Compare two audio spectra

Load your mix and a reference track, level-match them by loudness and see where their spectra differ.

Open your mix and a reference

A is your mix or recording, B is the track you want to compare it with. Both stay on your device.

Or drop two files here at once.

Loudness—
Biggest gap—
Close bands—
Smoothing1/6 oct

Files and sections
No file A yet (your mix)
No file B yet (the reference)
Band summary
BandA (dB)B (dB)A − B

Octave bands of the unsmoothed spectra. Rows of 3 dB or more are marked.

Level matching

With matching on, A's curve is moved by the loudness difference (ITU-R BS.1770 integrated LUFS of each section), so only tone differences remain.

Export and share
Recent results
    Both files stay on your device. Nothing is uploaded. A / B openS swapM match on/off

    How to compare two audio files

    1. Press Open A and choose your own mix or recording.
    2. Press Open B and choose the reference, for example a released track in the same style. You can also drop both files on the tool at once.
    3. Leave Match loudness on. Read the difference curve and the band summary under the charts.
    4. To compare just one part, such as the chorus, drag across each file's timeline under Files and sections, or type the From and To seconds.
    5. Press Swap to turn the comparison round, and save the curves with CSV or a picture with PNG.

    Files can be wav, mp3, flac, ogg or m4a, at any sample rate. If the two rates differ, the higher one is resampled to the lower one before anything is measured.

    How to read the difference curve

    A: your mix (level-matched) B: reference A − B −4 dB: less bass in A +3 dB around 3 kHz 201001k10kHz +30−3
    The two sample sounds: after level matching, A has about 4 dB less below 120 Hz and about 3 dB more around 3 kHz.

    The top chart shows both average spectra: A in amber and B in blue. The lower chart is the one to read. It shows A minus B at every frequency. Where the line sits above zero, A has more energy than B; where it dips below, A has less. The shaded strip covers ±3 dB, the range where differences are small enough to leave alone in most mixes.

    Each curve is the average of many short FFT frames across the chosen section, measured per channel and then averaged in power. That is why a stereo file and a mono file can be compared fairly, and why a stereo track whose two channels cancel when summed still shows its real level. The levels are relative: both files are measured the same way, so only the gap between the curves matters, not the numbers on the left axis.

    Smoothing decides how much detail you see. At 1/3 octave the curve shows the broad tonal balance, which is what you usually adjust with EQ. At 1/12 octave you see narrow peaks and dips, such as one resonant note in a bass line. The band summary adds the energy in each octave and puts the big differences into words, such as “A has 4.1 dB less 60–120 Hz than B”. Look for wide areas that are off by 3 dB or more. Narrow wiggles are normal, because two different songs never have the same notes.

    Use cases

    Mix vs a pro reference

    Load your mix as A and a released song in the same style as B, and compare the chorus of each. If the band summary says your mix has 4 dB less 60–120 Hz and 3 dB more 2–4 kHz, your low end is thin and the upper mids are pushy. Try a gentle low shelf and a small cut near 3 kHz, then compare again.

    Before vs after EQ

    Export the same part of a track before and after your EQ or a mastering chain. The difference curve shows exactly what the processing did: a 2 dB bump at 100 Hz, a cut of 1.5 dB at 400 Hz, a shelf that lifts everything above 8 kHz. It is a quick way to check that a plugin does what its display claims.

    Two microphones

    Record the same voice or guitar with two microphones at the same time and compare the two files. One mic may have 5 dB more presence around 5 kHz, the other a fuller sound around 200 Hz. Because both recordings hold the same performance, the curve shows the microphones, not the music.

    Two masters

    Compare two versions of a master, such as a streaming master and a louder club version. With level matching on, the loudness gap disappears and only real tone changes remain, for example the louder one losing some low bass to its limiter. Turn matching off to see the raw level difference too.

    Why level matching matters

    Our ears hear more bass and more treble when music is louder, so the louder of two tracks nearly always sounds better at first. On a chart, a louder file also sits higher everywhere, which hides the real differences. Matching by LUFS, the loudness measure that many streaming services use, moves A up or down so both have the same loudness. What is left on the difference curve is the tone balance alone. The readout shows how far apart the two files were before the match: −5.4 LU means A was 5.4 LU quieter than B, and its curve was raised by 5.4 dB. To measure loudness itself, use a LUFS meter.

    Limits

    The comparison shows the average over each section, so a short loud snare or a long quiet intro can tilt the result: pick similar parts of both songs. Different songs in different keys will always differ in narrow peaks, so trust the broad shape and your ears over single wiggles. Everything is measured from the files as they are; lossy formats such as mp3 often cut the top above about 16 kHz, which shows up as a steep drop at the right edge. To look at one file in more depth, analyze one file. Mixing tips are in our guide to the spectrum analyzer for mixing, and the LUFS meter measures loudness on its own. More tools are in the list of audio analyzer tools.

    FAQ

    Why level-match before comparing?

    Louder always looks and sounds “better”. Matching LUFS first shows real tone differences only.

    Should my mix match the reference exactly?

    No. Use the curve to spot big differences (3 dB or more over an octave), then trust your ears.

    Can I compare only the chorus?

    Yes, pick a section on each file's timeline.

    Are both files kept private?

    Yes, both stay on your device.

    Written and tested by the SpectrumAnalyzer.app team · Last checked

    How we tested: Checked with the site's test files: the −14 and −23 LUFS pink noise files read 9.0 LU apart and match within a fraction of a dB once level-matched, white against pink noise rises about 3 dB per octave, and the two sample sounds read about 4 dB less bass below 120 Hz and a bump of nearly 3 dB around 3 kHz.