// analyze · tool 29

Spectrogram

See sound. Render the full time–frequency picture of any audio file — synced to playback, exportable as PNG — or watch your microphone paint the spectrum live.

ANALYZED ON YOUR DEVICE — NOTHING UPLOADED

Drop your audio file here

MP3, WAV, M4A, OGG, FLAC and more — or click to browse

[ drop works anywhere on this page ]

The audiophile's X-ray machine

A spectrogram unrolls audio into a picture: time runs left to right, frequency climbs bottom to top on a logarithmic axis (matching how you hear), and brightness shows energy through the scientifically-uniform viridis colormap across a 70 dB dynamic range. Once you learn to read one, you can see kick drums as bright columns, basslines as a thick glowing floor, vocal harmonics as stacked horizontal stripes, and cymbals as a fine mist across the top.

Things worth looking for

  • The lossy ceiling — MP3 and AAC cut content above ~16–20 kHz. A hard horizontal shelf near the top is the fingerprint of lossy encoding — and the classic way to expose a "lossless" file that was secretly transcoded from MP3.
  • Hum and hiss — a thin unbroken line at 50/60 Hz is mains hum; an even haze across all frequencies is broadband hiss. (Both are exactly what our Noise Reducer removes.)
  • Hidden images — some artists have drawn pictures into their spectrograms. Now you can check any track for easter eggs.
  • Your own voice, live — switch to microphone mode and whistle: you'll see the fundamental and its harmonics dance in real time.

Click anywhere on a rendered spectrogram to seek playback to that moment — the playhead tracks the audio as it plays. The PNG export saves the full-resolution image for sharing or documentation.

FAQ

What does a spectrogram show?

Time left→right, frequency bottom→top, energy as color. It's literally a picture of the sound — basslines, harmonics and hi-hats are all visible.

Can a spectrogram reveal fake high-resolution audio?

Often yes — lossy encoders leave a hard frequency ceiling near 16–20 kHz that survives transcoding to "lossless" formats. This tool makes the shelf obvious.

What FFT settings does this use?

2048-point Hann-windowed FFTs, log frequency axis from 30 Hz to Nyquist, 70 dB range, viridis colormap.

Is my audio uploaded?

No — file analysis and the live mic mode both run entirely in your browser.