Quick Reference
| Parameter | Value | Notes |
|---|---|---|
| Integrated loudness | -14 LUFS (recommended) | EBU-gated per ITU-R BS.1770-4, recommended range -15 to -13 LUFS |
| Max true peak | -1.0 dBTP | Oversampled; headroom for OGG Vorbis / AAC encoding |
| Hard rejection | None | Spotify normalizes on playback instead of rejecting uploads |
| File format | WAV, FLAC, MP3, OGG accepted | Spotify transcodes every upload anyway |
| Playback: Loud | -11 LUFS | Positive gain uses a playback limiter |
| Playback: Normal | -14 LUFS | Default for most listeners |
| Playback: Quiet | -23 LUFS | No limiting needed, pure attenuation |
How Spotify Handles Your Master
Normalization instead of rejection
Unlike delivery platforms with automated QC, Spotify accepts any loudness and corrects it on playback. A master above -14 LUFS integrated is attenuated by the exact difference. The catch: the compression and limiting you used to get loud stays baked in, while the loudness advantage disappears. Next to a well-balanced -14 LUFS master, a squashed one plays at the same level and sounds flatter.
Quiet masters meet Spotify's limiter
When a master sits below the playback reference and the mode calls for positive gain, Spotify applies its own playback limiter to create headroom. That processing is outside your control. Delivering at -14 LUFS with -1 dBTP means neither direction of correction has work to do.
True peak headroom for lossy encoding
Every upload is transcoded to OGG Vorbis and AAC. Lossy codecs can raise inter-sample peaks by up to 3 dB, which is why the ceiling is specified as -1.0 dBTP oversampled true peak, not sample peak. Measure the master with a true-peak meter (BS.1770-4 Annex 2, 4x oversampling), because a sample-peak reading of -1.0 dBFS does not guarantee a pass.
Three playback modes, three references
Listeners choose between Loud (-11 LUFS), Normal (-14 LUFS, default) and Quiet (-23 LUFS). Your master cannot target all three, which is the point of mastering to the middle one: Normal plays it untouched, Quiet is pure attenuation, and Loud is the only mode that adds processing.
What This Means for Delivery QC
Because Spotify never rejects a file, every check on this page is advisory rather than pass/fail. DubCheck reports deviations from the -14 LUFS target and the -1 dBTP ceiling as warnings: the file will go live either way, but the numbers tell you exactly what Spotify's playback chain will do to it.
FAQ
Does Spotify reject masters that are too loud?
No. Spotify has no hard rejection mechanism for loudness. A master at -8 LUFS is accepted and then turned down by 6 dB on playback so it lands at the -14 LUFS reference. You lose nothing in delivery, but you gave up dynamic range in mastering that the normalization does not give back.
What happens to masters quieter than -14 LUFS?
In the default Normal mode Spotify can apply positive gain, and in Loud mode it applies a playback limiter to create the missing headroom. That limiter is not your limiter: it processes your master with settings you do not control. Delivering close to -14 LUFS integrated keeps your own processing in charge.
Why -1 dBTP when Spotify accepts anything up to 0 dBFS?
Spotify transcodes uploads to OGG Vorbis and AAC. Lossy encoding reconstructs the waveform and can raise inter-sample peaks by several dB. A master that sample-peaks at -0.1 dBFS can clip audibly after encoding. -1 dBTP measured as oversampled true peak keeps the encoded stream clean; for very loud, dense masters Spotify itself suggests -2 dBTP.
Is the -14 LUFS target measured with dialogue gating?
No. Spotify measures integrated loudness with standard EBU gating per ITU-R BS.1770-4, the same measurement DubCheck runs. Dialogue-gated measurements (like Netflix uses) will not match what Spotify computes for music or podcasts.
Can listeners turn normalization off?
Yes. Premium listeners can disable loudness normalization entirely, and then your file plays back exactly as mastered. This is another reason to keep the master itself clean rather than relying on playback processing: a meaningful share of playback happens un-normalized.