Public reference

Platform delivery targets

What each service wants, what it does to your file, and how much of that the service actually publishes. Generated from the same file the engine masters against, so this page cannot drift away from the tool.

Link to it, quote it, correct it. If one of these figures is wrong we would rather hear about it than keep mastering against it.

Platforms
6
Last checked
2026-08-07
Checks kept
1
Source
the engine's own table
A value marked consensus is not published by the platform and may be wrong.
PlatformTargetTrue peakCodecNormalisesConfidenceSourceWhat that means
Spotify-14 LUFS-1.0 dBTPlibvorbis 320kyesprimarySpotify for ArtistsSpotify normalises to -14 LUFS and publishes the figure itself. Listeners can override it: Loud -11, Normal -14, Quiet -19. Spotify's own guidance is to stay under -1 dBTP for a master at -14 LUFS, and under -2 dBTP for anything louder.
Apple Music-16 LUFS-1.0 dBTPaac 256kyesconsensusApple Digital MastersApple publishes no LUFS figure at all. The -16 LUFS everyone quotes for Sound Check is industry consensus, not documentation. Treat it as an estimate, because that is what it is.
YouTube / YouTube Music-14 LUFS-1.0 dBTPlibopus 128kyesconsensusnot publishedYouTube only ever turns tracks down, never up. A track quieter than the target stays quiet, which is the one case where mastering conservatively costs you something real.
TikTok / Reels / Shorts-14 LUFS-1.0 dBTPaac 192knoconsensusnot publishedThe feed does not normalise, so a louder upload really is louder here. We still deliver at -14 LUFS with a safe true peak, because pushing lossy source material harder buys distortion on phone speakers rather than loudness.
Amazon Music-14 LUFS-2.0 dBTPaac 256kyesconsensusnot publishedAsks for -2 dBTP where everyone else accepts -1 dBTP. A master aimed at -1 dBTP structurally cannot satisfy Amazon, and we would rather say that than quietly mark it green.
SoundCloud / Bandcamp-14 LUFS-1.0 dBTPlibmp3lame 320knoconsensusnot publishedNo normalisation is documented, so the master plays back exactly as loud as you delivered it.

Every check of this table, kept with its date, and what changed between them

Confidence. The platform publishes this number itself. The platform publishes no number. This is the industry figure, and it may be wrong. Of the 6 platforms here, 1 publishes its own figure. The rest are industry values, and every master made against them inherits whatever error is in them.

Per platform

Detail, with an anchor each

Spotify

Loudness target
-14 LUFSprimary
True peak
-1.0 dBTPprimary
Codec
libvorbis320k, .ogg
Normalises
yesplayback level is set by the platform, not by you
Receives the master
via distributorgets a vote on the master's ceiling

Spotify normalises to -14 LUFS and publishes the figure itself. Listeners can override it: Loud -11, Normal -14, Quiet -19. Spotify's own guidance is to stay under -1 dBTP for a master at -14 LUFS, and under -2 dBTP for anything louder.

Source. https://support.spotify.com/us/artists/article/loudness-normalization/ Checked 2026-08-07. Encoder speed: fast enough to measure on a whole track.

Apple Music

Loudness target
-16 LUFSconsensus
True peak
-1.0 dBTPconsensus
Codec
aac256k, .m4a
Normalises
yesplayback level is set by the platform, not by you
Receives the master
via distributorgets a vote on the master's ceiling

Apple publishes no LUFS figure at all. The -16 LUFS everyone quotes for Sound Check is industry consensus, not documentation. Treat it as an estimate, because that is what it is.

Source. https://www.apple.com/apple-music/apple-digital-masters/ Checked 2026-08-07. Encoder speed: slow, so long tracks are measured on their loudest passages and the report says so.

YouTube / YouTube Music

Loudness target
-14 LUFSconsensus
True peak
-1.0 dBTPconsensus
Codec
libopus128k, .opus
Normalises
yesplayback level is set by the platform, not by you
Receives the master
via distributorgets a vote on the master's ceiling

YouTube only ever turns tracks down, never up. A track quieter than the target stays quiet, which is the one case where mastering conservatively costs you something real.

Source. not published Checked 2026-08-07. Encoder speed: fast enough to measure on a whole track.

TikTok / Reels / Shorts

Loudness target
-14 LUFSconsensus
True peak
-1.0 dBTPconsensus
Codec
aac192k, .m4a
Normalises
noplayback level is exactly what you delivered
Receives the master
direct uploadgets its own render, so it cannot cost you loudness elsewhere

The feed does not normalise, so a louder upload really is louder here. We still deliver at -14 LUFS with a safe true peak, because pushing lossy source material harder buys distortion on phone speakers rather than loudness.

Source. not published Checked 2026-08-07. Encoder speed: slow, so long tracks are measured on their loudest passages and the report says so.

Amazon Music

Loudness target
-14 LUFSconsensus
True peak
-2.0 dBTPconsensus
Codec
aac256k, .m4a
Normalises
yesplayback level is set by the platform, not by you
Receives the master
via distributorgets a vote on the master's ceiling

Asks for -2 dBTP where everyone else accepts -1 dBTP. A master aimed at -1 dBTP structurally cannot satisfy Amazon, and we would rather say that than quietly mark it green.

Source. not published Checked 2026-08-07. Encoder speed: slow, so long tracks are measured on their loudest passages and the report says so.

SoundCloud / Bandcamp

Loudness target
-14 LUFSconsensus
True peak
-1.0 dBTPconsensus
Codec
libmp3lame320k, .mp3
Normalises
noplayback level is exactly what you delivered
Receives the master
direct uploadgets its own render, so it cannot cost you loudness elsewhere

No normalisation is documented, so the master plays back exactly as loud as you delivered it.

Source. not published Checked 2026-08-07. Encoder speed: fast enough to measure on a whole track.

Why this matters

A target is not a promise about what the listener hears

Every figure above describes the file you hand over. The file that reaches the listener has been through the codec in the fourth column, and that codec moves the peaks. On the demo dataset the largest lift was +0.86 dB.

See the measurementMaster a track

Verdicts used on this site: ok within the platform's own true-peak limit. tight under full scale, but above the true peak this platform asks for. no margin left. clips above 0 dbtp after the codec. this distorts audibly.