Spectrogram Viewer
See frequency over time. A hard cutoff near 16 kHz often reveals an MP3 that was converted to FLAC.
MP3, WAV, M4A, FLAC, OGG, Opus, AIFF, WMA, AMR and the sound of most videos. Files stay on your device.
How it works
- Choose an audio file.
- Pick the scale and colour map, and zoom into a time range.
- Export the view as a PNG if you want to keep it.
How to read a spectrogram
Time runs from left to right and frequency from bottom to top. The colour shows how loud each frequency is at each moment: brighter means louder. A steady tone is a horizontal line, a drum hit is a vertical stripe, and a voice is a stack of lines (the harmonics) that move up and down with the pitch.
The cursor reads out the time, the frequency and the level in dB under the pointer. You can move it with the arrow keys instead of the mouse, and the picture has a text description that names the brightest region.
Settings that change the picture
The FFT size trades time detail against frequency detail: 512 shows fast events sharply, 16384 separates close pitches. The window changes how much a strong tone smears into its neighbours (Hann is a good default). The frequency scale can be logarithmic, which matches how pitch is heard, linear, which shows the top of the range clearly, or mel. The loudest and quietest levels set the colour range; narrowing it brings out faint detail.
Colour maps are magma, viridis and grayscale. They are ordered by brightness so that quieter always looks darker, which makes them safe for people with colour vision differences and for printing.
Zooming and exporting
The whole file is drawn first. Zoom in time with the buttons, the keyboard, or Ctrl or Cmd with the mouse wheel; each view is recomputed from the audio at full detail instead of stretching a small picture. The lowest and highest frequency fields zoom the other axis. Export saves the current view as a PNG, with or without axes and a level scale.
To check for a fake lossless file, look for a hard horizontal cutoff: many MP3 encoders remove everything above 16 kHz or so, and the cutoff stays if the file is later saved as FLAC or WAV.
Questions
What FFT size should I use?
2048 is a good start. Use 512 or 1024 to see exactly when sounds start, and 8192 or 16384 to separate notes that are close together.
Why is there a hard cutoff at 16 kHz?
Lossy encoders such as MP3 remove high frequencies to save space. A cutoff at 16 kHz or 19 kHz in a file that claims to be lossless is a common sign that it was made from an MP3.
Can I see more than 20 kHz?
Only if the sample rate allows it: the top of the range is half the sample rate, so a 96 kHz file shows up to 48 kHz.
How long files can it handle?
Long files work: the audio is analysed in pieces and only the visible part is kept as a picture, so memory use stays small. Very long files take longer to draw because every part has to be read.