Loudness Meter

How to use it

1. Export the whole track, as it will be released: the finished master, not a loop or a section. Integrated loudness describes the entire file.
2. Drop it onto the meter or press the panel to choose it. WAV, AIFF, FLAC and MP3 work in every current browser.
3. Read the integrated loudness first, then the true peak. Point at the graph to find the loudest passages, and check the streaming list for what each service documents at that level.
4. Click any reading to copy it, or copy all of them as one line for a mastering note.

What the numbers mean

LUFS and LU use the same scale: LUFS is an absolute level (loudness units relative to full scale) and LU is a difference. A track at -9 LUFS is 5 LU louder than one at -14 LUFS.

What streaming services publish

Most loudness charts online give every service a number. Most of those numbers are measurements made by third parties, not figures the services publish. This meter only shows what each service documents itself, and links to the page that says so. Checked September 29, 2026.

• Spotify publishes the most: -14 LUFS by default, louder tracks turned down, softer tracks turned up at that level only while 1 dB of headroom remains below the true peak, Loud (-11) and Quiet (-19) options for Premium listeners, and no normalization in the web player or on third-party devices.
• TIDAL publishes a formula for developers in its open-source player SDK specification: ReplayGain referenced to -14 LUFS plus a 4 dB pre-amp on most devices, capped at the track's peak. It does not document what its own apps apply, and its listener help pages give no level, so the meter labels TIDAL's row as the SDK formula.
• SoundCloud says it normalizes on playback and advises mastering around -14 LUFS, but does not publish its playback level.
• Apple Music, Amazon Music and Deezer document a normalization setting that listeners can switch, but no level.
• YouTube documents no loudness normalization for music at all. Its one documented volume feature, Stable Volume, is turned off for YouTube Music and official music videos.

How loud should a master be?

There is no single right number. Where normalization is on, a louder master does not play louder: at its default setting Spotify turns a -8 LUFS master down by 6 dB, to the same level as a -14 LUFS master that kept more of its dynamics. Not every listener hears it that way. Spotify's Loud setting only turns masters down to -11, its web player and third-party devices do not normalize, and on the services with a switch the listener decides. What loudness buys you is a sound: the density, saturation and punch some genres expect.

A few points are firm. Keep true peak below -1 dBTP, and below -2 dBTP for very loud masters, so lossy encoding does not add distortion (Spotify and SoundCloud both say this). Broadcast work in Europe follows EBU R 128, which sets programme loudness at -23 LUFS. Anything else is a creative decision: compare your master with references you like, at matched loudness.

How accurate is it?

The meter implements ITU-R BS.1770-4: K-weighting, 400 ms gating blocks with 75 percent overlap, the -70 LUFS absolute gate and the -10 LU relative gate, with loudness range per EBU Tech 3342 and true peak by 4x polyphase oversampling. Its automated tests run the EBU Tech 3341 and 3342 style reference signals on every change:

• A stereo 1 kHz sine at -23 dBFS reads -23.0 LUFS integrated, short-term and momentary (tolerance 0.1 LU).
• The gating sequences (-36/-23/-36 dBFS and -72/-36/-23/-36/-72 dBFS) read -23.0 LUFS.
• Alternating tone patterns built to read a steady -23.0 LUFS short-term and momentary do.
• The four Tech 3342 loudness range sequences read 10, 5, 20 and 15 LU within 1 LU.
• Sines at fs/4, fs/6 and fs/8, sampled off their crests, read their true peak within +0.2 and -0.4 dB.
• Files shorter than 0.4 s are reported as too short for integrated loudness rather than silent, and loudness range and short-term maximum need at least 3 s.

The file is decoded at the sample rate its container declares (WAV, AIFF, FLAC, MP3, M4A and AAC, Ogg Vorbis), so nothing is resampled before measuring. Opus always decodes at 48 kHz, its native rate. AAC at a core rate of 24 kHz or less may carry SBR and is decoded at twice that rate, and a file that declares no rate is decoded at 48 kHz; the meter says which happened under the readings. Multichannel files up to 7.1 are weighted as BS.1770 specifies: surrounds count 1.41 and the LFE channel is left out. Tech 3341 case 6 (5.0, L and R at -28, C at -24, Ls and Rs at -30 dBFS) reads -23.0 LUFS.

Common mistakes

• Measuring a section. Integrated loudness is defined over the whole programme. Measuring only the drop reads louder than the song.
• Chasing -14 at any cost. Normalization already brings loud masters down. Turning a finished master down to -14 changes nothing a listener hears on Spotify; squashing a dynamic mix up to a target does.
• Watching only sample peaks. A master that peaks at -0.1 dBFS can have inter-sample peaks well above 0 dBTP. Use a true peak limiter and check the true peak reading.
• Judging an MP3 of the master. Lossy encoding changes peaks. Measure the WAV or FLAC you will deliver.
• Comparing at different levels. Louder sounds better for a few seconds. Level-match before deciding which version wins; the Referee mix check compares a mix with a reference on the same measurements.