LUFS Meter Online

How the Loudness Meter works

Lufs meter online: what it does and why

Peak level tells you how close a file is to clipping; it says nothing about how loud it sounds. Two tracks can both peak at 0 dBFS and differ by 15 dB in perceived loudness, which is why streaming platforms stopped normalising by peak and started measuring loudness with the ITU-R BS.1770 algorithm instead. This tool runs the same measurement and shows you what the platform will see.

Integrated loudness is the single number the platforms use. The audio is passed through a K-weighting filter that mimics the ear's sensitivity, the power is averaged over 400 ms blocks, blocks quieter than an absolute gate of -70 LUFS and a relative gate 10 LU below the ungated average are discarded so silence and quiet intros do not drag the figure down, and the remainder is averaged into one value. Spotify, YouTube and Tidal aim for -14 LUFS; Apple Music for -16; podcast platforms commonly recommend -16 for stereo and -19 for mono; European broadcast (EBU R128) is fixed at -23 and American (ATSC A/85) at -24 LKFS, which is the same unit under a different name.

Loudness range (LRA) describes how much the loudness varies through the file, in loudness units. A heavily compressed pop master might have an LRA of 3–5 LU; a film soundtrack or a classical recording can exceed 20. It is not a target, but a very small value on a podcast suggests the compressor was working too hard, and a very large one means listeners in a car will keep reaching for the volume.

True peak is where clipping is really decided. Sample peak only looks at the stored samples, but the analogue waveform reconstructed between them can go higher — up to a few decibels higher on a limited master — and a lossy encoder can push it higher still. The tool upsamples the file four times before searching for the peak, which is how a true peak meter finds those inter-sample overs. Each target has a true peak ceiling, -1 dBTP for most services, -2 for Amazon and ATSC, and the reading is flagged when it is over.

The verdict. The tool subtracts your integrated loudness from the target's to get the gain the platform will apply. If the file is louder than the target it will simply be turned down, which is harmless apart from the wasted dynamic range. If it is quieter, most services turn it up — but only until the true peak would exceed the ceiling, so a quiet file with high peaks ends up quieter than everything around it. The verdict points this out: when adding the required gain would push the true peak over the limit, it tells you to limit first. The Gain to reach target line is the number to type into the Audio Volume Changer, or you can let the Audio Normalizer do the whole job.

Short-term and momentary maxima are the loudest 3 s and 400 ms windows in the file. Some platforms and most broadcasters also cap these — for example, EBU R128 recommends a maximum short-term of -18 LUFS for some programme types — and a high momentary maximum with a low integrated value shows where a single loud moment sits.

Export writes a CSV with every reading on the page followed by a short-term loudness trace sampled every 100 ms, which can be plotted to see how the level moves through a track or an episode.

Related tools