audioJS

Spectrogram

Drop a file. See its frequencies over time, with peak frequency and bandwidth.

Drop an audio or video file here
or click to choose, or paste with Ctrl+V

Nothing is uploaded. Everything runs in this tab with @audio/stft and @audio/spectral.

Spectrogram analysis in the browser

The page computes a short-time Fourier transform with @audio/stft and renders it on a canvas: logarithmic frequency axis by default (what you hear), linear on request (what engineers measure). Audio and video files are accepted; nothing is uploaded.

Reading it

Numbers alongside the picture

Strongest frequency across the file, the highest frequency with meaningful content (useful to judge a transcode), and the spectral centroid, the balance point of the energy that tracks perceived brightness.

Questions

What does a spectrogram show? Time runs left to right, frequency bottom to top, brightness is level. Harmonics appear as stacked lines, drums as vertical stripes, noise as a wash.

How do I spot a fake lossless file? A hard cutoff at 16 kHz is the MP3 128 kbps signature, 19 to 20 kHz suggests 320 kbps or AAC. True CD-quality material reaches 22 kHz with a natural roll-off.

Why choose a different window? A longer window (4096) resolves close pitches but blurs time; a shorter one (1024) shows fast transients but smears pitch. 2048 is the usual compromise at 44.1 or 48 kHz.

Is the file uploaded? No. The STFT is computed in your browser with @audio/stft; nothing leaves your device.