What does Spotify do to your track's loudness?
Loudness · 2 min read
Spotify plays every track at -14 LUFS by changing the playback gain, not the file: a loud master is turned down, a quiet one is turned up only as far as its peaks allow. Only the optional Loud setting pushes quiet tracks into a limiter.
Apple states -16 dB for Sound Check. YouTube, Tidal, Amazon Music, SoundCloud and Beatport Streaming publish no playback level. None of these numbers is a club target.
A loud master is turned down, nothing more
Spotify measures loudness at upload and does not "process or change your audio tracks before playback". A loud master gets a fixed negative gain, and "no additional distortion occurs".
We applied each published target to our 299 finished club records: arithmetic on loudness we had already measured, not a platform measurement.
So the louder master does not play louder. It plays at the same level as everything else, carrying whatever its limiting did to the waveform.
A quiet master is turned up only to its peaks
On Normal and Quiet, Spotify leaves 1 dB of headroom when it turns a track up. On Loud (-11 LUFS, Premium) it ignores the peaks and a limiter catches them at -1 dB: the one documented case where a platform changes your waveform.
Of our 299 records, Normal raises none. Loud raises 43, by at most 3.1 dB, and all 43 would push their true peak past -1 dB. Spotify's limiter reads sample values, which never sit above true peak, so 43 is an upper bound on how many it engages on.
-14 or -16?
Both, says the AES recommendation TD1008: under album normalisation the loudest track plays at -14 LUFS, which typically puts the album at -16. For a single normalised on its own, TD1008 recommends -16 and Apple states -16 dB. So a club single in a playlist plays 2 dB lower under Sound Check than on Spotify.
Not a club target
Every number here is a player turning its own volume knob; the file is not changed. In a booth, a record meets the next one at its own level plus whatever gain the DJ, or the DJ software, applies.
What does carry over is Spotify's -1 dBTP peak advice, because it is about the lossy encoder, not the player. What club records actually measure is on Club loudness, measured.
What this does not tell you
- What the platforms actually do. We read their documentation; we measured no platform's output.
- Exact gains. We assume a platform measures a track as our analyzer does.
- Your track, or your DJ software. 299 records are not your master, and some DJ applications match gain on their own.
How we measured the gains, and what each platform says in full
Method.
- Sources: the platforms' own help and artist pages, Apple's Apple Digital Masters brief, and AES TD1008. Every page was read on 2026-10-03. No third-party measurement is quoted as fact.
- Arithmetic: the 299 finished club records in our reference corpus (
mastering/analysis/tracks.csv, rows wheresourceisreference), integrated loudness and true peak as already measured by our analyzer (libebur128, ITU-R BS.1770). No new audio was measured. - Script:
content-social/drafts/scripts/I2_normalisation_gain.py. It subtracts each published target from each track's loudness and reports the 10th percentile, median and 90th percentile, numpy linear interpolation.
What each platform says, side by side.
| Platform | Published playback level | Album or track | Quiet tracks turned up? | Listener switch | Source |
|---|---|---|---|---|---|
| Spotify | -14 LUFS (Normal). Premium: Loud -11, Quiet -19 | album when an album plays in order, track on shuffle and in playlists | yes, limited by the track's peaks; Loud uses a limiter | on/off and level, mobile and desktop apps | Spotify for Artists (two pages), Spotify Support |
| Apple Music | "-16 dB", stated on an Apple Podcasts page, not in Apple Music docs | per song, or per album | not documented | Sound Check on/off | Apple Digital Masters brief, Apple Support, Apple Podcasts |
| YouTube | none published | not documented | not documented | Stable volume, off for YouTube Music and official music videos | YouTube Help (two pages) |
| Tidal | none in its help centre | not documented | not documented | not documented | Tidal help centre |
| Amazon Music | none published | not documented | not documented | Loudness Normalization on/off | Amazon Customer Service |
| SoundCloud | none stated for playback; recommends mastering to -14 LUFS | not documented | not documented | not documented | SoundCloud Help Center |
| Beatport Streaming | none in its help centre | not documented | not documented | not documented | Beatport Support |
| AES TD1008 (a recommendation, not a platform) | -16 LUFS per track, or -14 LUFS for the loudest track of an album | prefers album | yes, with true-peak limiting where needed | not applicable | AES TD1008 |
"Not documented" means the official page is silent, not that the feature is absent.
Spotify.
Spotify is the only platform that documents the mechanism, so its page "Loudness normalization on Spotify" is the one worth reading closely.
It is playback gain, not processing. "Loudness normalization is applied during playback. We measure loudness levels during the upload process, but we don't process or change your audio tracks before playback." The measurement is ITU 1770, the same family of standard our analyzer implements. Spotify does not say which revision or which gating it uses.
Album or track. "We normalize an entire album at the same time, so gain compensation doesn't change between tracks." But: "We adjust individual tracks when shuffling an album or listening to tracks from multiple albums (e.g. listening to a playlist)." Spotify does not say what loudness the album gain is computed from. A club single heard in a playlist is track-normalised.
Turned down. "Negative gain is applied to louder masters so the loudness level is -14 dB LUFS. This lowers the volume in comparison to the master - no additional distortion occurs."
Turned up, on Normal and Quiet. "We consider the headroom of the track, and leave 1 dB headroom for lossy encodings." Spotify's own example: a track at -20 LUFS with a true peak of -5 dBFS is lifted only to -16 LUFS, which puts its true peak at -1. So on Normal a quiet, dynamic master can stay quieter than -14 LUFS.
Turned up, on Loud. Loud is -11 LUFS and works differently: "We set this level regardless of maximum True Peak. We apply a limiter to prevent distortion and clipping in soft dynamic tracks. The limiter's set to engage at -1 dB (sample values), with a 5 ms attack time and a 100 ms decay time." This is the one documented case where a streaming platform changes the shape of your waveform rather than its level.
Who controls it. Spotify's Volume normalization help page gives Normal as the default level. It and the artist page describe the level choice as a Premium setting, and the on/off switch sits under Playback (mobile) or Audio Quality (desktop). Two exceptions, in Spotify's words: "The web player and 3rd-party devices (like TVs or speakers) don't use loudness normalization", and audio settings cannot be changed while playing to another device over Spotify Connect.
Lossless is included. Spotify states the same normalisation applies to its FLAC playback, in the normalisation page and in "Track not as loud as others?".
The -1 dBTP advice is about the encoder, not the normaliser. Spotify asks for a true peak below -1 dBTP, and below -2 dBTP if the master is louder than -14 LUFS, because "louder tracks are more susceptible to extra distortion when encoded for streaming". Read with the line above about measuring at upload and gaining at playback, the order is: your master is encoded at its own level, then turned down in the player. Turning a loud master down 5 dB at playback does not undo an overshoot the encoder already produced. AES TD1008 says the same at the standards level: true peak no higher than -1 dBTP "at the codec input of lossy-encoded streams", with more margin needed as the bit rate drops.
Apple Music and Sound Check.
Apple's Apple Digital Masters brief describes Sound Check in words, not numbers: it "first determines the loudness of a track and then stores that information in the file's metadata. The metadata is then used to raise or lower the volume of each track". It "can also correctly set the volume per album, rather than per song". The brief names ITU BS.1770 as a similar approach, and separately recommends leaving "at least 1 dB of headroom" because oversampling on playback can clip a file that sits at 0 dBFS.
Where -16 comes from. No Apple Music page we found states a level. The figure appears on Apple's audio requirements page for podcasters: "Apple uses this mechanism via the Sound Check setting in the iOS Music app. When enabled, content with loudness or Sound Check metadata will consistently have playback levels at -16 dB." That is Apple's own statement about Sound Check, but it sits on a podcast page, the unit is "dB" rather than LUFS, and it does not say whether music is handled the same way. We report it as Apple's stated level, with that caveat attached.
Not documented by Apple: whether a quiet song is turned up, whether any limiter is involved, and whether Sound Check is on by default.
Who controls it. Apple Support describes the switch: on a Mac, in the Music app's settings under Playback; on iPhone and iPad, under Settings, Apps, Music.
YouTube, Tidal, Amazon Music, SoundCloud, Beatport.
YouTube documents Stable volume, which "balances the range between quiet and loud, continuously adjusting volume levels" and is on by default, with a switch in the player settings. That is a dynamic adjustment, not a single target level, and YouTube's help says it "is turned off for YouTube Music and official music videos". YouTube publishes no loudness target for music.
Tidal has no help-centre article on loudness normalisation. We listed its English help centre through its public article index on 2026-10-03: 131 articles, none on the subject. Figures that circulate for Tidal come from third parties (see also, below).
Amazon Music documents a Loudness Normalization switch in its Android, iOS and FireOS apps that plays "all songs at the same loudness level". It states no level, no method, and no default.
SoundCloud says in its help centre that it "applies Loudness Normalization to your tracks as they're played to listeners" and advises: target "-14 dB integrated LUFS and keep it below -1 dB TP (True Peak) max", or below -2 dBTP for masters louder than -14 LUFS, for its Ogg Vorbis and AAC encodes. That is mastering advice, worded almost exactly like Spotify's. The page does not state the level SoundCloud plays at, whether quiet tracks are raised, or whether listeners can switch it off.
Beatport Streaming has no help-centre article on playback loudness: a search of its help centre for "loudness" returns nothing, and its audio-quality article lists codecs and bit rates only.
Why -16 and -14 are both "right".
AES TD1008 gives both numbers, and says when each applies:
| Content | Distribution loudness | Upper tolerance |
|---|---|---|
| Music, track-normalised | -16 LUFS | +0.2 LU |
| Music, loudest track of an album (album normalisation) | -14 LUFS | +0.2 LU |
| Speech | -18 LUFS | +1 LU |
The reasoning: "the loudest tracks are typically 2 LU louder than the average loudness of their albums", so setting the loudest track to -14 "will typically result in an integrated album loudness of -16 LUFS". TD1008 strongly prefers album normalisation, "even when playing tracks outside the album context, including shuffle play", and recommends -16 per track only where album normalisation is not possible. It also notes that upward normalisation "may require peak limiting" and recommends keeping content above -20 LUFS so weak players are not left too quiet.
So -14 and -16 are not two opinions about how loud music should be. Under album normalisation they describe the same album. The gap is real in one case that matters here: a single track normalised on its own. There Spotify documents -14, TD1008 recommends -16, and Apple's stated Sound Check level is -16 dB. A club single in a playlist is in exactly that case, so it plays 2 dB lower under Sound Check than on Spotify Normal.
Spotify does not say whether its album gain targets the loudest track or the album average, so we cannot say Spotify follows TD1008, only that its per-track number matches TD1008's album number.
What that does to a finished club record.
Arithmetic, not measurement: each published target minus each track's integrated loudness, over our 299 reference records. Negative means turned down. The left column is the track at the quiet end (10th percentile), the right column the loud end (90th percentile). The loudness values themselves are on Club loudness, measured.
| Target | Quiet-end track | Median track | Loud-end track |
|---|---|---|---|
| Spotify Normal (-14 LUFS, default) | -2.6 dB | -4.7 dB | -6.9 dB |
| Spotify Loud (-11 LUFS) | +0.4 dB | -1.7 dB | -3.9 dB |
| Spotify Quiet (-19 LUFS) | -7.6 dB | -9.7 dB | -11.9 dB |
| Apple Sound Check (-16 dB, as stated) | -4.6 dB | -6.7 dB | -8.9 dB |
| AES TD1008, track-normalised (-16 LUFS) | -4.6 dB | -6.7 dB | -8.9 dB |
What the script also shows:
- On Normal, essentially nothing is turned up. The one record quieter than -14 LUFS would need +0.1 dB, but its true peak already sits above -1 dBTP, so Spotify's 1 dB headroom rule allows no lift at all.
- On Loud, 43 of 299 records (14%) are turned up, by a median 0.6 dB and at most 3.1 dB. All 43 would push their true peak above -1 dB, by a median 1.8 dB and at least 0.2 dB, so all 43 reach Spotify's limiter. Spotify's threshold is on sample values and our figure is true peak, which is never lower, so 43 is an upper bound.
- At -16 and -19, no record is turned up. Every one is turned down.
More limits.
- Dates and revisions. Most platform pages show no date and change without notice. Spotify names no BS.1770 revision, and Apple says "dB", not LUFS.
- Anything about the silent platforms. We do not fill their gaps with third-party figures.
- Album playback. Our records are singles, and Spotify does not say what its album gain is computed from.
- How the Loud limiter sounds. We counted which records reach it; we did not render one.
Related
- How is loudness measured, and why is LUFS not dB?
- What does "just normalise it" actually do to a track?
- What does a 320 kbps MP3 throw away compared to the lossless file?
- Club loudness, measured
Sources
All accessed 2026-10-03.
- Spotify for Artists, "Loudness normalization on Spotify". No date shown. https://support.spotify.com/us/artists/article/loudness-normalization/
- Spotify for Artists, "Track not as loud as others?". No date shown. https://support.spotify.com/us/artists/article/track-not-as-loud-as-others/
- Spotify Support, "Volume normalization". No date shown. https://support.spotify.com/us/article/volume-normalization/
- Apple, "Apple Digital Masters: Music as the Artist and Sound Engineer Intended", Technology Brief (PDF; file metadata: "Apple Digital Masters 2.4.1", created 2021-06-08). https://www.apple.com/apple-music/apple-digital-masters/docs/apple-digital-masters.pdf
- Apple Support, "If music in Apple Music sounds quiet". Published 2025-03-26. https://support.apple.com/en-us/109331
- Apple Podcasts for Creators, "Audio requirements". No date shown. https://podcasters.apple.com/support/893-audio-requirements
- YouTube Help, "Control video volume on your device". No date shown. https://support.google.com/youtube/answer/14106294
- YouTube Help, "Video & audio quality enhancements". No date shown. https://support.google.com/youtube/answer/16619284
- TIDAL Support help centre, English article index (131 articles, none on loudness normalisation). https://support.tidal.com/hc/en-us
- Amazon Customer Service, "Turn On Loudness Normalization for Amazon Music Playback". No date shown. https://www.amazon.com/gp/help/customer/display.html?nodeId=TgG1d8mPjGwf885hRP
- SoundCloud Help Center, "Will SoundCloud play my track at the level it's mastered?". Last edited 2026-02-26. https://help.soundcloud.com/hc/en-us/articles/360053660014
- Beatport Support, "What is the audio quality like with Beatport Streaming?". Last edited 2026-05-05. https://support.beatport.com/hc/en-us/articles/27893825412500
- Audio Engineering Society, AESTD1008.1.21-9, "Recommendations for Loudness of Internet Audio Streaming and On-Demand Distribution", 24 September 2021 (supersedes TD1004). https://www.aes.org/technical/documentDownloads.cfm?docID=731
- ITU-R BS.1770, "Algorithms to measure audio programme loudness and true-peak audio level". https://www.itu.int/rec/R-REC-BS.1770
See also (secondary, not used for any figure above):
- audioXpress, "Tidal Implements Album Loudness Normalization and Activates It by Default for Mobile Players", 2017-11-27. Reports Tidal adopting album normalisation at -14 LUFS on the recommendation of Eelco Grimm; not a Tidal document. https://audioxpress.com/news/tidal-implements-album-loudness-normalization-and-activates-it-by-default-for-mobile-players
- 9to5Google, "YouTube Music rolling out 'Consistent volume' on Android, iOS", 2025-04-18. Reports a normalisation switch in YouTube Music; no YouTube Help page for it was found. https://9to5google.com/2025/04/18/youtube-music-consistent-volume/
We build enguetedl.ch; the analyzer above is ours.