Audio file spectrum analyzer

Open a recording to see its whole spectrum, spectrogram, noise colour and file facts, all on your device.

No file open yet

Drop an audio file here to see what is in it

mp3, wav, flac, ogg or m4a. The file is read on this device and never uploaded.

0:00 / 0:00
Noise colour—
Peak level—
Highest real content—
File—

File facts
File
Open a file to see its facts.

Time range

Drag across the spectrogram, or type the start and end in seconds.

Showing the whole file.

Spectrum
Frequency scale

The frequency range is used by both charts. Leave “To” empty for the top of the file.

Spectrogram
Colours

Zoom with + and −, Ctrl + wheel or two fingers. Shift + wheel moves along the file.

Detail

Auto: about 6 Hz per bin at any sample rate.

Changing these analyses the file again. Which window?

Export and share

CSV lists the level at every frequency. The link keeps your settings, never your file.

Recent files
    Your file never leaves your device. Nothing is uploaded. Space play← → move+ − zoomEsc whole file

    How to use the audio file analyzer

    1. Press Open file, or drop a file onto the tool. No file to hand? Press Try a sample sound.
    2. Wait a moment while your browser reads it. A long file shows a progress bar, and Cancel stops it.
    3. Read the report: the spectrum on top, the spectrogram below it, and the noise colour, peak level and highest real content under the charts. File facts lists the rest.
    4. Drag across the spectrogram to pick a part, such as one word or one hit. The spectrum then shows only that part. Press Whole file to go back.
    5. Press play to hear the file while a cursor moves across the picture, then save a PNG of both charts or a CSV of the spectrum.

    How to read the report

    Peak: a short whistle Peak spectrum Average spectrum Slope −3 dB/octave: pink Highest real content 16 kHz → −20−40−60−801001k10kFrequency (Hz, log scale) · level in dB
    The spectrum part of the report for a noisy file with a short whistle and a lossy cut at 16 kHz.

    Average vs peak spectrum

    The amber line is the average: the whole file cut into short slices, each slice turned into a spectrum, and all of them averaged. It shows the overall tone, so a dull recording slopes down early and a bright one stays high. The coral line is the peak: the loudest level each frequency reached at any moment. A gap between them tells you that something was short. A whistle that lasts half a second barely lifts the average but makes a tall spike in the peak line.

    Highest real frequency

    This is where the sound in the file really stops. The tool looks for the point where the level falls by 20 dB or more within a sixth of an octave and stays down, which is how a lossy encoder or a low sample rate cuts the top off. A file whose content runs all the way up shows its top limit, half the sample rate, for example 24 kHz for a 48 kHz file. A cut at 16 kHz or lower often means the sound went through a 128 kbps mp3 at some point.

    Noise colour and slope

    The tool fits a straight line to the average spectrum from 50 Hz up to 16 kHz and gives its slope in dB per octave. Close to 0 means white noise, about −3 means pink and about −6 means brown. Music, speech and tones do not follow one straight line, so they read as “other”, with the slope still shown for the steady noise under them.

    Use cases

    Check a podcast for hum

    Open the episode and look at the low end of the average spectrum. Mains hum shows as a thin spike at 50 or 60 Hz with smaller spikes at 100, 150 or 120, 180 Hz. Drag across a quiet pause between sentences: if the spikes stay while the voice is gone, the hum is in the recording, often 20 to 40 dB above the floor. The live analyzer helps you find the cause while you move cables.

    Compare two mixes by eye

    Open the first mix, then open the second. In Recent files press Show as reference on the first one, and its average spectrum appears as a green dashed line over the new one. A mix that is 3 dB louder below 100 Hz, or 4 dB softer around 3 kHz, stands out at once. To lay two files over each other with more detail, compare two files with the comparison tool.

    Find out if a noise file is pink or white

    Sleep, masking and test noise files are often sold as pink but turn out to be white, or have a bump somewhere. Open the file and read the noise colour: a pink file should give about −3.0 dB per octave, and the dashed slope line should sit on the amber line from bass to treble. A reading such as −1.6 means it is neither, and the report says “other”.

    Check a sample library

    Before you buy or share samples, check a few files. File facts shows the sample rate, the peak level, so you can spot files that hit 0.0 dBFS and may clip, and the highest real content. A “96 kHz” sample with nothing above 20 kHz was made at a lower rate and converted up. To see if a lossless file hides an old mp3, check if a FLAC is real.

    Noise colours explained

    Noise colours describe how the energy of a steady noise is spread over frequency. Each one has a slope you can measure on the average spectrum, in dB per octave, where an octave is a doubling of frequency.

    ColourSlopeWhat it sounds like
    White0 dB/octaveBright hiss, like a detuned radio
    Pink−3 dB/octaveEven and full, like steady rain
    Brown−6 dB/octaveDeep rumble, like a distant waterfall

    White noise has the same energy in every hertz, so each octave holds twice as much energy as the one below it. Pink noise has the same energy in every octave, which sounds balanced to our ears, so speaker and room tests use it. Brown noise falls twice as fast and keeps most of its energy in the bass. For a live picture of noise changing over time, open the live spectrogram.

    Supported files and limits

    The tool opens what your browser can decode: wav, mp3, flac, ogg and m4a work in every current browser, and opus, aiff and webm in most. Files are read at their own sample rate where the browser allows it, from 8 kHz phone recordings to 192 kHz studio files. Wav files over 200 MB are read in parts, with a progress bar. Other formats over that size may be too big for your browser to decode in one go; save them as wav first. Copy-protected (DRM) files cannot be read by any web page, and the tool says so instead of showing an empty chart. Stereo files are measured channel by channel, so the L vs R view works on any stereo file; files with more channels, such as 5.1, use the first eight. Very short files, under one FFT frame, are padded, so the tool picks a smaller FFT size for them.

    FAQ

    Is my file uploaded?

    No. It is decoded and analysed by your browser on your device.

    What is the difference between average and peak spectrum?

    Average shows the overall tone of the file; peak shows the loudest moment at each frequency, which reveals short whistles or clicks.

    How do I know if noise is pink or white?

    Look at the slope: white noise is flat, pink falls about 3 dB per octave, brown about 6 dB. The tool gives the number.

    Why does my mp3 stop at 16 kHz?

    Many mp3 encoders cut the highest frequencies to save space. Check it with the Fake FLAC Checker.

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

    How we tested: Checked with the site's own test files: white, pink and brown noise read 0.0, −3.0 and −6.0 dB per octave, the fake FLAC test file shows its 16 kHz cut-off, a 1 kHz sine at −6 dBFS reads −6.0 dBFS, the silent file says it is silent, and the mono-as-stereo and out-of-phase files are spotted in the L vs R check.