BeatSync
Sign InGet Started
Free · runs in your browser · nothing is uploaded

Free Loudness Benchmark

Is your master competitive? Drop it in and see how its loudness, punch, width and tone compare with the benchmark.

Drop your track here or click to choose
WAV, AIFF, FLAC, MP3, AAC
Not sure what you get? See a sample report.

Loudness per ITU-R BS.1770-4 / EBU R128 (K-weighting, gated), true peak by 4x oversampling. "Loudest 20%" = 3 s windows at or above the 80th percentile of short-term loudness. Spectrum bands are relative to total energy so they compare shape, not level. Hover a term for its definition.

How to use it

1
Drop your masterWAV, AIFF or FLAC. MP3 and AAC work too, but read hotter on true peak.
2
Analysed in your browserA few seconds. Only the measured numbers go to the benchmark, never the audio.
3
Read, fix, drop againWork through the findings from the top, re-export, compare.

The score

38
Competitive
score
Real gapMost measurements sit outside the benchmark spread.
62
Competitive
score
CloseIn range on some, off on the ones that weigh most.
90
Competitive
score
CompetitiveInside the spread on nearly everything.

Share of measurements inside the benchmark spread, weighted towards chorus loudness and density.

Reading a gauge

Loudness 1.8 dB behind
you -8.2 LUFSvs benchmark -1.8
Your choruses are 1.8 dB quieter than the benchmark. This is the gap listeners hear in a playlist.
Loudest 20%, LUFS
  • Highlighted band: benchmark ±1 standard deviation. Inside it reads "matching".
  • Centre line: the benchmark average.
  • Marker: your track. Red below the band, amber above, green inside.
LoudnessLUFS of the loudest 20%
PunchPSR. Lower = denser, more limited
WidthSide/mid in the loudest 20%
ToneSpectral tilt, bright vs dark
Low endSub 20-60 Hz share of energy

Findings

Critical
Costs level or translation. Fix first.
Warning
A trade-off that needs a decision.
Info
A difference that may be intentional.
Good
Confirmed in the benchmark range.

Sorted by severity. Every finding that has a fix states it.

Loudest 20% vs whole track

loudest 20%: what listeners compareintegrated: what platforms normalise onchoruschorusbreakdownintro

Platforms normalise on the whole track, listeners compare choruses, so the report measures both. Choruses matching while the whole track is behind means the gap is arrangement, not mastering.

After normalisation

-14 LUFS platform targetYour track · -6.6 LUFS-7.4 dBBenchmark average · -8.0 LUFS-6.0 dB

Every platform pulls both to its target. What is left is density and tone, not level. "Limited" means the platform would have to raise your track above its true-peak ceiling.

Fixes, in the order that works

  1. 1True peak under the ceiling: -1 dBTP for lossy platforms, -0.3 dBTP if you accept some overs.
  2. 2Mono the low end below 100-150 Hz, high-pass below 30-40 Hz. This is where headroom goes.
  3. 3Tonal balance per the band findings before pushing level.
  4. 4Density from clipping on the drums and saturation before the limiter. Limiter last.
  5. 5Automate section levels if the whole-track gap is arrangement.
Glossary
LUFS
Loudness Units relative to Full Scale. The standard loudness measurement (ITU-R BS.1770). Higher (closer to 0) is louder.
Integrated
Loudness of the whole track, gated so silence and very quiet parts do not count. This is what streaming platforms normalise on.
Short-term
Loudness over a sliding 3 second window.
Momentary
Loudness over a sliding 400 ms window.
Loudest 20%
The 3 second windows at or above the 80th percentile of short-term loudness: effectively the choruses and drops.
LRA
Loudness Range: the spread between the quiet and loud parts of the track, in LU. High LRA pulls integrated loudness down even when the chorus is loud.
True peak
The highest peak of the signal after it is reconstructed between samples (dBTP). Above 0 dBTP it clips DACs and lossy encoders.
PLR
Peak to Loudness Ratio: true peak minus integrated loudness, whole track. Lower means denser, more limited.
PSR
Peak to Short-term loudness Ratio: true peak minus loudness over 3 s windows. Tracks how hard the loud sections are limited. Lower = denser, flatter transients.
Crest factor
Sample peak divided by RMS level, in dB. Lower means flatter, more compressed.
Correlation
How similar left and right are. +1 = mono, 0 = fully independent, negative = phase cancellation risk in mono.
Side/mid
Energy in the stereo difference signal relative to the centre signal. Higher = wider.
Side/mid below 120 Hz
Stereo width in the bass. Wide bass eats headroom and cancels in mono.
Mono sum
Level change when left and right are summed to mono. Negative values mean phase cancellation.
Tilt
Slope of the long-term spectrum from 50 Hz to 16 kHz in dB per octave. More negative = darker.
Centroid
The centre of mass of the spectrum in Hz. Higher = brighter.
Benchmark
The benchmark average and spread. "In range" means within one standard deviation of the average.
FAQ

Is my audio uploaded?

No. Decoding and measurement happen in your browser. The benchmark comparison sends only the measured numbers (a few kilobytes), never the audio, and the benchmark statistics themselves never leave the server.

Which formats work?

WAV, AIFF, FLAC, MP3 and AAC/M4A, mono or stereo, at any sample rate. Use the lossless export you would send to a distributor: lossy encoding changes the true peak reading.

What LUFS should I master to?

There is no single number. Spotify normalises to -14 LUFS, Apple Music to -16, YouTube and Amazon to -14, and all of them turn loud tracks down. Competitive releases are mastered well above those targets because density and tone survive normalisation and level does not. The benchmark shows where competitive releases actually sit, for the chorus and for the whole track.

Why the loudest 20% and not just integrated loudness?

Integrated loudness averages the whole track, intros and breakdowns included. Two tracks with the same integrated loudness can have choruses several dB apart. The loudest 20% (3 second windows at or above the 80th percentile of short-term loudness) isolates the sections a listener compares when your chorus follows someone else's in a playlist.

What is the benchmark?

A set of commercially released, competitively mastered tracks measured with exactly the same analysis. Every metric is compared with the benchmark average and spread. "Matching" or "in range" means within one standard deviation of the average.

Do I have to stay under -1 dBTP?

Lossy encoders overshoot the source peaks on decode, so -1 dBTP is the common recommendation and -0.3 dBTP is fine if you accept some overs. The report flags overs as a warning, not an error: many competitive masters have them and treat it as a trade-off for level.

Can I compare two versions of a master?

Drop them one after the other. Each report replaces the previous one, so note the score, the gauges and the findings you care about before dropping the next file.