Loudness targets for Spotify, Apple Music, YouTube, TIDAL & more (2026)

You don’t master to a number. But you can’t ignore the number either: every major streaming platform measures your master’s loudness and turns it up or down to land on its own target. Master too loud and it gets flattened. Master too quiet and, on some platforms, it gets pushed up until the noise floor shows. Knowing each platform’s target, and how it behaves, is the difference between shipping the mix you made and shipping whatever the platform decides to make of it.

Here are the current targets for 2026, what each platform actually does with them, and how to check your master against the right one before you upload.

→ Check a master against a target in MixLab

The targets at a glance (2026)

Platform	Integrated target	True-peak ceiling	What it does
Spotify	−14 LUFS	−1 dBTP (−2 if your master is above −14)	Turns loud down and quiet up
Apple Music	−16 LUFS	−1 dBTP	Sound Check; turns down and up
YouTube / YT Music	−14 LUFS	−1 dBTP	Only turns loud down, never boosts
TIDAL	−14 LUFS	−1 dBTP	Album-normalized; only turns down
Amazon Music	−14 LUFS	−2 dBTP (strictest)	Track-normalized; only turns down
Deezer	−15 LUFS	−1 dBTP	Track-normalized
SoundCloud	historically none	keep −1 dBTP	Long had no normalization, verify current policy
Broadcast EU (EBU R128)	−23 LUFS ±1	−1 dBTP	Legally required for TV
Broadcast US (ATSC A/85)	−24 LKFS ±2	−2 dBTP	LKFS is the same measure as LUFS
Netflix (video delivery)	−27 LKFS (dialog-gated)	−2 dBTP	Delivery spec, not playback
Apple Podcasts	−16 LUFS (−19 mono)	−1 dBTP	Sound Check active
Spotify Podcasts	−14 LUFS	−1 dBTP	Normalized like music

New for LUFS itself? Start with the explainer: what a loudness meter measures and which numbers matter. This page is about the targets and what to do with them.

The one behavior that actually changes your decisions

Everyone quotes the LUFS numbers. Almost no one mentions the part that changes what you should do: some platforms only turn loud material down; others also turn quiet material up.

• Turn-down-only (YouTube, Amazon Music, TIDAL): a master quieter than target just plays quieter. No penalty, no reward, the platform leaves it alone. Going over target gets you turned down, so you gave up dynamics for nothing.
• Turn-down-and-up (Spotify, Apple Music): a quiet master gets boosted to target. That’s usually fine, but the boost also lifts your noise floor, hiss, and room tone. A −20 LUFS bedroom master pushed up to −14 can reveal problems that were inaudible before.

So the practical rule is not “hit −14 exactly.” It’s: land near the target, keep true-peak headroom, and never exceed the strictest ceiling that matters to you. For most releases that means somewhere between −14 and −16 LUFS integrated with peaks at or below −1 dBTP (−2 if Amazon Music matters to you).

Per-platform notes

Spotify, −14 LUFS, −1 dBTP. The default listening mode (“Normal”) normalizes to −14. There’s a “Loud” mode (−11) and a “Quiet” mode, but −14 is what most people hear. Spotify’s own guidance: if your master is louder than −14 LUFS, keep true peak at −2 dBTP so their downward limiting doesn’t clip. Spotify boosts quiet tracks up.

Apple Music, −16 LUFS. Sound Check targets −16, two units quieter than the −14 crowd. It’s on by default on most devices. Apple both turns down and boosts up.

YouTube, −14 LUFS. Applies to music and to video (including podcast videos). YouTube only turns loud uploads down, it never lifts quiet ones. You can read the applied gain in YouTube’s “Stats for nerds.” Aiming a hair under −14 is safe.

TIDAL, −14 LUFS, album-normalized. TIDAL normalizes by album, not by track, so a quiet interlude keeps its relationship to the loud single. Turn-down-only.

Amazon Music, −14 LUFS, −2 dBTP. Same loudness target as most, but the strictest true-peak requirement, −2 dBTP. If Amazon is a priority, master to −2 dBTP and you’re covered everywhere else too.

Deezer, −15 LUFS. One unit quieter than the −14 group, track-normalized.

SoundCloud, historically no normalization. For years SoundCloud played your file at whatever level you uploaded. That means loudness-war masters can sound “louder” there, the one platform where over-compressing still “works,” at the cost of dynamics everywhere else. Policies change, so verify the current behavior before you master specifically for it.

Broadcast (EBU R128 / ATSC A/85). TV is a different world: −23 LUFS (EU) or −24 LKFS (US), and it’s a requirement, not a suggestion, with a −1 to −2 dBTP true-peak ceiling. If you’re delivering to broadcast, the tolerance is tight (±1 LU EU) and a certified meter is often contractually required.

Podcasts. Apple Podcasts targets −16 LUFS (−19 for mono); Spotify normalizes podcasts to −14 like everything else. Spoken-word content lives or dies on consistency, every episode at the same integrated LUFS matters more than the exact number.

The number that can actually hurt you: true peak

Miss the loudness target and you lose a little dynamics or a little level. Miss the true-peak ceiling and you get audible distortion.

A sample-peak meter reading −0.1 dBFS can still have inter-sample peaks over 0 dBTP, the real analog waveform between samples goes higher than any single sample, and lossy codecs (AAC, Ogg, MP3) push it higher still on decode. That’s why every platform above specifies a dBTP ceiling, not dBFS. Set peaks to −1 dBTP (−2 for Amazon or broadcast) measured with true-peak oversampling, and you have real headroom. (Why sample peak lets you down →)

Master for the music, then check against the target

The workflow that keeps you in control:

1. Master for the song, not for a LUFS reading. Get the balance, dynamics, and tone right.
2. Measure the finished master, integrated LUFS, LRA, and true peak.
3. Check it against the target for where it’s going. Over target? You’ll be turned down, decide if the loudness was worth the dynamics. Way under on a boost-up platform? Check your noise floor.
4. Fix the true peak first. That’s the one that distorts.

You don’t need a plug-in for step 2 and 3. Drop the file into MixLab and it reports integrated LUFS, LRA, and true-peak dBTP in the browser, nothing uploads, nothing leaves your device. Point it at a platform target and it tells you how far off you are and in which direction.

Shipping a catalog or an automated pipeline? The same engine is available as an API and MCP server: POST a file (or hand your AI agent the check_target tool) and get back “−13.2 LUFS, −0.8 dBTP, 1.2 LU over the Spotify target” as JSON, no DAW in the loop.

An honest caveat on the numbers

These are the platforms’ published and observed normalization targets as of mid-2026. They move: Spotify has changed its reference in the past, services add and remove modes, and normalization can be switched off by the listener. Treat the table as a well-informed starting point, verify anything mission-critical against the platform’s current spec, and, because the listener can always disable normalization, make sure your master sounds right at its own level too, not only after the platform touches it.

FAQ

What LUFS should I master to for Spotify in 2026? −14 LUFS integrated, true peak at −1 dBTP (−2 dBTP if your master runs louder than −14). Quieter is fine; louder gets turned down.

Is −14 LUFS a hard rule for all streaming? No. −14 is the common cluster (Spotify, YouTube, TIDAL, Amazon), but Apple Music is −16 and Deezer −15. Master once, sensibly, near −14 with −1 dBTP peaks and you’re in range for all of them, the small differences are handled by each platform’s own normalization.

If I master quieter than the target, do I lose loudness? On turn-down-only platforms (YouTube, Amazon, TIDAL) a quiet master simply plays quieter, no boost. On Spotify and Apple Music it gets pushed up to target, which can also lift noise. Either way, mastering louder than target buys you nothing but lost dynamics.

Which true-peak ceiling covers everything? −2 dBTP. It satisfies Amazon Music and broadcast, and it’s comfortably inside the −1 dBTP other services ask for.

Does normalization mean the loudness war is over? For streaming, effectively yes. Since every service pulls you back to its target, a hyper-compressed master doesn’t play louder than a dynamic one, it just plays flatter. The exception is a platform that doesn’t normalize (historically SoundCloud) or a listener who turns normalization off.

Further reading

• Online EBU R128 loudness meter: what it measures: the explainer for LUFS, LRA, and dBTP.
• MixLab: the working analyzer + target check + stem compare.
• API & MCP: the same measurement from code or an AI agent.
• Methodology: loudness: how we measure and validate against the reference.