≋
ULTIMATE METER
AUDIO REFERENCE LAB · V1© 2026 · Les disques Rubicon / Pure Obsessions & Red Nights · All rights reserved.
Analyze your track, compare it to your reference tracks, and hear it at your favorite streaming platform’s loudness level. No more surprises on release day!
Developed by musicians, for musicians.
Your music stays yours.
All analysis happens in your browser. No audio uploads. Your files never leave your device.
01 / IMPORT & ANALYZE
No files uploaded to a server+ Drop audio files hereWAV PCM / 32-bit float · MP3 · FLAC · AIFF*
or click to select multiple files
or click to select multiple files
Add masters and reference tracks.
Local library (0)
No analyses saved.
02 / KEY METRICS
No master selectedSpotify Normal: published −14 LUFS. Other service levels are approximate listening references. This is gain matching, not platform processing emulation.
—LUFS
BS.1770 · gated
—LU
EBU Tech 3342
—dBFS
sample peak
—dBTP
estimated 4× inter-sample
—dB
loudest 30% of blocks
—dB
vs. selected LUFS target
PLR : —P90 : —L/R : —Duration: —Sample rate : —
Import files once. Their measurements can then be compared instantly without reanalyzing.
* True Peak is an estimated inter-sample peak (4×), not a certified BS.1770 True Peak implementation. Very high-frequency signals may produce larger discrepancies than normal mastering content. LUFS/LRA are computed locally using K-weighting and BS.1770/EBU R128 gating rules. Loudness Penalty estimates the signed gain required to reach the chosen LUFS reference; actual platform normalization may differ, and positive gain can be limited by peak headroom. Results are intended for comparative analysis, not broadcast certification.
03 / A/B/C/D PLAYER · REFERENCE LISTENING
Click: seek · Drag: select a section
00:00 / 00:00No region selected
Select up to four A/B/C/D tracks and press ▶. Playback is normalized to −14 integrated LUFS.
PLAYBACK & ANALYSIS REGION
Select a region on the waveform or enter its boundaries. Each track keeps its own region.
Select at least 4 seconds on the waveform to recalculate measurements.
A/B/C/D playback at −14 integrated LUFS for the full track, without compression DSP. If the target would clip a quiet file, gain is capped at a conservative estimated −2 dBTP and a warning appears. A selected section may have a different loudness. Stored analyses contain no source audio; reconnect the original files after reopening the app. When browser cryptography is available, SHA-256 helps verify reconnections. Platform presets do not emulate the services’ codecs, limiters or other playback processing.
SELECTED REGION ANALYSIS
A/B comparison at −14 LUFS per sectionDraw a waveform selection, then click “Analyze A/B”. Region comparison uses A and B; C and D are also available for playback. If B has no region, the same time range is used.
04 / SPECTRAL TABLE · ZERO REFERENCE
Waiting for frequency-band analysis.
All tracks are matched to the same integrated loudness, −14 LUFS by default. + means more energy than the zero reference; − means less. Minimum and maximum in bold on each row. Frequency bands represent RMS energy (dBFS), summed in power rather than decibels. 16–20 kHz is included when the Nyquist limit allows.
05 / DIFFERENCE CURVES
dB vs zero referenceAdd at least two files to display the curves.
06 / LOUDNESS & DYNAMICS TABLE
Waiting for files.
P90: 90th percentile of short-term loudness (3 s). Local crest: median peak/RMS ratio of the loudest 30% of 100 ms windows. PLR = estimated True Peak minus integrated LUFS.
07 / STEREO ANALYSIS · VECTORSCOPE & MID/SIDE
—
−1 out of phase · +1 mono
—%
Side share of M² + S²
—dB
Side/Mid power ratio
—dB
Positive: left channel louder
VECTORSCOPE
Mono vertical · anti-phase horizontalPoint cloud built from real audio samples across the full track. This represents the overall stereo image, not a real-time oscilloscope.
WIDTH BY FREQUENCY BAND
Side energy share · %Select a track and reference to display both curves.
Waiting for stereo analysis.
Frequency-dependent correlation is calculated from the L and R spectra. Side share reflects (L−R)² / [(L+R)²+(L−R)²], not an absolute “width percentage.” Correlation near +1 in the bass is not a fault; negative values flag potential mono cancellation. Mono files are identified separately.
08 / SHORT-TERM TIMELINE
LUFS in 3 s windowsTimeline for the selected zero-reference track. Changes between verses and choruses remain visible, unlike in the integrated average.
09 / A/B VERSION COMPARISON
Choose two masters to compare loudness, dynamics, and spectrum.
Compare two analyzed masters. The local library restores saved measurements when possible; audio playback may require reconnecting the original files.