More than ever before, it's possible to reproduce the functions of a hardware mixer with software in a computer. But can you completely replace a mixer with software -- and, more to the point, should you be trying to? Paul White plays devil's advocate...
A few weeks ago, I took part in a series of industry seminars where the distributors of a well-known brand of mixers were defending the position of hardware mixers in recording against suggestions from an equally well-known software company that the way ahead was to do all your mixing inside the computer. Predictably, this left a few of the attendees a little confused, hence the reason for this article. As is often the case, my feeling is that the truth lies somewhere between these extreme positions. It may well be true that most project studios can manage quite nicely without a large multi-buss console, but it's my conjecture that few practical studio setups are workable without a hardware mixer of some kind. To explain why, I need to play devil's advocate and argue both sides.
The Case For Software
Let's examine first of all the case of a softwarebased computer studio running on a Mac or PC equipped with a good-quality stereo soundcard. Providing this system uses exclusively software-based instruments to provide any MIDI parts, there's no practical reason not to mix everything within the computer. Indeed, keeping the signal in the digital domain after recording means that the problems of introducing noise and distortion by successive journeys through A-D and D-A converters, external wiring looms and external analogue equipment are neatly sidestepped. Providing you take care at the recording stage, ground-loop hum should just be a bad memory.
The mixed files (which can be saved at 24-bit to allow for later dithering) can be opened directly within mastering software, and then burned to CD without ever having to be converted back to analogue signals, so the quality of the recording is affected only by that first analogue-to-digital conversion at the recording stage.
This is pretty much the pitch from the software company's point of view, but there are questions it doesn't address. Firstly, how good is the mixer inside a software sequencer? In order to maintain signal resolution when multiple channels are mixed, a digital mixer needs headroom, just the same as an analogue mixer does. While analogue headroom is the ability to handle higher voltage signals without clipping, digital headroom means having a signal path with enough bits to handle the large binary numbers that are generated when numerous 24-bit signals are EQ'd, boosted in level and mixed. In practice, this means having a 32-bit data pathway inside the mixer, something that Emagic's Logic has had for several years now and which Steinberg has recently introduced in its premium Cubase VST32 package. Even so, the width of the data pathway doesn't tell you everything about how the mathematics of mixing are handled, and because a software mixer relies on the host processor for everything, the designers are more likely to take processor-efficient shortcuts than the designers of hardware digital mixers who can apply all their DSP resources to the task in hand.
A hardware digital mixer will generally allow you to have fully parametric EQ on all channels and aux returns, whereas a software mixer usually provides EQ as insert effects to be used when needed. Like all plug-ins, EQ uses processing resources, so there's a limit to how many channels of virtual EQ you can run at once, especially when running other plug-ins at the same time. My personal opinion is that EQ is greatly over-used, but I'd be the first to agree that it ought to be available at all times if needed.
Staying with the subject of plug-ins for a moment, some of the available processing software is truly excellent, but the designers always have to keep one eye on processor overhead. In practice, this means that processor-hungry effects such as reverb tend not to be nearly as good as a serious hardware counterpart unless they use up a disproportionate amount of processing resources. Very often a decent hardware reverb is actually cheaper than its plug-in counterpart, but it can be difficult to patch a hardware reverb into a software-based system that doesn't use a hardware mixer. What would be really welcome is a high-quality PCI-card reverb unit that could be addressed within a VST application just like any other VST plug-in but without using the host processor resources.
Hardware Mixer As Nerve Centre
What else does the hardware mixer offer that the software mixer might not? Aside from mixing, any hardware mixer, whether analogue or digital, provides certain essential control functions that all studios need. For example, it allows you to connect additional playback and recording machines and to route signals between them. At the push of a button you can play a CD through your studio monitors for reference, or you can copy a mix to cassette to play in the car. There's a control for adjusting the monitoring level, there's a headphone output, and there's a system of pre-fade aux sends that make it very easy to set up a monitor mix. A soundcard studio can only provide a separate monitor mix if it has an additional pair of outputs that can be fed from a pre-fade mix buss.
What's more, impressive though virtual instruments are, I would feel seriously deprived if I had to use them for everything. Nothing on the market compares with my Korg Wavestations, my Emu Morpheus or my expanded Roland JV2080, but to use these instruments within a mixerless soundcard system would require two or more additional soundcard inputs for each hardware synthesizer, sampler or drum machine. And in many cases, you need far more inputs to do justice to a multi-output sound module.
By using a small hardware mixer (digital or analogue) to combine the outputs from your synths and soundcard, you're effectively bridging the hardware and software worlds in an effective and controllable way. Each synth has its own physical mixer input(s), EQ and physical fader control and any channel can be processed using a connected hardware signal processor, such as a reverb unit. If the soundcard has multiple outputs, these too will have their own level faders, pan controls and effects send capabilities, so if you have a lead vocal recorded in the computer and you assign it to its own soundcard output, you can treat it with highquality outboard reverb. At the risk of giving a product a gratuitous plug, I've found the still-current Mackie LM3204 mixer makes an ideal companion for a computer system. It is a balanced line mixer with 16 stereo channels, four stereo returns and two patchable mic preamps. Each channel is controlled by a knob rather than a fader, but in a computer system where mixes tend to be automated, the mixer level controls are more likely to be used as level trims rather than for dynamic mixing.
What about a future in which everything has digital inputs and outputs? Won't that make it easier to connect everything to a computer-based system without requiring a mixer? In theory, the answer is yes, as digital I/O is cheaper to provide than analogue I/O, but as anyone who has tried it will probably have discovered, interfacing digital equipment can be far from painless. Even if everything can run at the same sample rate, you have to work out what should be master and in which order to connect everything, and unless the equipment is very well designed, clock jitter may compromise the audio performance. The only really satisfactory system is to have everything controlled by a master word-clock generator, but a good one tends to be expensive, and you know how reluctant project studio owners are to spend money on a box that doesn't make an interesting noise! Even if you are prepared to meet the expense, you may find that many pieces of studio equipment don't have a word-clock input.
Back To Analogue?
Though mixing signals in the analogue domain might be seen as a retrograde step, bear in mind that an analogue mixer is oblivious to issues such as sample rate, sync status or bit depth. All sound sources are welcome in the analogue domain, and very often mixing in analogue also sounds better. After all, analogue mixer technology is decades old and has been refined to a very high level. Digital mixing, on the other hand, is a relatively new technology, and has a long way to go before it can be considered optimal. What's more, analogue EQ tends to sound more musical than all but the best digital equalisers, and even a budget 12:2 analogue mixer will never show you a 'DSP overload' message when you try to EQ all channels simultaneously.
Summing up this section, then, adding an analogue mixer to a computer system provides an easy means to integrate your external MIDI instruments and signal processors, it provides greater mixing and processing flexibility where the soundcard has multiple outputs, and the amount of EQ that can be used is not limited by DSP resources.
The Human Interface
Finally, let's consider what is perhaps the most important aspect of a hardware mixer -- a tactile control surface. Even if you're used to a computer mouse, you can only adjust one thing at a time, whereas a traditional recording engineer will often be reducing a channel level with one fader while increasing another with the other hand. Mixing is a type of performance art where immediate and simultaneous access to all level controls is very important.
The software brigade will no doubt counter by saying that you can add an inexpensive MIDI fader box to control the mix levels within the software mixer, and this is unarguably better than doing everything one part at a time using a mouse, but none of the more affordable MIDI controllers uses motorised faders. Why is this important? If you use mix automation, the level settings will change while the physical faders stay put, so to edit a mix you need to move the fader to the null point before you can make a meaningful edit. This is tedious at the best of times. Furthermore, you may have a box with 16 faders controlling a 24-track mix, so you have to keep switching between the fader functions. Again this means the faders are in the wrong position most of the time. Perhaps you also wish to switch the faders to change the aux send levels, but again, the gain settings and the physical fader settings don't match up. You only have to try to work with such a system for a few hours to realise that a moving-fader control surface is the only really useable option.
Apparently both Steinberg and Emagic have such controllers in the pipeline, but the problem is that moving faders and the associated support hardware are not cheap. In fact, when I've spoken to hardware designers, most of them have admitted that it's almost as cheap to build a complete digital mixer as a moving-fader control surface. Even if an affordable moving-fader control surface were practical, it still wouldn't do away with the need for an external mixer in situations where any number of external synths and effects processors need to be accommodated.
Currently, a cheap digital mixer such as the Yamaha 01V or 03D can double as a MIDI control surface, yet costs less than any serious moving fader controller I can think of. It's simply a matter of switching the fader banks between real and virtual mixer channels. At the same time, it's still a mixer, it has lots of EQ and onboard effects, and it can tie in all your hardware instruments and sound processors. Such a mixer can also handle your monitoring and provide central access to your other recorders and playback machines.
Things look good for mixers in the future too, though I'm sure they'll change considerably from what we have today. The most radical step (which may not be that far away) will be when mixers are produced with FireWire/mLAN computer interfaces -- or some other fast interface if that's the way progress takes us -- which permit direct connection to a computer. This will eliminate the need for a soundcard, and remove the bottleneck caused by limited soundcard I/O. The mixer of the future will probably be a more conscious combination of mixer and remote-control surface, and to make this work effectively, there'll need to be a great deal of cooperation between mixer manufacturers and the main music software developers. In turn, this may bring about the need for more standards to be introduced, following on from the likes of ASIO and VST, in order that products can be made compatible without special drivers having to be written for every combination of hardware and software the user is likely to encounter. It's certain to be an interesting future, but in my opinion mixing consoles in one form or another are likely to remain a significant part of it for a long time to come.