Audio Processing: Exploring Dynamic Range Compression
First Steps: What is Audio Compression?
The first step in becoming a master of digital audio compression is to understand exactly what it is we mean when we say it! Most tech-saavy people think of digital compression as digital data compression i.e. the practice of making computer files smaller so that they can be more efficiently stored, transported etc. While data compression can be very useful in audio processing, it actually has very little in common with digital audio compression!
So what is it then?
Digital audio compression (henceforth to be known simply as 'compression') is the process of reducing the dynamic range of an input signal. More simply put, the purpose of compressing is the reduce the difference in volume between the loudest part of a signal and the quietest part of a signal. This is usually accomplished by attenuating the portions of the input that are above a certain level. Compressors can have many parameters, but usually the core controls are:
-
Threshold - the level at which the compressor activates. This is how loud a signal must be in order to be affected by the compressor. Typical values are between -6dB and -25 dB depending on level of input signal.
-
Ratio - the measure of how much any signal above the threshold should be compressed. Usually expressed in the form X:Y (i.e. 3:1) - this means that for every 3dB we are above the threshold at the input we will be 1dB above the threshold at the output. Typical values are between 1 and 10
-
Attack Time - the attack time governs how quickly the compressor reduces the level of any signal above the threshold. Typical values are between 6 microseconds and 6 ms.
-
Release Time - measure of how long it takes after the input has dropped below the threshold for the compressor to completely deactivate. Typical values are between 1 ms and 1.5s.
Additional controls you may find on a compressor include:
-
Input Gain - used to boost or attenuate the level of the input signal before it is compressed
-
Output Gain - same as above, but at the output of the compressor
-
Makeup Gain - similar to Output Gain, Makeup Gain is used to counteract the reduction in volume that often is a consequence of compressing a signal. Remember the purpose of using a compressor is to reduce the dynamic range of the input which is accomplished by reducing the level of the loudest parts of the signal. This leads to a reduction of the average level of the signal as a whole - Makeup gain is provided as a way to return the signal to its original average level.
-
Knee - knee is a subtle parameter which affects how the compressor handles the transition between the level range that will not be compressed and the level range that will be compressed. For the purposes of this tutorial and project we will stick to what are known as "hard-knee" compressors (i.e. compressors whose full ratio is immediately applied at the threshold)
Below is a graph commonly known as a 'compression curve' which illustrates the threshold and ratio parameters.
Figure 1: Illustration of the impact a compressor has on an input signal. The threshold is the point where the slope of the line changes and the slope beyond that point is the ratio
Displaying attack and release time as succinctly as threshold and ratio can be a bit of a challenge, but you may gain an intuitive understanding of these parameters by considering the graph below. The waveform shown could represent any type of input signal.
Figure 2: Illustration of the impact attack and release have on the output gain of a signal over time. The blue line represents the original square wave input while the red line represents the output from a compressor