Integrated loudness in LUFS (ITU-R BS.1770-4, EBU R128) in about 100 lines of dependency-free JavaScript. Runs in the browser (Web Audio) and in Node.
- Integrated loudness with absolute (-70 LUFS) and relative (-10 LU) gating
- Max momentary (400 ms) and max short-term (3 s) loudness
- Sample peak and an estimated true peak (4× oversampling)
- K-weighting coefficients computed for any sample rate (44.1, 48, 96 kHz …), not just 48 kHz
- Verified against
ffmpeg -af ebur128: within 0.03 LU on pink noise, tones and bursts at 44.1, 48 and 96 kHz (npm test)
Try it without code: Free online LUFS meter. Drop a WAV, MP3 or video file, nothing is uploaded.
import { measureFile, measureLoudness } from "./lufs.js";
// Browser: a File from <input type="file"> or drag and drop
const r = await measureFile(file);
console.log(r.integrated.toFixed(1), "LUFS", r.truePeak.toFixed(1), "dBTP");
// Anywhere: your own Float32Array channels
const result = measureLoudness([left, right], 48000);measureLoudness returns { integrated, momentaryMax, shortTermMax, samplePeak, truePeak }.
| Where | Integrated | True peak |
|---|---|---|
| Spotify, YouTube | -14 LUFS | -1 dBTP |
| Apple Music, Apple Podcasts | -16 LUFS | -1 dBTP |
| EBU R128 broadcast | -23 LUFS | -1 dBTP |
| Instagram Reels (measured, not official) | about -14 LUFS |
The Instagram value comes from our own measurement of 135 public Reels, TikToks and Shorts: Short-form video quality study 2026 (data as CSV, CC BY 4.0).
npm test
Needs ffmpeg on your PATH. The test generates signals with ffmpeg, measures them with both ffmpeg and lufs.js and fails if they differ by more than 0.1 LU.
- Only the first two channels are used (stereo). 5.1 channel weighting is not implemented.
- True peak is an estimate (Catmull-Rom 4× interpolation), close to but not identical with the BS.1770 polyphase filter.
MIT. Made by Diego Brügger, video editor in Zürich, for Shortform Kit: free tools for creators.