Spend any time reading about streaming loudness and you’ll run into EBU R128, LUFS, integrated loudness, true peak, and LRA. These get used interchangeably in tutorials, but none of them mean the same thing. Mixing up the distinctions is how you end up with mixes that don’t translate.
This article covers what EBU R128 actually is, what the numbers mean, and how to use Phantom’s analyze_loudness tool to check where your mix stands before it goes to mastering.
What is EBU R128?
Back in the early days of digital streaming, listeners were constantly riding the volume knob. Different broadcasters and platforms ran at wildly different loudness levels, and there was no standard reason to change that. EBU R128 was the fix: a loudness measurement and normalization standard published by the European Broadcasting Union that gave everyone a common target.
Before EBU R128, the industry normalized to peak levels, making sure audio never exceeded a certain ceiling. Peak normalization doesn’t correlate well with perceived loudness, because our ears respond differently to sustained sound vs. transient spikes. A heavily limited track and a quiet acoustic recording can share the same peak level and sound completely different in volume.
EBU R128 fixes this by measuring perceived loudness with an algorithm called K-weighting, which applies filters that model how the human ear responds to different frequencies. The unit is LUFS: Loudness Units relative to Full Scale. (You’ll also see LKFS, which is the same measurement with a different name used in broadcasting.)
LUFS, dBFS, and RMS: what the numbers mean
dBFS (decibels relative to full scale) is a peak measurement. 0 dBFS is the maximum digital level; anything above clips. True peak measurement (dBTP) measures the reconstructed analog waveform after digital-to-analog conversion, which can run slightly higher than the digital peak because of intersample peaks.
RMS (root mean square) is an older average loudness measurement. It’s better than peak for correlating with perceived loudness. No K-weighting though, so it’s less reliable when you’re comparing across platforms.
LUFS is the EBU R128 loudness measurement. It comes in several flavors:
- Integrated LUFS: average loudness over the entire file (the number streaming platforms care about)
- Short-term LUFS: loudness averaged over 3 seconds (useful for section-by-section analysis)
- Momentary LUFS: loudness averaged over 400 milliseconds (useful for catching loud spikes)
- LRA (Loudness Range): the range between quiet and loud passages in LUFS units
For mastering-readiness checks and streaming compliance, integrated LUFS is the number that matters.
Streaming platform loudness targets
Each platform normalizes to a different target, and the behavior is asymmetric. Too loud and the platform pulls it down. Too quiet and most platforms leave it alone, no upward normalization, so your track just sounds softer than everything around it.
| Platform | Target (Integrated) | True Peak Max |
|---|---|---|
| Spotify | -14 LUFS | -1 dBTP |
| Apple Music | -16 LUFS | -1 dBTP |
| YouTube | -14 LUFS | -1 dBTP |
| Tidal | -14 LUFS | -1 dBTP |
| Amazon Music | -14 LUFS | -2 dBTP |
Say your mix comes in at -8 LUFS (heavily limited). Spotify knocks it down 6 dB to hit -14. Your mastering decisions about punch, attack, and compression tail are still there. The loudness advantage isn’t. Mixes that rely on loudness for impact tend to fare worse after normalization than mixes that used that headroom for dynamics.
The -1 dBTP true peak limit is there because intersample peaks can cause distortion during format conversion. Most platforms enforce this at upload and will reject or transcode files that exceed it.
LRA and dynamic range
LRA (Loudness Range) is the spread between quiet and loud passages in your track, measured in LUFS units. EBU R128 doesn’t set a universal cap; it leaves that to broadcasters based on genre and context. Many European broadcasters cap LRA around 15–20 LU for general programming, but music gets more latitude.
- Electronic and heavily produced tracks often have LRA values of 4–8 LU
- Rock and pop typically land in 6–12 LU
- Classical and jazz can reach 15–20 LU or higher
An LRA under 4 LU usually means heavy compression and not much left in the way of dynamics. A very high LRA can point to an uneven mix where some passages are dramatically quieter than others.
How to check loudness with Phantom
Phantom’s analyze_loudness tool runs a full EBU R128 analysis and returns:
- Integrated LUFS (the primary streaming compliance number)
- True peak in dBTP
- LRA (loudness range)
- Short-term and momentary loudness range
Ask Claude to run it on your bounce and compare against streaming targets:
“Analyze the loudness of my mix bounce and tell me if it meets Spotify’s -14 LUFS target. Flag any true peak issues.”
Phantom returns the measurements with context. It flags when true peak is approaching or exceeding -1 dBTP, and it tells you how far off you are from target rather than just handing you a raw number.
For a full pre-mastering check, full_diagnostic runs loudness analysis alongside phase, spectrum, and problem detection. You get a prioritized list of everything that needs attention before the session.
Practical workflow
- Bounce a reference mix at your normal export settings (24-bit, 44.1 or 48kHz)
- Run analyze_loudness via Phantom and check integrated LUFS against your target platform
- Check true peak: should be -1 dBTP or lower
- Review LRA: is the dynamic range appropriate for the genre?
- Adjust if needed: if you’re at -6 LUFS targeting Spotify, consider pulling back the limiting rather than pushing it further
If you’re heading toward mastering, hand off a mix with headroom. Not a pre-limited bounce. Mastering engineers typically want integrated loudness in the -16 to -20 LUFS range (sometimes lower) with plenty of true peak headroom so they can make their own loudness calls.
For more on the full mix analysis workflow, see the audio analysis hub page. It covers the complete set of Phantom tools and how they work together at each stage.