Guide · Analysis
How to Read a Spectrogram
A spectrum analyser shows you one moment. A spectrogram shows you the whole piece: every transient, every sustained note, every frequency, all at once. It is the most information dense picture in audio, and it takes a little practice to read.
What It Is
A spectrogram is a stack of spectra. The analyser takes a window of audio, measures the frequencies in it, and plots that as a column of coloured pixels. It then does the same for the next window, and the next, and slides the whole thing past you.
So the two axes are not the same as a spectrum analyser's. On a spectrum analyser, the horizontal axis is frequency. On a spectrogram, the horizontal axis is time and the vertical axis is frequency.
The colour at any point is the level of that frequency at that moment. Bright means loud. Dark means quiet or absent.
The Axes
The vertical axis runs from 0 Hz at the bottom to 20 kHz or more at the top. It is normally logarithmic, so that the low end, where most of the interesting decisions are made, is not crushed into the bottom tenth of the screen.
The horizontal axis is time, and the scale is whatever suits the material: seconds for a whole track, bars for a loop, or anything in between.
The colour scale is a legend you need to read. Most spectrograms use a dark-to-bright ramp where each step is a set number of decibels, but the specific colours differ between tools. What matters is finding the darkest and brightest ends of the legend and working out the range between them.
What to Look For
Vertical stripes
Each stripe is a transient: an event where a lot of the spectrum appears at once. A kick drum produces a tall vertical line because it has energy across the whole low end. A hi-hat produces a shorter one, higher up.
When the stripes are evenly spaced, you are looking at a rhythm. The gaps between them are the space between hits, and that space is where you can hear whether a part has room.
Horizontal lines
A horizontal line means a sustained pitch. A bass note held for a second draws a line at 55 Hz, and its harmonics draw parallel lines above it at 110, 165 and 220 Hz.
This is the single most useful thing about spectrograms. A spectrum analyser tells you there is energy at 55 Hz. A spectrogram tells you it is a note, that it lasts 400 ms, and that it repeats every bar. You can read the actual performance.
The bright floor
A continuous band of colour across the bottom of the display means there is continuous energy in that range. Sometimes that is the music, and sometimes it is noise floor, hum, or bleed. If a track is supposed to be clean and there is a bright line at 50 Hz running the whole time, that is hum and you have a problem.
Harmonic stacks
A line at 100 Hz with parallel lines at 200, 300, 400 and 500 is a harmonic series: a sawtooth, a square wave, or a piano note. The spacing between the lines tells you the fundamental, and the relative strength of the upper harmonics tells you the timbre. A bright 3rd harmonic is a hard, nasal sound. A weak one is soft.
Curved lines
A line that drifts upward in frequency is a glide, a bend, or a pitch drift in the recording. A line that curves downward after a hit is the natural pitch drop of a percussive body. A sudden vertical smear at the start of a note is a click.
What It Is Good For
- Finding clicks and pops. They show as thin vertical lines extending far above the surrounding material, and they are much easier to spot here than in a waveform.
- Checking whether a part is playing what you think. You can read the notes.
- Seeing bleed. If a microphone picked up a drum that was not meant to be there, it is a faint version of that drum's pattern, at a lower level.
- Finding sibilance. A dense bright band between 5 and 9 kHz across a vocal is sibilance, and you can see exactly which syllables.
- Timing. Whether two parts actually line up, or whether one is early.
What It Is Not Good For
It is not good for making precise level judgements. A spectrogram shows level as colour, which is hard to read quantitatively. For an exact number, use the readouts on the analyser.
It is also not good for judging a single momentary problem, because the window length sets how much time is averaged into each column. A long window is precise in frequency but smears transients; a short window is the opposite.
And it is not a substitute for listening. A spectrogram will show you that two parts overlap at 120 Hz. It will not tell you whether that is a problem.
Spectrum Against Spectrogram
The two are the same measurement at different scales, and it is worth knowing which one suits the question.
| Question | Use |
|---|---|
| What is the peak level of this file? | Neither. Use the numeric readouts. |
| What is the tonal balance right now? | Spectrum analyser. |
| Where are the clicks? | Spectrogram. |
| What notes are being played? | Spectrogram. |
| Is the low end clean throughout? | Spectrogram. |
| Is the bass and the kick masking? | Both, but the masking finder is faster, because it compares the two directly. |
Questions
What is the difference between a spectrum and a spectrogram?
A spectrum analyser shows one moment in time. A spectrogram shows a whole piece of time: the horizontal axis is time, the vertical axis is frequency, and the colour is the level at each point. It is the same measurement drawn as a film rather than a photograph.
What do the colours mean?
They mean level. The scale runs from dark to bright, and each step is a set number of decibels. Exact readings depend on the colour map, but the shape of the scale matters more than the specific colours: find the darkest and brightest values on the legend.
Why do the vertical stripes matter?
Each stripe is a transient. A kick produces a vertical line because the whole spectrum appears at once. Regular stripes at a fixed spacing are a rhythm, and gaps between them are the space between hits.
What are the horizontal lines?
A sustained tone. A pitch that holds for a second draws a horizontal line at its frequency, and any harmonic of it draws parallel lines above. This is the most useful thing about spectrograms: you can see the actual notes being played.