I've been told that I should use an analogue compressor or voice channel with compression as a front end when recording digitally, as this will reduce the dynamic range of the input signal and make better use of the A-D converter's available bit depth. Is this true?
Features Editor Sam Inglis replies: This is a widely held belief, but shouldn't be adhered to blindly. For one thing, it's important to consider the role of the compressor's attack time. In lots of sources, the highest signal levels occur during transient peaks, for instance at the start of guitar chords or drum hits. Many compressors are not capable of reacting instantly to these peaks, even with their attack time at its fastest setting.
If this is the case, the highest peaks in the signal will be passing straight through the compressor whilst the quieter portions of the signal are attenuated (see figure 2, right). With no make-up gain applied, the compressed signal will therefore tend to present lower overall levels to the A-D converter — yet because the transient peaks have not been attenuated, there is no additional headroom to apply make-up gain, and doing so risks clipping the converter.
If you do want to raise the average level of the signal prior to A-D conversion, then, you need to make sure you're using a compressor with an attack time fast enough to squash these transients, or teaming your compressor with a peak limiter that can catch what the compressor misses (see figure 3). Some input channels, such as the Focusrite ISA430, do offer both compression and limiting, whilst some A-D converters also feature 'soft limiting'. Other signals, such as vocals, present fewer problems because they contain fewer sharp transients.
However, it's worth thinking about why you would want to muck about with the dynamics of your input signal in this irreversible way. Modern A-D converters offer plenty of dynamic range — enough to capture the most demanding sources and still leave plenty of headroom to avoid clipping, especially if you're recording at 24-bit. It should therefore be unnecessary to compress a signal in the analogue domain simply in order to capture it at acceptable quality in a digital system.
On the other hand, lots of engineers just like the sound of analogue compression, and feel that it can't be reproduced using plug-ins or other digital compressors. In digital recording systems, the most convenient place to apply analogue processing is on the way in, since this avoids sending recorded signals through another stage of D-A and A-D conversion. So if you want to use analogue compression because of the way it sounds, it often makes sense to do so at the input stage.