An audio spectrum analyzer shows how much of a sound sits at each frequency, from deep bass on the left to high treble on the right. Press Start with the microphone, or open an audio file, and the graph updates live. The yellow dot marks the strongest frequency, shown below with its musical note. Switch to 1/3 octave for bands like a sound engineer's analyzer, or Spectrogram to see how the spectrum changes over time.
Reading the graph
- Across: frequency, on a logarithmic scale from 20 Hz to 20 kHz, so each octave takes the same width.
- Up: level in dBFS, where 0 dB is a full-scale sine. If you have calibrated your microphone in the decibel meter, levels switch to calibrated decibels.
- Green line: the current spectrum. Amber line: peak hold, the highest level reached at each frequency since you turned it on.
- Hover (or touch) anywhere to read the frequency, nearest note and level.
FFT size and resolution
The analyzer uses a fast Fourier transform (FFT) of the newest block of samples. A bigger block separates frequencies more finely but reacts more slowly.
| FFT size | Resolution at 48 kHz | at 44.1 kHz | Block length at 48 kHz |
|---|---|---|---|
| 2,048 | 23.44 Hz | 21.53 Hz | 43 ms |
| 4,096 | 11.72 Hz | 10.77 Hz | 85 ms |
| 8,192 | 5.86 Hz | 5.38 Hz | 171 ms |
| 16,384 | 2.93 Hz | 2.69 Hz | 341 ms |
| 32,768 | 1.46 Hz | 1.35 Hz | 683 ms |
At 2,048 points some low 1/3-octave bands (such as 31.5, 40 and 80 Hz) contain no FFT line and show empty; use 8,192 or more for bass.
1/3-octave bands
The 1/3-octave view adds up the energy in bands one third of an octave wide, ten bands per decade, centred on 1000 × 10k/10 Hz. Bands are labelled with the familiar nominal frequencies (25, 31.5, 40 … 16k, 20k Hz) from the preferred-frequency series in ISO 266, Acoustics: Preferred frequencies. Band levels use the same scale as our decibel meter, so with a calibrated microphone they read in decibels of sound pressure.
Weighting
Flat shows the signal as measured. A and C apply the hearing-related weightings from IEC 61672-1, the same as the decibel meter: A-weighting turns the bass down sharply, C-weighting only slightly. See what is a decibel for the curves.
What you can use it for
- Find a hum or whine: mains hum shows up at 50 or 60 Hz and its multiples; a whine shows as a sharp peak you can read off.
- Feedback: a ringing microphone shows one frequency climbing above the rest, so you know what to cut on an equaliser.
- Speakers and rooms: play pink noise and watch the 1/3-octave bands. Dips and peaks show what your speakers and room do to the sound.
- Music and voice: see the fundamental and its harmonics; whistle a note and read its frequency and pitch.
- Audio files: open a recording to see where its energy lies and whether it has content up to the top of the range.
How accurate is it?
In our tests with generated tones, the analyzer found the peak within 0.4 Hz of 1 kHz at every FFT size, and its 1/3-octave and total levels matched the calculated values to 0.01 dB. A-weighting and C-weighting matched the IEC values at 100 Hz. The peak level of a pure tone can read up to about 0.2 dB high because of the window. The bigger limits are outside the maths: phone and laptop microphones do not respond equally at all frequencies, often losing deep bass, and they add their own noise. Treat the shape of the curve as approximate unless you use a measurement microphone.
Export
Save PNG saves the graph as an image. Download CSV saves the current spectrum, one row per FFT line or per band, with your calibration applied, for Excel or Google Sheets. Nothing is uploaded; files are made in your browser.
To play test tones, use the tone generator; to measure loudness over time, use the decibel meter.
Common questions
What does a spectrum analyzer show?
How much of a sound is at each frequency. Bass is on the left, treble on the right, and the height shows the level.
How do I find the frequency of a sound?
Start the analyzer and make the sound. The yellow dot marks the strongest frequency, shown below in hertz with the nearest musical note. Use a larger FFT size for finer resolution.
What is 1/3-octave analysis?
Adding the spectrum into bands one third of an octave wide, ten per decade. It gives a steady, readable view of broadband sound such as pink noise or room noise.
Can I analyse an MP3 or WAV file?
Yes. Press Analyse an audio file. It plays the file and shows its spectrum as it goes, entirely in your browser.
Why does the bass look low?
Most phone and laptop microphones lose sensitivity in the deep bass, so the analyzer shows less bass than is really there. A measurement microphone gives a flatter result.
Last reviewed 14 November 2026 by Sri Murugan Systems. Found a mistake? Tell us.