Best free online LUFS meters: 11 sites, the same files.

The results: same files, 11 sites

Tested on 24 October 2026. The first row is what our reference tools read. Master is the music file. EBU −23 is the EBU's real programme test file, whose answer is −23.0 LUFS; the error is in brackets. EBU +3.0 is the true-peak test file: every sample sits at 0 dBFS, the true peak between them is +3.0 dBTP, and EBU accepts +2.6 to +3.2.

What each site gives you for free, whether the file left the computer, and whether you can hear the master the way streaming services will play it:

Level means the site plays your original file turned up or down to the service's level. Codec means it also plays the lossy version the service streams. “Numbers only” and “target check” mean the site tells you how far services turn the track down but doesn't play it that way.

How we tested

• One real master. “The Boy Who Tried” by Astari Nite, a 2015 single released under CC0. It's loud and modern: −8.4 LUFS integrated, sample peak at 0.0 dBFS and true peak well above it.
• Four EBU test files from the EBU loudness test set, each with a published answer in EBU Tech 3341 or 3342: a 1 kHz tone at −23 LUFS, a real programme excerpt at −23 LUFS with an LRA of 5 LU, the true-peak test at +3.0 dBTP, and a two-tone LRA test at 10 LU.
• Reference values from three independent BS.1770 implementations: libebur128, FFmpeg's ebur128 filter and pyloudnorm. Two of them, libebur128 and FFmpeg, disagree on the master's true peak, so for that reading we also used iZotope RX and a band-limited reconstruction of the waveform at 16× and 64× oversampling.
• Each site got each file in a fresh browser tab, the way you'd drop it in. We read the numbers it displayed and kept a screenshot.
• Loudmeter changed its true peak measurement on the test day and now oversamples 16× at 44.1 and 48 kHz. Its row shows the readings after that change; its LUFS and LRA readings were the same before and after.
• For uploads, we logged every request each page sent with its size, and watched the computer's outgoing traffic while each site analysed the 42 MB master. A 5 MB test upload showed up clearly in both.
• Free and paid features come from each site's own page on the test date.

Sites change their tools and prices. These are the results on 24 October 2026.

What the numbers show

Integrated LUFS: no real differences. Every site that shows integrated loudness read the EBU files correctly, and all read the master at −8.4 LUFS except Peak Mastering at −8.5. Loudness Penalty shows only the turn-down per service, and its numbers fit the same values. If integrated loudness is all you need, any of them will do.

True peak: most pass, one doesn't. On the EBU true-peak file the sites read +2.8 to +3.1 dBTP, inside EBU's tolerance, except Jukeblocks at +1.94, about 1 dB low. Loudmeter read +3.0, within 0.01 dB of the answer. The master splits the field. Our reference tools read +2.2 (libebur128) and +2.8 (FFmpeg), and iZotope RX read +2.76: on a master clipped this hard, oversampling filters give different answers, and all of them put it well over 0 dBTP. Six other sites read +2.2 to +2.23, in line with libebur128. Loudmeter read +2.8, in line with FFmpeg and RX, and a band-limited reconstruction of the file also gave +2.8. Peak Mastering was the exception at +0.9 on the master, though it read the EBU file correctly. Youlean keeps true peak for its paid plan, and Loudness Penalty and Maztr don't show it. Why it matters: true peak vs sample peak.

Loudness range: check a real master, not only test tones. Every site that shows LRA passed both EBU LRA files. On the master, Loudmeter, Darwin's Cat, Venia and Sound and Go read 3.7 LU, in line with the reference tools at 3.6 to 3.8 LU. SoundBoost read 7.4 and Jukeblocks 7.2, about twice that. More on the metric in what is LRA.

Uploads: none. Every site says it measures in the browser, and the traffic agrees. While each site analysed the 42 MB master, the whole computer sent at most 1.8 MB, page analytics included. Privacy no longer separates these tools.

Hear it on streaming: why listening beats the number

A loudness penalty is one number: how far a service turns your track down. It doesn't tell you how the master sounds once it's down there, next to records that weren't limited as hard. The limiting that made it loud is still in the file after the turn-down, so the way to judge it is to listen at that level.

Three of the 11 sites let you do that. Loudness Penalty and Venia play your file at each service's level: their page code turns your original audio up or down, with no codec. Loudmeter does the same, and it also plays the lossy version each service streams.

What Loudmeter's preview does. It's the Hear it on streaming player under the readings in the online analyzer:

• It plays the file at each service's level: Spotify −14 LUFS, Apple Music −16, YouTube, Amazon Music and TIDAL −14, Deezer −15, and SoundCloud as uploaded. Spotify and Apple Music turn quiet tracks up only until the peaks reach −1 dBTP. YouTube, Amazon Music, TIDAL and Deezer only turn loud tracks down.
• You pick the quality too. Spotify: Ogg Vorbis 160 or 320 kbps, or lossless. Apple Music: AAC 256 kbps or lossless. YouTube: Opus 160, AAC 128 or AAC 256 kbps. Amazon Music: Opus 320 kbps or lossless. TIDAL: AAC 320 kbps or lossless. Deezer: MP3 128 or 320 kbps, or lossless. SoundCloud: AAC 160 or 256 kbps, or MP3 128.
• Lossy qualities are encoded on your device with that codec and bitrate, decoded back and lined up with your file sample for sample. Lossless qualities play your original samples.
• It shows where the peaks land after the codec.
• You can switch services and qualities while it plays. It carries on from the same moment at the new level, without a restart.
• The audio never leaves your browser.

What it showed on the test master. Spotify turned it down 5.6 dB, and after 160 kbps Vorbis its peaks sat at −2.9 dBTP. YouTube's 160 kbps Opus left them at −2.8, Amazon Music's 320 kbps Opus at −3.0, and Deezer's 128 kbps MP3, 6.6 dB down, at −5.3. Turned down, the master had room after every codec we tried. SoundCloud doesn't normalize, so its 128 kbps MP3 version still peaked at +1.3 dBTP.

How to compare with a reference.

1. 01Drop your master into the analyzer and pick the service your listeners use most.
2. 02Play the loudest section, then switch to Original. The difference you hear is what the turn-down does to it.
3. 03Drop in a released track you know well, from a file you own, and pick the same service. Both now play at that service's level, so you compare sound, not volume.
4. 04Listen for punch, low end and how present the vocal is. If your master sounds smaller at the same level, the extra loudness cost more than it gave.
5. 05Ease the limiter, bounce again and check.

Loudmeter analyzes one file at a time, so this is a back-and-forth. If you want both tracks loaded side by side, Venia's separate Streaming Preview page has a slot for a reference song. It changes the level only, without a codec.

What the preview can't tell you.

• It uses the service's codec and bitrate, not the service's own encoder and settings, so the lossy version is close, not identical.
• It plays the level normalization would use. A listener who has normalization switched off, or uses Spotify's Loud setting, hears something else.
• AAC qualities use your browser's own AAC encoder. A browser without one marks them Not available in this browser; our Linux test browser was one, so Apple Music fell back to lossless there.
• It's a check of your file before release, not a recording of the real stream.

Site by site

• Loudmeter. Every reading is free: integrated, short-term and momentary LUFS, true peak with each over listed by time, LRA, PSR, PLR and correlation, plus the codec preview above. True peak is oversampled 16× at 44.1 and 48 kHz. It read the EBU true-peak file at +3.0 dBTP, within 0.01 dB of the answer, and the master at +2.8, as FFmpeg did and close to iZotope RX's +2.76; libebur128 read +2.2. One file at a time.
• Youlean File Loudness Meter. From the makers of the Youlean Loudness Meter plug-in. Integrated, short-term max and momentary max are free and matched our references to 0.1 LU on every file. Its FAQ says the paid Plus plan unlocks LRA and true peak, so the free version can't check peaks.
• Loudness Penalty. By MeterPlugs, which also offers it as a plug-in and a desktop app. It shows how far YouTube, Spotify, TIDAL, Apple Music, Amazon Music, Pandora and Deezer turn your track down, and plays it at each level. It respects Spotify's peak limit when a quiet track is turned up. It doesn't show LUFS, true peak or LRA values.
• SoundBoost. A full set for free, with the time of the true peak and a loudness graph. True peak was exact on the EBU file and the EBU LRA files read correctly, but the master's LRA showed 7.4 LU against 3.6 to 3.8 from the reference tools.
• Jukeblocks. Integrated, LRA, true peak, short-term and momentary max, and the Spotify turn-down, with a free account for comparing a whole album. True peak read +1.94 on the +3.0 dBTP EBU file, outside EBU's tolerance, and the master's LRA read 7.2 LU.
• Peak Mastering. Shows sample peak next to true peak, which makes inter-sample overs easy to spot. No LRA. It read the EBU files correctly, but showed the master's true peak at +0.9 dBTP in repeated runs, where the other meters read +2.2 to +2.8.
• Darwin's Cat. One of the most complete: integrated, true peak, LRA, short-term and momentary max and PLR, all free, with a check against 17 targets from Spotify to EBU R128 and a downloadable report. Its integrated, true peak and LRA values were in line with our references on every file. It shows numbers only, with no playback at service level.
• Venia Mastering. Free with no sign-up, and its page says it takes up to 20 files at a time. Integrated, true peak, PLR, LRA, tonal balance, a plain-English summary and seven services including YouTube Music, with a level-only preview. Its values were in line with our references on every file.
• Maztr. Integrated loudness from the file, plus momentary and short-term while it plays. No true peak or LRA. Integrated was correct on every file; the page carries ads.
• Neural Analog. Integrated loudness, true peak and sample rate, nothing more, all within 0.1 of our references.
• Sound and Go. Integrated, LRA and true peak, all in line with our references, the turn-down for ten targets including podcasts and audiobooks, and several files at once. It can also export a normalized WAV or MP3, which none of the other pages we tested do.

Which one should you use?

• Just the integrated number: any of them. Youlean's free version and Maztr are the simplest.
• Every reading free, in line with our references on every file, master included: Loudmeter, Darwin's Cat, Venia or Sound and Go.
• To hear what normalization does: Loudmeter for level and codec, Loudness Penalty or Venia for level.
• A reference loaded next to your master: Venia's Streaming Preview page.
• Many files at once: Venia, Sound and Go, or Jukeblocks with a free account.
• A normalized copy of your file: Sound and Go.
• A meter while you mix: none of these, since online meters only measure files. Use a plug-in on the master bus, or LOUDMETER: Menu Bar on a Mac.

For what each kind of meter does well, see choosing a loudness meter. For how Loudmeter compares with plug-ins and desktop software, see Loudmeter vs other loudness meters. To check your own master, drop it into the online analyzer and press play under Hear it on streaming.

FAQ

Keep reading

Measure your file now.

Free, in your browser. Your file never leaves your device.