Loudness normalization is why you rarely need to reach for the volume control when a playlist moves from one track to the next. Players measure how loud each item is and turn it up or down at playback so everything plays at a similar level. It also changed the case for making masters as loud as possible: on a normalizing platform, extra loudness is simply turned down. This article explains how normalization works, what the major platforms actually document, and how to master with it in mind.
How normalization works
- Measure. The platform measures the integrated loudness of each track or episode. Spotify, for example, measures according to the ITU 1770 standard, which reports LUFS.
- Compare. It compares that value with a reference level, such as -14 LUFS.
- Adjust at playback. It applies the difference as gain while playing: a -10 LUFS track is turned down 4 dB, and a -18 LUFS track is turned up 4 dB if it has the headroom.
On Spotify, the file itself is not changed. Spotify states that it does not process or change audio tracks before playback; the gain is applied in the player, which is also why Premium listeners can choose between different levels.
Turning down is easy, turning up is not
Turning a loud file down is harmless: it cannot clip. Turning a quiet file up can push its peaks past full scale. Spotify handles this by limiting positive gain to the file's headroom, leaving 1 dB of headroom for lossy encodings. In its own example, a track at -20 LUFS with a true peak of -5 dBFS is raised only to -16 LUFS, not to -14.
The exception is Spotify's Loud mode, which raises tracks regardless of their true peak and relies on a limiter set to engage at -1 dB (sample values), with a 5 ms attack and a 100 ms decay.
Spotify: documented, for music
Spotify for Artists publishes a detailed description of its music normalization. Premium listeners can choose between three levels:
| Mode | Reference level |
|---|---|
| Loud | -11 LUFS |
| Normal (default) | -14 LUFS |
| Quiet | -19 LUFS |
Whole albums are normalized together, keeping the level differences between their tracks; individual tracks are adjusted when shuffling an album or playing tracks from several albums, as in a playlist. The web player and third-party devices such as TVs and speakers do not normalize. Spotify's advice for music masters is to target -14 LUFS integrated with true peak below -1 dBTP, and below -2 dBTP if the master is louder than -14 LUFS.
All of this is about music. Spotify for Creators publishes no loudness target for podcast episodes and does not say whether this normalization applies to them, so "Spotify requires -14 LUFS for podcasts" is a misreading of the music page.
YouTube: stable volume, no published number
YouTube does not publish a loudness target. Its help centre describes two related features:
- Stable volume, on by default, which works by "continuously adjusting volume levels to reduce variations in sound". It is turned off for YouTube Music and official music videos.
- Automatic adjustments to volume levels and sound mixing, one of YouTube's video and audio quality enhancements. Creators can opt out for future uploads in YouTube Studio under Settings, Channel, Advanced settings.
As described, Stable volume adjusts levels continuously within a video rather than applying one fixed offset per video. The commonly quoted -14 LUFS figure comes from third parties who read the "content loudness" value in the Stats for nerds overlay and measured playback; YouTube's own Stats for nerds help page does not mention loudness. It is a reasonable working level for video, but it is an observation, not a YouTube specification.
Apple: Sound Check and -16
For podcasts, Apple asks creators to precondition audio to about -16 LKFS ±1 dB, with true peaks no higher than -1 dBFS, before encoding. Apple states that content with loudness or Sound Check metadata will consistently play back at -16 dB. What loudness should a podcast be covers the podcast side in detail.
Official numbers and observed ones
| Platform | What is published | Status |
|---|---|---|
| Apple Podcasts | -16 LKFS ±1 dB; true peak no higher than -1 dBFS | Official |
| Spotify music | -14 LUFS (Normal), -11 (Loud), -19 (Quiet) | Official |
| Spotify podcasts | No loudness target | Not published |
| YouTube | Stable volume, with no LUFS figure | -14 LUFS is a third-party observation |
| Amazon Music | Can play all songs at the same loudness level; no LUFS figure | Not published |
| Amazon Alexa skills and flash briefings | -14 LUFS average; true peak no higher than -2 dBFS; may reject content below -19 or above -9 LUFS | Official |
LUFS figures you may see quoted for Amazon Music, Deezer or Tidal do not appear in those services' own help pages, as far as we could find. When a number is not documented, treat it as an estimate that can change without notice. The loudness targets reference lists the documented values with their sources.
Why mastering too loud backfires
Before normalization, a louder master stood out. On a normalizing platform it is turned down to the same level as everything else, so loudness bought with heavy compression and limiting gains nothing, while the costs remain:
- Flattened delivery. After normalization, a heavily limited master plays at the same loudness as a dynamic one, but with its emphasis and contrast squashed.
- Distortion. Hard limiting adds audible distortion, and turning the file down later does not remove it.
- More noise and breaths. Compressing speech hard raises everything quiet, including room noise, breaths and mouth sounds.
- Peak problems after encoding. Very loud masters leave little headroom, so inter-sample peaks and encoder overshoot are more likely to clip. Spotify, for instance, asks for a true peak below -2 dBTP on masters louder than -14 LUFS. See true peak vs sample peak.
Masters that are too quiet have the opposite problem: a player may not be able to raise them fully without exceeding its headroom limit, so they end up quieter than their neighbours. The goal is to land on the reference level for your main platform, not above or below it.
What to master to
- Podcasts: -16 LUFS integrated with true peak at or below -1 dBTP; -19 LUFS for mono files.
- Online video: -14 LUFS is a common working level, with true peak at or below -1 dBTP. YouTube publishes no target.
- ACX audiobooks: the ACX RMS window of -23 to -18 dB, not a LUFS figure.
- Broadcast: the standard your broadcaster specifies, such as EBU R 128 or ATSC A/85.
The free gain calculator works out how many decibels a file needs to reach any of these targets and whether its peaks will still fit.
LufsReady's presets follow the same split: Podcast at -16 LUFS (mono at -19), Online video at -14 LUFS, labelled as a commonly observed level rather than an official YouTube figure, and broadcast presets for EBU R 128 and ATSC A/85.
Sources
- Spotify for Artists: Loudness normalization
- Spotify for Creators: Publishing audio episodes
- YouTube Help: Control video volume on your device
- YouTube Help: Video and audio quality enhancements
- YouTube Help: Stats for nerds
- Apple Podcasts for Creators: Audio requirements
- Amazon Music Help: Loudness normalization
- Amazon Alexa: Normalizing the loudness of audio content
- ITU-R BS.1770: Algorithms to measure audio programme loudness and true-peak audio level
Check your file in seconds
Measure loudness, true peak and noise floor for free in your browser, or master a file to the exact target with a report.