ihateaudio

Loudness targets by platform

What every major platform expects, in LUFS and true peak, and what it does to your file if you miss.

Every streaming service normalizes playback loudness. That single fact makes the loudness war pointless: if you master a track four decibels louder than the target, Spotify turns it down four decibels on the way out, and all you have achieved is squashing your own dynamic range for nothing.

The number that matters is integrated LUFS, loudness averaged across the whole programme, weighted for how human hearing actually works and gated so that silence between phrases does not drag the reading down. It is defined by ITU-R BS.1770, and it is what all of these platforms measure.

Music streaming

PlatformTargetTrue peakTypeNotes
Spotify-14 LUFS-1 dBTPMusicNormalization is on by default. Louder masters are turned down; quieter ones are turned up only in the “Loud” setting.
Apple Music-16 LUFS-1 dBTPMusicSound Check is on by default on iOS. Apple’s mastering guidance also asks for -1 dBTP.
YouTube-14 LUFS-1 dBTPMusic & videoTurns loud uploads down but never turns quiet ones up, so undershooting costs you real loudness.
Amazon Music-14 LUFS-2 dBTPMusicSlightly more conservative peak ceiling than most.
Tidal-14 LUFS-1 dBTPMusicNormalization can be switched off by the listener.
Deezer-15 LUFS-1 dBTPMusicSits between the streaming consensus and Apple’s target.
SoundCloud-14 LUFS-1 dBTPMusicApplies normalization on playback for most listeners.

Podcasts and spoken word

PlatformTargetTrue peakTypeNotes
Apple Podcasts-16 LUFS-1 dBTPPodcast (stereo)The de facto podcast standard. Use -19 LUFS for mono files.
Spotify Podcasts-14 LUFS-1 dBTPPodcastMatches Spotify’s music target rather than the podcast convention.
Google / YouTube Podcasts-16 LUFS-1 dBTPPodcastFollows the -16 convention.
Audible / ACX-18 to -23 RMS LUFS-3 dBTPAudiobookACX specifies RMS, not LUFS, and requires a noise floor below -60 dBFS. Its check is strict and automated.

Broadcast and video

PlatformTargetTrue peakTypeNotes
EBU R128 (Europe)-23 LUFS-1 dBTPBroadcastTolerance ±0.5 LU. The origin of the modern LUFS approach.
ATSC A/85 (US)-24 LUFS-2 dBTPBroadcastThe CALM Act enforces this for US television.
Netflix-27 LUFS-2 dBTPStreaming videoMeasured with dialogue gating rather than programme loudness.

Which number should you actually aim for?

If you are releasing music to every service at once, master to-14 LUFS integrated with a -1 dBTP ceiling. That satisfies the majority outright, and Apple Music turning it down a further two decibels is inaudible and harmless. Chasing each platform separately is not worth the effort for anything other than a dedicated Apple release.

For podcasts, use -16 LUFS for stereo and-19 LUFS for mono. The difference exists because a mono file played through two speakers measures roughly 3 LU louder than the same material in stereo, so mono needs a lower target to sound equally loud.

For anything going to broadcast, use the number the broadcaster gives you and do not improvise. Broadcast delivery is automatically checked and rejected.

True peak is not the same as peak

A digital file stores samples, but playback reconstructs a continuous waveform that passes through them. And that curve can rise higher than any individual sample. A file that shows 0.0 dBFS on a normal peak meter can genuinely reach +1.5 dB once a converter or an MP3 encoder reconstructs it, which clips.

That is why every target above specifies a true peak ceiling below zero. Lossy encoding makes it worse: the encoder discards information and the decoded waveform is not the one you measured, so peaks move. A -1 dBTP ceiling is the standard insurance, and -2 dBTP is safer if you are encoding to a low bitrate.

Loudness range, and why quiet passages matter

Loudness range (LRA, measured in LU) describes how much the loudness varies across a programme. A heavily compressed pop master might sit around 3–5 LU; an orchestral recording can exceed 15 LU; a well-produced podcast usually lands between 5 and 8 LU.

There is no correct value, but there are wrong ones for a context. A podcast with an 18 LU range is unlistenable in a car, because the quiet parts vanish under road noise while the loud parts make you reach for the volume. If that is your situation, compression before normalization is the fix, not a louder target.

Measure and fix

You can check any file against this table with theloudness meter, which reports integrated LUFS, loudness range and true peak using the same BS.1770-4 measurement described above. The audio normalizerwill then move a file to any target on this page while holding the true peak below its ceiling. Both run entirely in your browser.