LUFS Meter

Peak level tells you the largest single sample in a file, which turns out to say almost nothing about how loud it sounds. A heavily compressed track and a dynamic one can both peak at exactly 0 dBFS while one is obviously far louder to listen to. LUFS measures perceived loudness instead: the signal is filtered to approximate how the ear weights different frequencies, then averaged over time with the quiet parts excluded. That is why every streaming service and broadcaster specifies LUFS and none of them specify peak.

No. It is decoded and measured entirely in your browser, and nothing is transmitted. There is no file size limit for the same reason. It also means unreleased music, client work and confidential recordings can be measured without leaving your machine.

Spotify, Amazon Music and YouTube normalise to about −14 LUFS. Apple Music and Apple Podcasts target −16, which is also the broad convention for spoken-word podcasts. European broadcast is EBU R128 at −23, and North American is ATSC A/85 at −24. True peak should sit at −1 dBTP for streaming and −2 for ATSC. If you are unsure, −14 for music and −16 for speech will not embarrass you anywhere.

Because the service turns it back down. A master delivered at −8 LUFS against a −14 target is simply attenuated by six decibels on playback. What does not come back is the dynamic range crushed to reach −8 in the first place, so the track plays at the same volume as everything else and sounds flatter than it needed to. Hitting the target is not a compromise; it is the entire point of the target existing.

Loudness range, or LRA, measures how much the level moves across a programme — the span between its quiet and loud passages, in loudness units. It is measured from short-term loudness rather than momentary, because the question is how much the level varies between passages, not between syllables. Roughly: a heavily compressed pop master might be 3 to 5 LU, a well-produced podcast 5 to 8, a film or a classical recording 15 or more. There is no correct value, only one appropriate to the material — but a very low LRA on speech usually means it has been over-compressed.

Sample peak is the largest value stored in the file. True peak is the largest value the waveform actually reaches between those samples, once a converter or a lossy encoder reconstructs the curve running through them. A file reading exactly 0.0 dBFS at every sample can still overshoot above full scale in between and distort on playback — which is why delivery specifications ask for −1 dBTP rather than 0. This meter estimates true peak by reconstructing between the samples rather than reporting the samples themselves.

The integrated measurement deliberately ignores the quiet parts. Two gates do it: anything below −70 LUFS absolute, then anything more than 10 LU below the average of what survived. Without that, a podcast with long pauses would measure far quieter than it sounds, because the silences would drag the average down. The meter reports what share of your file was excluded, so the figure is not a black box — a high number simply means a lot of silence, which is normal for speech and unusual for music.

The gap between your true peak and your integrated loudness, in decibels — a rough measure of how much dynamic range survives. A pure sine wave has about 3 dB. A crushed modern master might have 6 to 8. Something dynamic and unlimited will show 14 or more. It is not a quality score, but a very low figure on material that should breathe is a sign of heavy limiting.

The K-weighting filters are derived from the frequencies and Q values the standard specifies, rather than from its printed 48 kHz coefficients, so the measurement stays correct at 44.1 kHz and every other rate — using the printed values at 44.1 shifts both filters up by nine per cent and biases the reading. Checked against the reference tone the standard defines, this reads within 0.01 LU at both 48 and 44.1 kHz, comfortably inside the ±0.1 LU an EBU-compliant meter is allowed. Loudness range returns exactly 0.0 on a constant signal and exactly 10.0 on two halves ten units apart.

No, deliberately — a meter that changes the thing it measures is a worse meter. Our audio editor masters to any of these targets: it measures, applies the gain, limits the true peaks and then measures again, because limiting removes energy and so lowers the very loudness it was applied to protect. A single pass always lands short.

Yes. Drop in an MP4 or MOV and the soundtrack is measured, which is the usual case for anyone delivering to a broadcaster or checking a video before upload.