What LUFS should you master to?
Your song is ready, but how loud should the final version be? You may have seen different LUFS numbers for Spotify, Apple Music and YouTube. Do you need a different version for each?
Start with a reference, then judge the music
Usually, one master is enough for all three. Around −14 LUFS is a useful starting point, but every song does not need to hit that exact number. If making it louder leaves the drums sounding flat or the voice distorted, pushing harder will not help.
Here’s how to choose a suitable level, check your peaks and understand what happens when each platform plays your song.
| Platform | Playback reference | What to do with it |
|---|---|---|
| Spotify | −14 LUFS: Normal | Also Spotify’s published mastering recommendation. Check its true-peak advice below. |
| Apple Music | Approximately −16 LUFS: Sound Check comparison | Use as a playback comparison; it is not a universal stereo delivery requirement in Apple’s Digital Masters guide. |
| YouTube | Approximately −14 LUFS: comparison reference | Preview the level, then check the actual video. The general upload guidance does not mandate this LUFS value. |
The approximate Apple Music and YouTube figures are comparison references discussed in iZotope’s streaming-mastering guide. They are useful for a listening test, but should not be presented as guaranteed playback on every device or mandatory upload numbers.
This guide covers stereo music. A broadcast commission, an immersive release or a client delivery sheet can set different requirements. Follow that specification when one exists.
Spotify: understand −14 LUFS and peak headroom
Spotify’s mastering guidance recommends −14 LUFS integrated and true peaks below −1 dBTP. For masters louder than −14 LUFS, it advises true peaks below −2 dBTP. Normal playback uses −14 LUFS; Quiet uses −19 and Loud uses −11, with limiting available in Loud mode.
Louder tracks receive a downward gain adjustment. Quieter tracks can be raised only as far as the available headroom allows. Spotify also documents differences between album and shuffled playback, and exceptions for the web player and some third-party devices.
In your session, compare a conservative master with a denser version at equal listening loudness. If extra limiting makes the snare smaller or the vocal rougher, do not keep that processing just because its original meter reading is higher.
Apple Music: use Sound Check as a listening comparison
Apple’s Digital Masters guide describes Sound Check as a playback-level adjustment and asks for masters that preserve the intended sound without clipping. It does not prescribe −16 LUFS as a universal stereo-master delivery target. A number in a comparison table should not replace that distinction.
Listen to your preferred master at the lower comparison level. Is the vocal still present? Can you follow the bass notes? Does the chorus retain its lift? If something disappears, investigate balance and masking before adding more limiting. A guitar covering the vocal remains a balance problem when the whole mix becomes louder.
Apple’s audio mastering tools let you audition AAC encoding and inspect clipping. Use an encoded preview to hear whether a clean-sounding source develops roughness after conversion, especially around bright percussion and strong transients.
YouTube: measure the finished soundtrack
For a music video, start with the approved music master. If you add dialogue, effects or an extended introduction, measure the complete soundtrack again. The song’s original reading no longer describes the whole video. YouTube’s recommended upload settings cover encoding details without setting a mandatory integrated-LUFS target.
YouTube’s Stable volume feature continuously adjusts differences between quiet and loud portions. YouTube says it is unavailable for some videos and turned off for YouTube Music and official music videos. This is a separate behavior from applying one constant gain to an entire song. Check which playback features are active when reviewing your upload.
If the introduction is too loud beside the following dialogue, fix that transition in the edit. Turning down the complete soundtrack preserves the mismatch. Check the quietest important sentence and the busiest music passage, not just the final integrated number.
Read integrated LUFS and true peak together
| Measurement | Question it answers |
|---|---|
| Integrated LUFS | How loud is the complete track overall? Measure from start to finish. |
| Short-term LUFS | How loud is this passage? It uses a moving three-second window. |
| True peak · dBTP | How high can reconstructed peaks reach, including between samples? |
| Loudness range · LU | How much does sustained loudness vary across the programme? |
Youlean’s explanation of loudness measurements shows why a chorus reading and a full-song reading can differ. A chorus around −10 LUFS short-term can coexist with a complete-track result of −14 LUFS integrated. You do not need to force every passage to the integrated target.
Measure the complete exported song with LUFS Meter. Keep the start, ending and channel format identical when comparing versions. Repeatedly measuring only the loudest chorus answers a different question from measuring the finished release.
Example: what happens to a −9 LUFS master?
Take a worked example with integrated loudness of −9 LUFS and true peak of −2 dBTP. To compare it at −14 LUFS, apply −5 dB of constant gain: −14 − (−9) = −5. Its true peak moves down by the same amount, to approximately −7 dBTP.
A lower playback peak does not mean you should limit the master again. The gain reduction has lowered the whole signal. It has not restored drum attacks flattened earlier, nor removed distortion already printed into the file.
The distinction between gain adjustment and dynamic processing is explained in iZotope’s guide to audio normalization. A “loudness penalty” usually describes the estimated turn-down; it is not a quality grade or evidence of a recommendation-system penalty.
Example: why a quiet master cannot always be raised
Now take a master at −18 LUFS with a true peak of −3 dBTP. You want −14 LUFS while keeping a −1 dBTP ceiling. The loudness gap suggests +4 dB, but the peak headroom permits only +2 dB. Adding the full four would put the estimated peak at +1 dBTP.
You have three useful options: keep the dynamics and accept the lower loudness; control a few excessive peaks if that improves the sound; or revisit the mix if one hit is consuming disproportionate headroom. A gain-only adjustment cannot satisfy both targets in this example.
Listen to the strongest peak before treating it. A single accidental bump may need a local edit. A repeated kick may need a balance change. An intentional acoustic transient may be worth preserving. These are different decisions even when the peak meter shows the same maximum.
Compare two masters without rewarding extra volume
Save version A with lighter limiting and version B with more limiting. Suppose A measures −14 LUFS and B measures −9. Turn B down by about 5 dB after processing for the comparison. Leave both source exports intact. You are changing audition level, not rebuilding either master.
- Hold the listening setup fixed: same headphones or speakers, same playback passage and no extra enhancement between versions.
- Compare the start of a snare, the first consonant of a vocal phrase and the bass note underneath a strong kick. Identify a specific improvement or loss.
- Listen across a verse-to-chorus transition. Does the bigger arrangement actually feel bigger, or has the limiter reduced that contrast?
- Play the full song at a comfortable level. A version that impresses for ten seconds may become tiring over four minutes.
Neither the higher LUFS number nor the greater dynamic range wins automatically. Keep the version that serves the arrangement. Dense, sustained music and a sparse acoustic performance can need different treatment.
If the master still sounds wrong, change the right thing
| What you hear or measure | Try this next |
|---|---|
| Quiet overall, with low peaks too | Test a gain increase within your chosen ceiling; extra compression may be unnecessary. |
| Quiet overall, but a few high peaks | Locate and listen to those peaks. Decide between local edits, mix changes and controlled peak reduction. |
| The vocal disappears beside the reference | Match levels and compare similar sections. Check whether instruments mask the vocal before making the whole mix louder. |
| The chorus pumps or drums lose attack | Reduce limiter drive and compare again at matched loudness. Check whether bass hits trigger the unwanted reduction. |
| Distortion remains when you lower the output | Check an earlier mix or bypass suspect processing. Lowering final gain cannot restore already clipped audio. |
| Only one app or device sounds different | Check normalization, EQ and playback settings before changing the approved master. |
Integrated LUFS describes the whole programme. Two songs can match overall while their choruses differ considerably. Compare equivalent sections rather than a restrained intro against a full arrangement. Use your ears to identify the imbalance; use the meters to test the explanation.
Check the export you will actually release
- Export the approved version. Confirm its start, ending, channel format and file name before measuring.
- Reset the meter and analyze the complete track. Record integrated LUFS and true peak together.
- Listen to the busiest passage, the quietest important phrase and a strong transient. Confirm that the processing still supports the music.
- Compare playback at a reference level. If you change processing afterward, export and measure again.
- Follow the distributor or client’s actual delivery specification. Keep the approved export and a note of the measurement settings.
Check the final file with True Peak Meter after export. A limiter ceiling is a setting; the exported measurement verifies the result. Conversion or later processing can change peaks, so an earlier session reading is not the final check.
Use Streaming Loudness Checker to compare reference levels and available headroom. Its constant-gain estimates help you inspect the level change; they do not reproduce every player’s processing or judge the musical quality of the master.
Do you need a different master for every platform?
Usually, begin with one carefully checked stereo master that meets the delivery requirements. A different playback reference alone is not a strong reason to make another export. A copy that is simply 2 dB quieter retains the same underlying dynamics.
Separate versions make more sense when the actual deliverable changes: a video with dialogue, an instrumental, a club version with a different creative brief, or a commissioned format with specific limits. Name these versions clearly so the wrong export does not reach the distributor.
For an album, listen to the sequence as well as each track. A quiet interlude does not need the density of the main single. Preserve deliberate differences and check transitions. A set of identical LUFS readings does not guarantee a coherent album.