What is a frequency spectrogram?

A frequency spectrogram lets us see sounds. Unlike the waveform diagram, which visualises different levels, a spectrogram shows frequencies of different loudness. The X-axis represents the course of time. So usually the beginning of a song is on the left, the end on the right. The X-axis represents the pitch. Low frequencies are at the bottom of the diagram, high frequencies at the top (and therefore the mid-range in the middle). This is similar to music in notes.

In addition, a colour scheme helps to distinguish loud from quiet frequency ranges. The scale for this is usually found to the left of the X-axis in uebervinyl.de’s diagrams.

What does a frequency spectrogram show?

The frequency spectrum of a piece of music is a complex structure of many frequencies. As a rule, several notes are played simultaneously in a piece of music. However, each individual tone already consists of many frequencies – a fundamental and many overtones. The overtones define the timbre.

Crystal-clear sine tones without overtones can only be produced electronically. In practice, a concert pitch A of 440 hertz (Hz) sounds different when sung by a singer and completely different when played on a guitar or trumpet.

These differences in sound arise from overtones that are different in volume and produced in different ways.

What can be read from frequency spectrograms?

Uebervinyl.de uses spectrograms to create a sonic fingerprint of a song. This way, we can compare two different pressings and share with users what we have heard.

For example, whether there are differences in certain frequency ranges. Quite typically, different loud bass components can be detected. The frequency spectrogram above compares two different versions of Revolver by the Beatles. Here, the larger purple areas in the lower part of the diagram clearly show that the version in the diagram above reproduces considerably louder bass.

This example shows the song Let Me Out from the album Get The Knack. Here we see much larger swings in the treble in the lower diagram. While here the yellow peaks reach the upper edge of the diagram at more than 20,000 hertz, in the upper diagram the yellow peaks rarely reach beyond 18,000 hertz. So the lower compression sounds brighter and reproduces highs down into the range that is barely audible to humans. The distribution of energy across the different frequency ranges also has an impact on the perception of loudness. A few basics about loudness and LUFS are explained in another article.

What is the logarithmic scale of a spectrogram suitable for?

Spectrograms can be scaled in different ways. The logarithmic scale in the example above is well suited to make differences in low frequency ranges visible. This is because it is not the numerical value of a quantity to be represented that is plotted as a distance on the y-axis, but the logarithm of its numerical value. With the increase in frequency, the graduation lines of the skakla move closer and closer together. This leads to the fact that in the low frequencies, for example, the range between 43 and 86 hertz gets a good bit of area, although “only” 43 hertz are bridged in the frequency spectrum. In the high frequencies, distances of 43 Hetz can no longer be detected; here, the jumps from graduation mark to graduation mark bridge areas with hundreds or thousands of frequencies.

What is the linear scale of a spectrogram suitable for?

In contrast to the logarithmic scale, the linear axis scaling makes it possible to read differences in the mid and treble range particularly well. Since all graduation marks are equally spaced, the high frequencies fan out particularly well. In the bass range, on the other hand, differences are more difficult to detect.