overwhelming the speech, and the overall loudness must comply with the target of the distribution platform.
YouTube's normalisation means that mastering a video's audio too loud will result in the platform turning it down, which can exaggerate any dynamic imbalance between dialogue and music. A better approach is to mix the audio so that dialogue sits at a comfortable level, typically around -16 to -20 LUFS for the dialogue alone, with music and effects balanced underneath, and then ensure the overall programme loudness lands near the platform's target.
For video creators working with separate audio and video files, keeping the audio well-organised and clearly labelled, with loudness targets documented, prevents problems when handing off to an editor or colourist. A cloud workspace that supports annotations can be useful for noting loudness targets and mixing decisions alongside the files themselves.
The end of the loudness war
Streaming loudness normalisation has done what decades of criticism from audiophiles and mastering engineers could not: it has removed the commercial incentive to master as loud as possible. When every track plays back at the same perceived loudness regardless of how hard it was limited, the only thing that heavy limiting costs you is dynamic range. There is nothing left to gain.
This does not mean that every modern master should be whisper-quiet. Some genres, electronic music, certain styles of pop, heavy metal, are meant to sound dense and energetic, and moderate limiting is part of their character. But the extreme limiting that characterised the worst of the loudness war, pushing tracks to -5 or -6 LUFS with less than 3 dB of dynamic range, no longer serves any purpose on streaming platforms.
The winning strategy now is the same one that mastering engineers have advocated for years: master the music to serve the song. Use limiting where it enhances the energy, but preserve the dynamics that give the music its shape and emotion. A track that breathes, that has quiet moments and loud moments, that builds and releases, will sound more engaging on a streaming platform than one that has been crushed flat.
For music creators delivering finals for distribution, this shift is unambiguously positive. You can focus on making the music sound its best rather than making it sound its loudest. The platforms have taken the volume knob out of the competition, and what remains is craft.
Keeping your masters, project files, and delivery assets well-organised means you can revisit loudness decisions as platform targets evolve. What Spotify targets today may shift in the future, and having access to your lossless masters and session files, rather than only a final limited bounce, gives you the flexibility to re-master without starting from scratch. A workspace like Fabric that lets you find and retrieve those files quickly makes that future-proofing practical rather than theoretical.
Frequently asked questions
What LUFS should I master to for Spotify?
Spotify normalises to -14 LUFS integrated. Mastering to around -14 LUFS means your track will play back at its native level without being turned up or down. However, if your music is naturally louder or quieter than -14 LUFS, it is better to serve the music than to hit the number exactly. A dense electronic track at -10 LUFS will be turned down by 4 dB, which is fine; its character is preserved.
Does loudness normalisation change my audio file?
No. Normalisation on streaming platforms applies a gain offset during playback. Your uploaded file is not modified. The platform stores the original and adjusts the volume in real time when it is played. This is a non-destructive process.
Should I master differently for Apple Music and Spotify?
The targets differ (-16 LUFS for Apple, -14 LUFS for Spotify), but mastering a single version at around -14 LUFS works well for both. Apple will turn it down by about 2 dB, which is a negligible adjustment. If your primary audience is on Apple Music and you want to maximise dynamic range, mastering closer to -16 LUFS is a reasonable choice.
What is the difference between LUFS and dB?
Decibels (dB) measure signal amplitude. LUFS measures perceived loudness over time, using a frequency weighting that reflects human hearing sensitivity. A track can have a high peak level in dB but a low integrated LUFS (dynamic music with occasional loud peaks), or a moderate peak level with a high LUFS (heavily compressed music with consistently high average levels). Both measurements are useful but describe different properties.
Can a track be too quiet for streaming?
In theory, yes. If a track is significantly quieter than the platform's target, the platform may turn it up, which can introduce clipping if the peaks are already close to 0 dBFS. In practice, most well-mastered music falls within a range that platforms handle comfortably. If your music is intentionally very quiet (ambient, drone, certain classical recordings), ensure that you have adequate headroom above your peaks to accommodate any positive gain the platform might apply.
What is the loudness penalty?
The "loudness penalty" refers to the amount of gain reduction a platform applies to audio that exceeds its target. A track at -8 LUFS on a platform targeting -14 LUFS receives a 6 dB loudness penalty. Tools like the Loudness Penalty website allow you to check how much gain adjustment each major platform will apply to your master.
Do podcasts need loudness normalisation?
Yes. Podcast platforms apply normalisation, and listeners expect consistent volume between episodes and between shows. Aiming for -16 LUFS integrated with a -1 dBTP true peak ceiling is a widely accepted standard. Consistent loudness also improves the accuracy of automated transcription and accessibility features.
Does mastering loud damage audio quality?
Aggressive limiting to achieve extreme loudness does reduce audio quality, though not in the same way as low-bitrate encoding. The damage is to dynamics and transient detail rather than to frequency content. Heavy limiting flattens the difference between quiet and loud passages, softens transient attacks, and can introduce audible distortion on dense material. The audio is not corrupted, but it loses expressive qualities that cannot be recovered.
What LUFS target should I use for YouTube?
YouTube normalises to approximately -14 LUFS, similar to Spotify. For video content with dialogue, music, and effects, the overall programme loudness should be around -14 LUFS, with dialogue sitting at a level where it is clearly intelligible against the other elements. A true peak ceiling of -1 dBTP is recommended.
Can I check the LUFS of an existing track?
Yes. Import the track into any DAW with a LUFS meter (most modern DAWs have one) or use a standalone loudness meter plugin. Play the track from beginning to end, and the meter will display the integrated LUFS. Free tools like Youlean Loudness Meter make this easy, and several online tools and mobile apps offer quick measurements as well.