LUFS Loudness Targets for Spotify, Apple Music & YouTube (2026)
Short answer: For most releases — AI-generated or not — master to about -14 LUFS integrated with true peaks under -1 dBTP. Streaming platforms measure your track's loudness and turn it toward their own house level (roughly -14 LUFS on Spotify, YouTube, Amazon, and Tidal; roughly -16 LUFS on Apple Music), so mastering much louder than -14 doesn't make you louder — it just gets turned back down while your crushed dynamics stay crushed. One master at -14 LUFS / -1 dBTP travels well across every major platform.
What loudness normalization actually does
Every big streaming service measures your track's integrated loudness (its average level across the whole song, in LUFS) and adjusts playback toward a house reference. Upload something louder than the target and it plays quieter than your file; upload something quieter and most platforms turn it up.
The key thing to internalize: normalization equalizes volume, not impact. A dense commercial master and a thin raw export played at the same meter level do not sound equally powerful — the master still sounds bigger because of its dynamics control, peak management, and tonal balance. That's why "just turn it up" never closes the gap with commercial releases. (More on that in why your Suno song sounds quiet on Spotify.)
Per-platform loudness targets (2026)
Here's the practical reference. Treat every number as a ballpark — platforms tune these over time and don't always publish exact figures.
| Platform | Approx. playback target | True-peak guidance | Notes |
|---|---|---|---|
| Spotify | ~ -14 LUFS | aim < -1 dBTP | Normalization on by default; a "Loud" listening mode exists but most listeners stay on default |
| Apple Music | ~ -16 LUFS (Sound Check) | aim < -1 dBTP | Plays slightly quieter than the -14 crowd; Sound Check is user-toggleable |
| YouTube / YT Music | ~ -14 LUFS | aim < -1 dBTP | Turns loud uploads down; lossy encoding makes brittle highs worse |
| Amazon Music | ~ -14 LUFS | aim < -1 dBTP | Broadly in line with the -14 group |
| Tidal | ~ -14 LUFS | aim < -1 dBTP | Normalization applied on playback |
| SoundCloud | ~ -14 LUFS (approx.) | aim < -1 dBTP | Historically lighter/looser normalization; treat as approximate |
These figures are approximate as of 2026-07-04 and change over time — verify current specifications on each platform's official documentation before relying on exact numbers.
Why -14 LUFS is the practical default
You can't master one file to perfectly satisfy every platform at once — they don't share a single number. But -14 LUFS integrated is the widely used compromise:
- On Spotify, YouTube, Amazon, and Tidal it lands right around the target.
- On Apple Music it plays a touch quieter — which is fine. Quieter and clean beats loud and crushed.
Mastering to, say, -8 LUFS doesn't buy you loudness on streaming. The platform turns it back down toward its target, and now you're stuck with whatever dynamic damage you did to reach that level — pumping, clipped transients, a squashed low end. You paid the price and got none of the reward.
If you want the deeper "why the number matters for AI tracks specifically," see the AI music loudness guide.
Don't forget true peak (dBTP)
Loudness is only half the spec. You also need a true-peak ceiling so the file doesn't clip when it's converted to lossy formats — encoding to MP3/AAC can push reconstructed peaks above 0 dBFS even when your sample peaks looked safe.
- Target -1 dBTP as a sane ceiling for most releases.
- Being conservative for heavy lossy encoding (like YouTube)? -1.5 dBTP gives extra margin.
- Use a true-peak limiter, not a plain sample-peak one. Sample-peak limiting lets inter-sample peaks slip through and clip after encoding (why that matters).
Integrated vs. short-term vs. momentary
Most meters show three LUFS readings, and it's easy to chase the wrong one:
- Integrated — the average across the whole song. This is the one platforms normalize to. Aim for ~-14.
- Short-term — a rolling 3-second window. Useful for spotting sections that jump out (a loud chorus, a quiet intro).
- Momentary — a 400 ms window. Useful for catching transient spikes.
For setting your release level, integrated is the number that counts. Don't try to make short-term readings hit -14 everywhere — that flattens the song.
A simple workflow to hit the target
- Master for quality first — EQ → compression → limiting, in that order. Loudness comes from this sequence, not from force.
- Set your limiter's output ceiling to -1 dBTP (true peak).
- Adjust gain into the limiter until the integrated reading sits around -14 LUFS over the whole track.
- Reference against a commercial track in your genre at matched loudness — level-match before you judge, or the louder one always "wins."
That's it. One master at -14 LUFS / -1 dBTP covers the field. For a fuller walkthrough on AI exports specifically, see how to master Suno songs and the Suno V5 mastering guide.
Common mistakes
- Mastering to -9 LUFS "to be safe." Streaming turns it right back down toward -14, and you keep the crushed dynamics. You lose twice.
- Chasing loudness with the limiter alone. Slamming 8 dB of gain reduction hits the number but pumps and crackles. Real loudness comes from the EQ and compression before the limiter.
- Ignoring true peak. A master that peaks at -0.1 dBFS can clip after lossy encoding. Leave headroom at -1 dBTP.
- Metering short-term instead of integrated. Platforms normalize to integrated loudness; if you flatten every 3-second window to -14 you kill the song's dynamics for no benefit.
- Re-encoding through MP3 twice. Master from the highest-quality export you have and convert to lossy once, at the very end — every extra lossy pass bakes in artifacts.
If you'd rather not meter by hand
Anti-AI Master runs the full EQ → compression → limiting sequence and lands on a streaming-ready loudness target automatically, exporting a 24-bit master in your browser in about ten seconds. Preview is free with no account — a handy baseline to A/B your manual masters against, so you can hear where your hand-metered version lands relative to a consistent reference.
FAQ
What LUFS should I master to for Spotify? Around -14 LUFS integrated with true peaks below -1 dBTP. That sits at full volume after Spotify's normalization without sounding crushed. Master much louder than about -11 and normalization simply turns you down while the squashed dynamics remain.
Is -14 LUFS the same for every platform? Roughly, for Spotify, YouTube, Amazon, and Tidal (all approximately -14 LUFS as of 2026-07-04). Apple Music plays a little quieter (around -16 LUFS via Sound Check). A single master at -14 LUFS / -1 dBTP works well across all of them — you don't need a separate file per service. Verify current numbers on each platform's official documentation.
Does mastering louder make my song louder on streaming? No. Platforms normalize playback toward their target, so a louder upload just gets turned down. You keep the dynamic damage without any loudness gain. The only "louder" that survives normalization is perceived density from a proper chain — not raw level.
What is dBTP and why -1 instead of 0? dBTP is true-peak level, which accounts for inter-sample peaks that appear when audio is reconstructed or encoded to lossy formats. A file that reads 0 dBFS on sample peaks can exceed 0 dBTP after encoding and clip. A -1 dBTP ceiling leaves margin so that doesn't happen.
Should I use integrated or short-term LUFS to set my level? Integrated. It's the whole-song average that platforms normalize to. Short-term and momentary readings are diagnostic tools for spotting loud or spiky sections — useful, but not the number you target for release level.
Platform loudness behavior changes over time; all figures above are approximate as of 2026-07-04 — verify current specifications on each platform's official documentation.