One of the less-clear concepts in audio is that of serial versus parallel processing. As is the case with series and parallel electrical circuits, serial processing provides a single path for an audio signal on its way to the mix bus, while parallel processing allows multiple paths through which an audio signal may reach the mix bus. Each approach has its strengths and weaknesses, and knowing when to apply either of the two (or in some cases, both) facilitates better control over your mix.
The signal flow theory behind the two types of processing holds true regardless of whether you are dealing with software or hardware processing (or subgroups). Serial processing often (but not always) employs insert sends and returns; parallel processing typically uses auxiliary effects sends and aux effects returns. Let's take a closer look to set the record straight.
Dedication to Audio
Before we examine the signal flow, we need to have a general understanding of when and why we need to process an audio signal using either method. Serial processing is used when you want 100 percent of a signal processed, without allowing it to pass through in its original state. For example, let's say that you are using a graphic EQ to tune a house PA system. You want the EQ to process all of the audio reaching the speakers, not some percentage of that audio. To accomplish this task, we typically patch the house graph on the main L/R insert (software or hardware), or "in-line" after the main outputs of the console (hardware). See Diagrams 1A and 1B – all of the audio reaching the house speakers is EQ'd.
If we apply this concept to a single audio channel such as a kick drum, we usually use the channel insert, which interrupts the "normal" channel signal flow, re-routing it through an optional processor (See Diagram 2). This is true whether the insert is hardware or software. If we insert a graphic EQ on a kick drum channel, the ‘graph becomes part of the signal flow for that particular channel. This yields good news and bad news. The good news is that we have an equalizer that can be fine-tuned specifically for our kick drum. The bad news in the hardware world is that you cannot use that graphic EQ on any other channel. If you want graphic EQ on the snare drum, you'll need another outboard graph. Of course, if you have purchased a plugin, you can probably run it on more than one channel without extra charge – provided the CPU can handle the load. (Remember: The more plugins you add to a system, the more the CPU is taxed). Some digital consoles include a configurable software "rack" providing "X" number of processors, which you may freely assign to channels or outputs as you like. In this case, resources are treated more like hardware – once you assign a processor to a certain channel or mix bus, it is not available for simultaneous use elsewhere. Under most circumstances, serial processing is used for applying EQ, compression or expansion/gating, because these are processes where you don't want any of the original unprocessed signal "leaking" through to the mix. There are advanced mixing techniques that we'll discuss next time which twist this concept.
You Want Fries with That?
Parallel processing, on the other hand, adds an additional, processed version of a signal to the mix, while leaving the original "dry" version of the signal intact. The most popular example of parallel processing is using an auxiliary send to get signal into a reverb, and using an aux effects return to bring the reverb back into the mix. A quick refresher: An aux send is an additional "tap" on a channel from where the signal is split and routed "somewhere else." Where? Anywhere you want, such as input to a reverb or delay, or a feed to a monitor or headphone mix. Generally, we don't use an aux send to route signal to compressors or gates, because, under most circumstances, we want "absolute" control of dynamics. In other words, if you are compressing a vocal, you generally don't want the uncompressed version of the vocal coming through the mix.
Diagram 3 shows the routing for parallel processing using a reverb – routing that works with hardware or software. Signal is tapped off the channel using the aux send (shown in green). This extra path is connected to the input of the reverb, and the output of the reverb is routed to an auxiliary return, which is really just a stripped-down input channel. In the old (hardware) days, aux returns offered the most basic elements of control: level, pan and assignment to the L/R master (and, possibly, to the subgroups). Gain, equalization and straight faders were omitted as a means of saving space and money in console design. So an aux return treats our reverb like another input, which must be routed to the mix in addition to the original channel. That's why the aux effect return shown on the right feeds the L/R bus. We can have our original "dry" drum channels in all their wide-range frequency response (the red bus), and the "wet" reverb (often with a limit frequency response) on the side mixed in as a separate signal.
Parallel = Conserved Resources
Parallel processing is very useful because many channels can share the same processor. Looking again at Diagram 3, we can see that if we want to send Tom 1 to the same reverb that the snare is using, we simply turn up the aux send on the Tom 1 channel. In the software world, this conserves DSP resources, because you won't need a reverb on every channel. In the hardware world, it saves money and rack space, because you can have one or two really good reverb units for channels to share. It also increases efficiency in mixing. Suppose you have five channels of backing vocals all requiring reverb. If you decide to edit or experiment with the reverb sound, you can change the parameters once, and that change is applied to the reverb for all backing vocal channels. If you insert the reverb on individual channels, you'll need to make that same parameter change five times.
Suppose we try something less conventional by substituting a compressor for the reverb in Diagram 3. Our aux send would now route the drum channels into a compressor. The compressor "does its thing" and passes the compressed drums back to the aux effect return where they are added into the mix along with your original uncompressed drum channels. Why would you want to do this? You'll have to tune in next month!
