Compressors
Compressors were introduced in music recording technology as a mean to help engineers cope with rapid sound level changes, especially when recording live music. A compressor balances the difference between the loudest and the quietest parts of a piece of music by automatically reduce the gain when a signal exceeds a predetermined level. The amount of reduction depends on the ratio set on the compressor.
Modern compressors are essentially Voltage Control Amplifiers (VCA) whose gain decreases as its input level increases. A VCA is a little amplifier circuit that has an input connection, a control port, and an output connection. The control port determines how much gain there is between the input and output, based on a simple DC voltage. The amplifier is designed to have unity gain (0 dB) with no reference voltage. An amplifier with no gain will have absolutely no effect on anything that passes through it, it will simply multiply everything passing thought it by 1, so the input will remain unchanged. A basic compressor circuit might look as below.
In the above diagram a signal passes through the VCA to reach the output. The output then enters a feedback loop by sent back to the output through the Side-chain and a Rectifier / Log Convertor. The side-chain will allow the implementation of external devices to control the compression for certain frequencies. A classic example is the introduction of a parametric equalizer into the side-chain in order to control the unwanted sibilance (emphasized letters s or t in voices). The Rectifier will convert the audio signal from an AC (Alternating Current) signal into a DC (Direct Current) signal and the Log Converter will produce a DC output proportional to the log of the average level of the audio signal (http://www.fm-transmitter.com). The log signal is now used as the reference voltage for the VCA. As the output level increases, so does the reference voltage which is fed to the VCA and as this reference voltage increase, the gain of the VCA (which is unity with no reference voltage) will decrease, thus providing gain reduction.
Besides VCA compressors, other designs are also available. We can find compressors using valves, FETs (field effect transistors), or computers with digital implementations. While the first two mentioned as not as accurate as VCAs, each one of them gives a certain sonic character to the input sound and might be more suitable for certain types of music.
Valve compressors tend to add a certain amount of distortion. Frequencies that are not detected by the compression might still be non-linear affected (attenuated). This sometimes results to a more musical sound than the VCA counterparts. Many modern compressors use VCAs for the actual compression to avoid this non-linear behavior while using valves for the amplification stage to soften the sound
http://www.soundonsound.com/sos/dec00
The Field Effect Transistor (FET) is often used as for compression within more cost-effective designs nevertheless this is not always true as in the case of LA Audio's Classic II stereo compressor/limiter. An FET uses a transistor to vary its gain and they have similar non-linear response as to valves, distorting the signal in a similar way.
Digital compressors use as a reference the sonic characteristics of the previous mentioned circuits to output their sound. They offer the ability to process audio in the digital domain under the precise control of a computer. The result output signal can be predicted in pre-recorded sounds as the computer performs sample to sample analysis of the input signal prior the output. .
Most compressors have at list three basic controls. The Threshold control, the Compression ratio control and the output gain control. The Threshold Is set in dBs and determines the level from which the compression will start processing the input signal. If the signal level is below the threshold, no processing takes place and the gain reduction meter, if any available, will read 0dB. If a signal exceeds the Threshold then the signal reduce in level according to the ration setting.
When the input signal exceeds the Threshold, gain reduction is applied according to the ratio setting. The ratio is also expressed in dBs. If a ratio of 2:1 is set then an input signal of 2dB above the threshold will result in 1dB of output and if the signal falls back below the threshold the gain returns to normal. The Output Gain is used to compensate in level from the overall level reduction caused by the compression. Two additional common controls in compressors are the Attack and Release. The attack time determines how long a compressor takes to pull the gain down once the input signal has reached or exceeded the Threshold level. With a fast attack, the signal is controlled almost immediately, whereas a slower attack time will allow the start of a transient or percussive sound to pass through unchanged, before the compressor does something about it. The Release control sets how long it takes for the compressor's gain to return to normal once the input signal has fallen back below the Threshold. In many compressors is also possible to select if the side-chain chain responds to average signal levels (RMS) or to absolute signal peaks (PEAK). The compressor will behave differently depending on the selection. RMS offers the closest approximation to the way in which our ears respond to sound, thus provides a very natural-sounding dynamic control, but short signal peaks will get through unnoticed, even if a fast attack time is set. This could be a problem when making digital recordings and we want to avoid clipping. In this circumstance, the Peak type will be more suitable as it will more accurately track the peak levels especially in brief signal peaks.
A drawback with most compressors is that high frequency sounds are attenuated in the mix despite the possible fact that we might only want to attenuate low frequency sounds. This is natural as high-frequency sounds tend to be much lower in level than low frequency sounds and conventional compressors do not have a complete solution to compensate that. However, more complex designs have been implemented in which the frequency spectrum is divided into two or more frequency bands and compression take place on each band separately
(http://www.soundonsound.com/sos/jan01/articles
/advanced.asp). Although this can introduce phasing problems, a well-designed compression device can produce very good results.
Dimitris Barnias 2004
©sonicspace.org 2005