Strata Documentation: Decibel Maths

Decibel Maths And The Shape Of The Strata Graph

If you've used Strata for a little while, you might have noticed that the shape of the top of the Strata graph is not the same shape as that of the frequency spectrum of your master track. There's a reason for this, and this page explains.

The purpose of Strata is to show you the frequency spectrum of ALL of your tracks together so that you can see where every track is sounding in the spectrum and how loud they are relative to each other. Strata Vision simply takes each track's spectrum and layers them on top of each other. This allows the volumes of the tracks to be compared with each other. Often a glace at the Strata Vision graph will reveal that one track is much louder than all the others (when it shouldn't be).

While this view is excellent for comparing tracks, it leads to a distortion that means that the top of the Strata Vision graph is NOT the same as the frequency spectrum of the master track. This is because the Y axis is in decibels and decibels don't sum linearly!

Decibels are a logarithmic measure of the amplitude of an audio signal. We use them because they more intuitively represent our intuitions about audio volumes. But they do not sum linearly. Here's what ChatGPT had to say about it:

It is the amplitude of the waveforms that add up linearly. The decibel measure is a logarithmic measure of the amplitude.

This means that if I have three (uncorrelated) tracks, each at -6dB, the master track will be at -1.23dB.

Strata represents the volume of each track at each point in the spectrum by calculating the decibel level above an arbitrary Min dB level (which is a parameter of the plugin and defaults to -72dB). This is what most spectrum display software does.

So if you have three tracks at -6db, each track will be plotted as -6 - -72 graphical "units" high on the graph which is 66 units. Whereas if you have one track at -1.23db, its height will be -1.23db - -72 which is 70.77 units. If you plot the three tracks 66 units high they will stand 198 units high whereas if you plot the sum of the tracks it will be 70.77 units high) which is about 2.79 times less!!

What this all means is that if you plot a bunch of tracks with the height of each track being its decibel level above a baseline like -72dB, the sum of the height of the tracks is going to be much higher on the graph than if you took the audio sum of the tracks (i.e. the master channel) and displayed that in the same way. This means that, on a graph like this, the number of tracks active at a particular point in the spectrum is going to have far more influence on the height of the graph at that point than the dB of the sum of the tracks at that point. As there are usually far more tracks active in the midrange of a typical song than the bottom or the top, the Strata Vision graph will tend to be high in the middle and low at the ends, leading to a Strata Vision graph that typically has the overall shape of a normal distribution curve.

Apology

This concept is somewhat slippery, and the explanation above is not as clear as it could be. I apologise and hope to find better ways of explaining this at a later date. For now, the main thing is that you don't expect the Strata Vision graph to have the same shape as the spectrum of the master track.

References

Mmmmmmm! Cookies! This site uses cookies.