LUFS Loudness Meter

About the LUFS Loudness Meter

Find out how loud a recording really is, the way streaming services and broadcasters measure it. Choose an audio file — MP3, WAV, M4A, FLAC, OGG, Opus, AIFF — or a video, and the page decodes it on your device and measures its integrated loudness (the figure streaming services and broadcasters work with), the loudness range (LRA), the loudest short-term (3 s) and momentary (400 ms) loudness, the true peak (4× over-sampled) and the sample peak, for the whole file and for each channel.

The report then compares the file with published targets: whether Spotify will turn it down or up and by how much, how it fits the Apple Podcasts recommendation, broadcast delivery to EBU R 128 (−23 LUFS) and ATSC A/85 (−24 LKFS), or a target of your own. A graph shows the loudness over time with the target line, so you can see which parts are loud and which are quiet. The measurement follows ITU-R BS.1770-5 and EBU Mode (EBU Tech 3341 and 3342), and nothing is uploaded.

Common uses

• Checking a master before uploading it to Spotify, Apple Music or YouTube.
• Making sure a podcast episode is near −16 LUFS before publishing.
• Checking an advert or a programme against EBU R 128 or ATSC A/85 delivery specs.
• Finding the loud and quiet parts of a recording on the loudness graph.
• Comparing the loudness of several takes or versions of a song.

What each measurement means

• Integrated loudness (LUFS): the average loudness of the whole file, measured with K-weighting (a filter that hears roughly like a person) and two gates that leave out silence and very quiet passages: blocks below −70 LUFS, then blocks more than 10 LU below the level of the rest (ITU-R BS.1770). 1 LU is 1 dB; LUFS and LKFS are the same unit.
• Short-term (3 s) and momentary (400 ms) loudness: the same measurement over a sliding window. The report gives their maxima, searched every 10 ms, and the graph shows both every 100 ms.
• Loudness range (LRA): how much the loudness changes in the file, from the 10th to the 95th percentile of its short-term loudness after gating (EBU Tech 3342). A few LU for a compressed pop song or an advert, 15 LU and more for a film. It needs at least 3 seconds, and EBU R 128 advises against it for programmes under a minute.
• True peak (dBTP): the highest point of the waveform between the samples, found by over-sampling 4 times with the filter of BS.1770-5 Annex 2. It is what a decoder or converter can produce, so it can be above 0 dB even when no sample is. Sample peak (dBFS) is the highest sample.
• PLR (peak to loudness ratio): true peak minus integrated loudness, a rough measure of how much the peaks stand out.

The targets, and where they come from

• Spotify: tracks play at −14 LUFS in the default Normal setting (Loud −11, Quiet −19). Louder tracks are turned down; quieter ones are turned up only as far as 1 dB of headroom allows — Spotify’s example is a −20 LUFS track peaking at −5 dBTP, which plays at −16 LUFS. Spotify advises masters with a true peak below −1 dBTP, or below −2 dBTP when they are louder than −14 LUFS (Spotify for Artists).
• Apple Podcasts: about −16 LKFS with a tolerance of ±1 dB, and true peaks no higher than −1 dB (Apple Podcasts audio requirements). Apple does not publish a figure for Apple Music’s Sound Check, so the report does not invent one.
• EBU R 128: programme loudness normalised to −23.0 LUFS, with ±1.0 LU only where the target cannot practically be met (live programmes) and ±0.2 LU for measurement differences in quality control; true peak at most −1 dBTP (EBU R 128).
• ATSC A/85 (US television): −24 LKFS, with about ±2 dB anticipated from measurement differences (content should still aim at −24), and true peaks at most −2 dBTP. A/85 measures long programmes on their dialogue; this page measures the whole mix, which is what A/85 uses for short content such as adverts (ATSC recommended practices).

How the measurement is checked

The meter follows EBU Mode: momentary loudness over a sliding 400 ms window, short-term over 3 s, integrated loudness over 400 ms blocks that overlap by 75 %, and the loudness range from short-term values every 100 ms. Its unit tests run the minimum-requirement signals of EBU Tech 3341 — tones at −23 and −33 dBFS, gated sequences, a 5.0-channel signal, short bursts for the momentary and short-term maxima, and the true-peak cases (for example a sine at a quarter of the sample rate whose samples read −9.0 dBFS while its true peak is −6.0 dBTP) — and the loudness-range signals of EBU Tech 3342, all within their published tolerances.

In surround files the centre counts like left and right, the surround channels of 5.1 and the side pair of 7.1 are weighted by 1.41 (+1.5 dB), the back pair of 7.1 by 1.00, and the LFE channel is left out, as BS.1770-5 says (7.1 is read in the WAV and FLAC channel order: front, centre, LFE, back, side).

Limitations

• This is a measurement, not a correction: to change the loudness use the Audio Normalizer.
• Dialogue-gated loudness (used for long-form programmes in ATSC A/85 and by some film and TV platforms) needs speech detection, which this meter does not do; it measures the whole mix.
• Files the browser cannot decode (some DRM-protected, very old or exotic codecs) cannot be measured. Long recordings take a while: the progress bar shows how far it has got.
• Platforms change their targets from time to time; the report names each one’s published figure with its source.
• The free preview shows whether the file meets each target and a watermarked graph, with the integrated loudness and every other figure hidden; the full report, its copy and its downloads come with a pass.

Privacy

On this device; its files download once

Everything happens in your browser. What you enter or open here is not uploaded or stored by MySmartCoPilot. The audio engine (and, for some files, a WebAssembly decoder) downloads from MySmartCoPilot the first time you open a file. Your audio and video are never uploaded.

Open-source components

This page can load these open-source components, each only when a task needs it:

• It uses FFmpeg under the LGPLv2.1 and the LAME MP3 encoder under the LGPL, through libav.js: licence and source code.
• It uses FFmpeg under the LGPLv2.1, through the Mediabunny add-ons: licence and source code.
• It uses the LAME MP3 encoder under the LGPL, through the Mediabunny MP3 add-on: licence and source code.
• It uses Mediabunny under the MPL 2.0: licence and source code.

You may replace them with your own builds. Their licences and the credits of every library the site uses are on the open-source credits page.