Loading tool…
How loud is this file, really? Drop in a song, podcast or video and get the readings broadcasters and streaming services use: integrated LUFS, short-term and momentary maximum, loudness range and true peak. A graph shows where the loud and quiet parts are, and a table shows what each platform would do to it.
Processing locally on your device
Overview
Loudness is measured in LUFS (Loudness Units relative to Full Scale) according to ITU-R BS.1770-4. The meter first passes the audio through K-weighting, a pair of filters that roll off the deep bass below about 38 Hz and lift the range above about 1.7 kHz by 4 dB, roughly matching how sensitive our ears are. It then averages the power of each channel (surround channels count 1.41 times, the LFE channel not at all) and converts it to LUFS. One LU equals one decibel, so a file at −10 LUFS sounds about 4 LU louder than one at −14.
Different windows answer different questions. Momentary loudness uses a 400 ms window and reacts to individual hits and words. Short-term loudness uses a 3-second window and follows phrases and sections; it is the highlighted line on the graph. Integrated loudness is the figure platforms normalize to: the whole program averaged in 400 ms blocks after two gates remove silence and quiet passages (an absolute gate at −70 LUFS and a relative gate 10 LU below the average). Loudness range (LRA, from EBU Tech 3342) is the spread between the 10th and 95th percentile of short-term values, so a heavily compressed pop song might read 3–6 LU while a film mix or a classical recording can exceed 15 LU.
Peak level matters too. Sample peak is simply the largest stored value. True peak oversamples the signal four times to find the peaks between samples, which appear after digital-to-analog conversion or lossy encoding. Delivery specs commonly ask for −1 dBTP or lower, and the meter highlights a true peak above that. The measurement code was verified against FFmpeg’s ebur128 filter on test tones and noise, agreeing on integrated loudness, short-term maximum and LRA to within 0.1 LU.
All analysis runs in JavaScript on audio your browser decodes locally; nothing is uploaded, and there’s no engine download. A three-minute song is measured in well under a second on a laptop. Very long files take more memory: an hour of stereo audio decodes to roughly 1.3 GB of samples at 48 kHz, so phones may struggle beyond about 30 minutes.
| Reading | Window | What it tells you |
|---|---|---|
| Momentary | 400 ms | Instant loudness of words and hits |
| Short-term | 3 s | Loudness of phrases and sections |
| Integrated | Whole file, gated | The number platforms normalize to |
| Loudness range (LRA) | Spread of short-term values | How dynamic the material is |
| True peak | 4× oversampled | Risk of clipping after conversion or encoding |
| Sample peak | Single samples | Highest stored value |
Drop an audio or video file onto the meter.
Read integrated loudness, short-term and momentary max, LRA, true peak and sample peak.
Hover over the graph to see the loudness at any moment, or play the file and click the waveform to jump around.
Choose a target to draw it on the graph and see the difference in LU.
Check the platform table, and copy the readings as text if you need them for notes or a QC log.
Use cases
See how far a master sits above the roughly −14 LUFS reference used by several streaming services, and therefore how many decibels of its loudness will simply be turned down on playback.
Confirm each episode lands near −16 LUFS with true peak under −1 dBTP before uploading, and use the graph to find the segment, often a guest on a quiet mic, that drags the average down.
Check commercials and programs against EBU R128 (−23 LUFS) or ATSC A/85 (−24 LKFS). The table shows the correction needed to comply, which you can apply with the normalizer.
Measure a few commercial tracks in your genre to understand typical integrated loudness and loudness range before mixing your own.
Pro tips
Answers to common questions about the loudness meter.
dBFS measures signal level relative to digital full scale, usually for peaks. LUFS measures perceived loudness over time, with frequency weighting for human hearing and gating to ignore silence. A file can peak at 0 dBFS and still be quiet in LUFS.
Integrated loudness averages the whole file after gating, while the short-term maximum is the loudest 3-second stretch. In most music the chorus or drop sits a few LU above the integrated value.
Streaming services turn loud tracks down to their reference level. Some, like Spotify and Apple Music, also turn quiet tracks up, but only as far as the peak headroom allows. Others, such as YouTube and Tidal, leave quiet tracks alone. The table applies those rules to your readings; the targets are commonly cited values and platforms change them from time to time.
True peak is an estimate that depends on the oversampling filter. This meter uses the 4× interpolator from BS.1770-4, and other meters may use different filters or higher oversampling, so differences of a few tenths of a dB are normal.
No. It analyzes files, including the soundtrack of video files. Load a recording to measure it.
No. ToolEnhancer processes files on your own device inside the browser. They are never sent to, stored on or seen by our servers.
Yes. The loudness meter is free to use with no sign-up, no watermarks and no daily limits.