Loudness checker
Drop audio or a clip and the loudness checker returns its integrated loudness in LUFS, measured to ITU-R BS.1770 with both gates applied, plus the distance to what each platform normalises against. Everything is decoded and measured in your browser; nothing uploads.
Given an audio or video file, produce its integrated loudness in LUFS and the distance to each platform's target, in under twenty seconds.
Your file is decoded and measured inside this browser tab and released when you leave. Nothing uploads, nothing is stored, and there is nothing to delete.
Why loudness decides how your video sounds before anyone hears it
Every platform that plays your video normalises it, which means the loudness you exported is not the loudness anyone hears. Land above the target and the platform turns it down, taking your carefully placed peaks with it. Land below and it is either left quiet next to everything else in the feed, or turned up along with the room tone and the air conditioning.
The loudness checker measures the number that decision is made against. It is one integrated figure for the whole file, in LUFS, and the distance from the target — not the peak, not the average level on the meter in your editor, both of which can be right while the loudness is wrong.
How to check loudness by hand
- Use a meter that says LUFS. A meter showing dBFS or VU is measuring something else. Integrated LUFS is the number platforms use.
- Measure the whole file. Integrated loudness is defined over the entire programme. A reading taken over the loudest thirty seconds is a short-term reading and will be several LU high.
- Compare, then apply gain once. The difference between your reading and the target is the gain to apply. Applying it in stages, or with a compressor in between, means measuring again from scratch.
- Re-measure after every change. Music beds, ambience and a limiter all move the integrated figure, and they move it by more than people expect.
Where the manual way breaks
Checking loudness by hand breaks because one number does not tell you where the problem is. A file measuring 4 LU under target can be uniformly quiet, which one gain move fixes, or it can be a loud presenter over quiet guests, which gain makes worse in both directions at once.
The other failure is repetition. Loudness is a per-deliverable property, so eight clips from one recording are eight measurements and eight corrections, and the Editing Tax dataset puts colour, audio levels and music together at 20 minutes for a single recording before any clip is exported.
Never runs a static chain — amplify into a limiter, measured per deliverable — rather than a dynamic loudness pass that pumps, and an automated audio QA gate checks timeline drift, joints and ambience holes on the rendered result before anything is published. What loudness actually measures is the longer version of the number this page gives you.
Read the transcript
Loudness checker Drop audio or a clip and the loudness checker returns its integrated loudness in LUFS, measured to ITU-R BS.1770 with both gates applied, plus the distance to what each platform normalises against. Everything is decoded and measured in your browser; nothing uploads.
Frequently asked questions
What does LUFS measure that a peak meter does not?
Perceived loudness over time rather than the highest sample. A file can peak at 0 dBFS and still be quiet, because peak says nothing about how much of the runtime is loud. LUFS applies a frequency weighting that follows human hearing, then averages the energy with two gates that ignore silence.
What are the two gates?
An absolute gate discards any 400-millisecond block quieter than −70 LUFS, which removes room tone and pauses. A relative gate then discards blocks more than 10 LU below the mean of what survived. Without the second gate, a recording with long pauses measures several LU quieter than it sounds.
Is the peak figure a true peak?
No, and the loudness checker says so on the result. A true peak needs four-times oversampling to catch inter-sample peaks that appear after conversion; this reports sample peak, which is what it measured. Treat anything above −1 dBFS as close enough to clipping to fix.
Are the platform targets published?
Some are and some are not, and the table marks which. YouTube, Spotify, Apple Music, Apple Podcasts and EBU R128 publish theirs. The short-form feeds do not publish a number; the figure shown for them is the community-measured consensus and is labelled as measured rather than published.