Take a look at the diagram that I’ve prepared below. It shows a number of things. The first three waveform displays show each stage of compressing a signal. On the left is the uncompressed signal. The wide vertical differences show the short and long term dynamics of the particular piece of music (this is the end of Mosaic again, which is available for download at the FTP site). The middle waveform shows what happens when a 20:1 compressor/limiter (a limiter is simply an extreme case of compression) is applied. Any part of the sound that exceeds the “threshold” level is reduced by a specified ratio…in this case 20 to 1, which is a lot. If a signal comes in below the threshold it is allowed to pass through undisturbed.
Figure 1 – Three images of a file subjected to compression and a transfer function showing the input and output from a compressor/limiter. [Click to enlarge]
The overall RMS level of the song is reduced and the resulting waveform shows the telltale plateau appearance of most pop music. This is how things are portrayed in “The Distortion of Sound” film. But the processing doesn’t stop there.
Not only are the engineers trying to protect their recordings from overages but they are also tasked with making the final output as loud as possible. This additional processing doesn’t happen during the initial recording. It’s the job of the mastering engineer to crank up the amplitude so that the tune “punches” when played on FM radio, on your iPhone or through your Bluetooth portable speaker.
It’s not necessarily the best thing for the music but it is standard operating procedure in the commercial music business.
There are lots of creative uses of compression but normalizing is not one of them. Normalizing happens after the compressor/limiter brings down the peaks and flattens the waveform…and the dynamics of the music selection. This is the process that boosts the whole track up to the maximum allowed. If you’re using a PCM system at 16-bits, then normalizing will find the loudest instant in your track and amplify it to the 16-bit maximum AND then raise the rest of the track by the same amount. If there are no amplitude differences in the track (which is likely because of the heavy compression that was applied) then the whole track get maxed out.
That what should have been explained and animated in “The Distortion of Sound”, but it wasn’t. I don’t know why. When tens of thousands of people see this film online or at the Grammy museum, they’re being misinformed.