Loudness Targets and Mastering for Streaming Platforms
Streaming platforms normalize loudness. That single fact changes nearly everything about how a master behaves after it leaves the room, and it is still one of the most misunderstood aspects of preparing music for release.
This article covers what loudness normalization actually is, what the current targets are by platform, how loudness range fits into the picture, and how to think about all of it without letting the numbers make your decisions for you.
What Loudness Normalization Is
Before normalization existed, the only way to compete for attention on a playlist was to master louder than the track before yours. That dynamic produced some of the most fatiguing records in the history of recorded music. Streaming platforms introduced loudness normalization to solve the listener's problem: no more reaching for the volume knob between songs.
The mechanism is straightforward. Each platform measures the integrated loudness of your track after upload, compares it to their target, and applies a gain offset at playback. If your master is louder than the target, it gets turned down. If it is quieter, some platforms turn it up. The key word is playback. The gain adjustment happens in the stream. Your file is not modified. Think of it as the platform politely adjusting the volume at the party. Your music stays exactly as you made it. They just control the speakers.
Integrated loudness is measured in LUFS, Loudness Units Full Scale, a standard defined by the ITU-R BS.1770 specification. Unlike a peak meter, which shows the maximum amplitude of individual samples, a LUFS meter measures perceived loudness over time using a K-weighted frequency response that approximates how human hearing works. The result is a number that correlates meaningfully with how loud a track actually sounds, not just how loud its peaks are.
Current Normalization Targets by Platform
These numbers are current as of 2026 but are subject to change. Platforms update their specs without announcement. Verify directly with your distributor or check platform documentation before a major release.
| Platform | Integrated Target | True Peak Limit | Boosts Quiet Tracks | Normalization Mode | Lossy Format | Lossless Available |
|---|---|---|---|---|---|---|
| Amazon Music | -14 LUFS | -2 dBTP | No | Track only | MP3 320kbps | Yes — FLAC up to 24-bit/192kHz (HD / Ultra HD) |
| Apple Music | -16 LUFS | -1 dBTP | Yes | Track and album | AAC 256kbps | Yes — ALAC up to 24-bit/192kHz |
| Bandcamp | No normalization | -2 dBTP recommended | N/A | None | MP3 / AAC | Yes — FLAC and WAV available for download |
| Deezer | -15 LUFS | -1 dBTP | No | Track only | MP3 320kbps | Yes — FLAC 16-bit/44.1kHz (HiFi) |
| SoundCloud | ~-14 LUFS (unconfirmed) | -1 dBTP | Unconfirmed | Not officially documented | Ogg Vorbis / AAC | No |
| Spotify | -14 LUFS | -1 dBTP (quiet) / -2 dBTP (loud) | Yes | Track and album | Ogg Vorbis | Yes — FLAC up to 24-bit/44.1kHz (Premium, rolling out from Sept 2025) |
| Tidal | -14 LUFS | -1 dBTP (quiet) / -2 dBTP (loud) | No | Album (all playback) | AAC 320kbps | Yes — FLAC up to 24-bit/192kHz |
| YouTube Music | -14 LUFS | -1 dBTP | No | Track only | AAC | No |
A few things worth noting about this table.
The difference between -14 and -16 LUFS is not trivial. A master sitting at -14 LUFS integrated will be turned down 2 dB on Apple Music. That gain reduction is applied cleanly (no dynamic processing, just a level offset) but it means a master optimized for one target will behave differently on another.
YouTube, Amazon, and Tidal only turn loud tracks down. They do not boost quiet ones. A master at -18 LUFS will play at -18 LUFS on YouTube, not -14. This matters for quiet, dynamic records: they will sound quieter than everything around them on platforms that do not boost.
Spotify and Apple Music will boost quiet masters up to their targets. Apple Music does this without limiting, which means the dynamic character of a quiet master is preserved through the gain adjustment.
The true peak limits are not suggestions. They are the headroom the platform's lossy encoder needs to do its job cleanly. When a WAV file gets transcoded to Ogg Vorbis or AAC, the encoding process can push inter-sample peaks higher than they were in the source file. A master sitting at 0 dBTP before encoding can produce audible distortion after. The -1 dBTP ceiling exists to prevent that. For a deeper look at why, see True Peak vs Inter-Sample Peaks.
On lossless streaming: Apple Music, Tidal, Amazon, Deezer, and Bandcamp all offer lossless tiers. For Apple Music, Tidal, and Amazon, lossless playback means your master reaches the listener without lossy codec artifacts, but loudness normalization still applies. The gain offset happens before the stream reaches the listener regardless of format. Lossless does not exempt a track from normalization. What it does mean is that true peak management matters even more, because there is no lossy encode absorbing headroom issues before playback.
What Integrated LUFS Does Not Tell You
The integrated LUFS value is the number most people focus on, and it is the least complete part of the picture.
Integrated loudness measures average loudness over the entire track. Two masters can share the same integrated LUFS reading and sound completely different. One might be heavily limited, with transients compressed flat and dynamics squeezed into a narrow band. The other might be open and dynamic, with real peaks and genuine quiet. Both read -14 LUFS integrated. Only one sounds like music. The meter cannot tell you which one. Your ears can.
This is where loudness range becomes important.
Loudness range, or LRA, is a supplementary measurement defined in EBU R128. It describes the statistical spread of loudness within a track: the difference in LUFS between the quieter and louder sections, expressed in loudness units. A track with an LRA of 2 LU is almost uniformly loud throughout. A track with an LRA of 12 LU has significant dynamic variation, with loud passages that are meaningfully louder than quiet ones.
LRA does not have a target the way integrated LUFS does. There is no correct number. What it tells you is whether a master has been compressed into submission or left with room to breathe. For most music, an LRA somewhere between 6 and 14 LU reflects a master that has been processed intentionally rather than flattened. Heavily produced pop and hip-hop often runs lower. Classical and jazz often runs higher. The genre and the music determine what is appropriate.
The practical takeaway: if your master hits the integrated LUFS target but the LRA is 2 LU, you have not necessarily won. You have just squeezed the life out of the record to hit a number. Normalization will bring the level down to match everything else on the platform, and what is left is a dense, fatiguing block of audio with nowhere to go dynamically.
What Happens When Your Master Is Louder Than the Target
If your integrated LUFS reading is hotter than the platform target, the platform applies negative gain at playback. The track gets turned down. This is clean (no processing, no compression, just a gain offset) and the dynamic character of your master is preserved, just quieter.
The implication is one of the most important things to understand about mastering for streaming: there is no loudness advantage to pushing past the target. Every decibel of integrated loudness you gain above the platform ceiling by squashing dynamics is a decibel the platform removes at playback. You have traded dynamic range for nothing.
This does not mean loud masters are wrong. It means the motivation for them has changed. If a heavily limited master is what the music calls for (dense, punchy, aggressive) that is a valid creative choice and the master will still sound that way after normalization, just at a lower playback level. The problem is when heavy limiting is done in pursuit of a competitive loudness advantage that does not exist.
What Happens When Your Master Is Quieter Than the Target
The behavior here depends on the platform.
Spotify and Apple Music will boost a quiet master up to their target. Apple Music does this without limiting, preserving the dynamic character of the master through the gain adjustment. Spotify leaves 1 dB of headroom for the boost to avoid inter-sample peaks on the way up.
YouTube, Amazon, and Tidal will not boost. A quiet master plays at its actual level on these platforms. If you make a sparse, dynamic record at -18 LUFS, it will sound quieter than the records around it on YouTube. Whether that is a problem depends on the music and your intentions.
Bandcamp applies no normalization at all. Whatever you deliver is what the listener hears, at full level. It is also the one platform where listeners can download lossless files directly. That combination (no normalization, lossless download) makes it the most unmediated path between your master and the listener of any major platform.
How to Think About This When Mastering
The platforms set targets. The music sets the actual approach.
For most releases, the practical workflow is: master for the music, check the integrated LUFS and LRA at the end, and make sure true peak is managed correctly. If the integrated reading lands somewhere in the range of -9 to -16 LUFS, you are probably in reasonable territory depending on the genre. If true peak is at or below -1 dBTP, the encoder has what it needs.
For a loud pop or hip-hop record, -8 to -10 LUFS integrated is not unusual. The platform will turn it down to -14. It will still sound dense and energetic, because the limiting that produced that loudness is baked into the master. That is an intentional trade-off, not a mistake.
For a quiet, dynamic record (folk, classical, jazz, ambient) the integrated reading might land at -16 to -18 LUFS naturally. On platforms that boost, it will come up to target. On platforms that do not, it will play quieter than the records around it. Whether that matters depends on where the music lives and who is listening.
The one number that is non-negotiable regardless of genre or loudness approach is true peak. Keep it at or below -1 dBTP. This is not about the loudness target. It is about giving the encoder headroom to do its job without introducing distortion that was not in your master.
Beyond that, master for the music. Check the numbers when you are done. If they tell you something useful, act on them. If the master sounds right and the numbers are in reasonable range, trust what you hear.
Meters do not attend the listening party. People do.
