What's best: hardware or software? Analogue or digital? Most studio tools have their strengths, but how best do you combine hardware and software in a modern project studio?
Whether you're a wet‑behind‑the‑ears newbie, or have many years' recording experience under your belt, the chances are that you do a lot of your mixing work now 'in the box' (ITB), on your computer. Indeed, after the recording stage itself, it's possible now to do everything inside the computer, with no other hardware except a mouse, a VDU, and a pair of speakers or headphones. The ITB approach to mixing gives you the ability to fine‑tune every parameter in your mix, and of course you get full recall of your project.
At the other end of the scale is 'outside‑the‑box' (OTB) analogue mixing, using only traditional analogue hardware and tape. There are fewer people working this way now, partly due to cost and partly inconvenience, but partly also due to the improvements made in DAW design and plug‑in instruments and processors.
A few issues ago, we explored in detail what makes analogue gear sound like it does, why that sound might — or might not — appeal, and what advantages it still has in the studio. I won't retread that ground here (you can find the article on-line at /sos/feb10/articles/analoguewarmth.htm), but how you can integrate select pieces of outboard hardware into your software‑based system is another question altogether — and one that we're frequently asked by readers. So what follows is an attempt to unpack the various issues you might face, and decisions you might have to make if you want to operate a software‑hardware hybrid studio at home.
Tracking through a processor such as a compressor or de‑esser, or through an effects unit such as a delay pedal for a guitar, is relatively straightforward: you just take a feed from the last unit in the chain into your audio interface input and press record; and if you want a dry line or DI feed for safety, just take that before any processing appears in the signal chain. Job done. The same goes for tape: track to tape as normal, and then play it back into your DAW — although there's a useful time‑saving tip when working with tape that I'll discuss later. There's little more to say in this context about the recording stage.
When it comes to mixing or mastering, though, the options for routing to and from your hardware are more diverse, the demands for multiple 'instances' of your hardware become potentially greater, and the integration of all that outboard into your system can therefore become much more complex. When we talk about running a 'hybrid studio', it's really this mix‑oriented system that we have in mind: a setup where the DAW software remains the central hub of the studio, but the hardware and software is configured in such a way that you can simply patch in hardware to add a little character, quality and immediacy where you need it most, without having to leave your otherwise ITB mix. To some, the term might also extend to how best to sum the various signals in your mix, whether individual tracks or stems, although to my mind that's a slightly different question. Hugh Robjohns explores the arguments in favour of both analogue and digital summing on page 32102 .
So, if you're interested in augmenting your software setup with hardware — or if you're one of those rare folk who have only just decided to invest in a computer and would like to know what's possible — how do you go about plumbing your hardware into your DAW in the most efficient and flexible way? What hardware is worth using? What options become available when working with a desk alongside your DAW? What about recalling settings on your hardware? And what tools have the software designers given us to make setting up a hybrid studio that little bit easier?
Before we get on to anything to do with your choice of software or hardware processors, let's deal with one of the main practicalities. If you're planning to add any outboard gear to your system (other than on the Master stereo bus — and even then if you want a separate channel for monitoring), you'll need a means of routing several channels of audio from your DAW to your hardware and back. That may or may not mean using a patchbay or mixer, both of which I'll discuss later, but it definitely means using an audio interface with multiple analogue input and output (I/O) channels.
It's not terribly expensive these days to purchase a USB or Firewire audio interface that includes several channels of analogue I/O: something like the MOTU 828 Mk3 is typical, giving you eight mono (four stereo) analogue audio channels going in and out of your computer. At first, that may sound like plenty... but you'll need at least two outputs to monitor your DAW software's stereo mix, which leaves you with eight inputs and six outputs. A single dual‑channel hardware compressor could easily use up three of those outputs (one audio input per channel and one side‑chain input) and two inputs to take the processed signal(s) back to your DAW. In this scenario, you'd be left with seven more inputs, and only three remaining analogue outputs on your interface.
Although there are traditional ways around such limitations — you can always 'print' (record) the output of each effect to its own track(s) — realistically, if you plan to use more than three or four pieces of outboard simultaneously, you'll need to start thinking about adding more analogue I/O to your setup. Most multi‑channel interfaces, like the MOTU device mentioned above, include at least some digital I/O in addition to their analogue channels. Digital inputs and outputs come in various shapes and forms, but the most popular are ADAT 'lightpipe', which offers up to eight channels, depending on the sample rate, stereo S/PDIF via phono plugs, and AES/EBU, which can carry two channels on a single cable via three‑pin XLR connectors, and is often configured to carry eight channels on D‑sub to XLR breakout loom.
If you opt to go the PCI or PCI Express route, there are some other setups designed to give you more than enough I/O. MOTU's PCI424 system, for example, enables you to connect up to four of their breakout boxes: adding four of their 24IO boxes would give you a system with 96 I/O in only 4U of rack space. See the 'I Need More I/O' box for more advice on expanding your setup.
With your audio travelling back and forth between your computer and the analogue domain, you'll need to get the best converters available, to avoid nasty noise and distortion elbowing its way into the recording, right? Well, yes... to an extent. Obviously, the better your converters, the fewer unwanted distortion artifacts will creep into your mix at each stage of conversion; and the fewer times you go through A‑D or D‑A conversion stages the better. So, in theory, it makes sense to get as good quality converters as you can afford and, if you plan on using more than one outboard processor in series, to connect one device to the other, rather than use A‑D and D‑A stages for each.
That said, converters would almost never be top of my shopping list — and you might be surprised by just what you can get away with, given the quality of audio interfaces currently on the market. Although there are also a few high‑end converter manufacturers such as Burl, whose products are deliberately designed to colour the sound (albeit subtly) in a pleasing way, most converters are designed to be as clean and neutral as possible... and the fact is that, as bad as some designs were in the past, the converters on most audio interfaces these days are actually rather good. (Note that I'm not thinking about the mic preamps in your interface here, where there's a more tangible difference between different models!)
Even with modestly priced models such as the Behringer ADA8000 ADAT expander, audio signals should stand several 'generations' of conversion before there's audible degradation of the signal quality. Furthermore, if you're sending signals out of the box and back, it's usually to add 'flavour' or 'colour': there's no point sending audio out and back to get clean gain when you have infinitely tweakable access to that in your DAW. So even in the event some tiny artifacts do creep in, you might discover that it doesn't present the problem you'd imagined, or that it's at least a price worth paying to get the tonal colour you're seeking.
If you have everything else you need, then maybe it's time to think about upgrading your converters... but your ears and your wallet have to be the final arbiter on this point. I'd suggest that any cash burning a hole in your pocket might be better spent on more noticeable elements of the signal chain, such as EQs, compressors, mics or preamps where your money is likely to have greater impact. I guess the old adage "if it sounds right, it is right” applies perfectly here.
With the physical requirements of your audio interface out of the way let's move on to the software side of things. What's the best way to configure your DAW software to route signals to and from your hardware? The answer to that question will always be similar, but it does vary slightly from DAW to DAW. You can also toss into the equation the routing software that comes with your audio interface (such as RME's Totalmix and MOTU's CueMix, for example), because these offer another layer of routing flexibility between your DAW and the outside world. The answer will also depend on how many hardware devices you're using, how much flexibility you want in your effects and processing chains in the analogue domain, and how you plan to connect them to your interface. Let's start with the approaches taken by most of the leading DAW software.
At the most basic level, in any system, all you have to do is connect your hardware via the analogue inputs and outputs of your audio interface, then route the output of a track in your software to that physical output — and then monitor the sound via the interface's input on a different channel in your DAW. This is rather a primitive way of working, though, because it requires that you use a whole DAW channel each time you want to go out into the analogue domain, and another when you come back: if you want, for example, to use a hardware compressor, followed by a plug‑in, followed by another hardware device, then you're going to be using several channels in place of one, which quickly becomes cumbersome and confusing.
For this reason, most modern DAWs now include an insert plug‑in of some sort, which will send your audio out of a user‑defined audio-interface output, and receive the feed back through a similarly defined input. These plug‑ins enable you to connect the hardware device via a single insert slot on a DAW channel, just as you would with any plug‑in — except, of course, that you're limited to a single instance of each piece of hardware. See the box on the opposite page for more on how to set these plug‑ins up using some of the most popular DAWs.
It's a good idea to think about the names that you give to these 'hardware plug‑ins', and to my mind there are three or four broadly sensible approaches to consider. By default, most systems will give them the same name as the physical interface I/O that you select for the plug‑in — and you might be perfectly happy leaving it like that, particularly if you're likely to be plugging different pieces of hardware in and out of your interface for different mixes.
Perhaps you prefer to perform software routing tasks within the (often) more flexible software that ships with your interface. I've done this in the past using RME's Totalmix software, rather than specifying specific outputs, because it enables me to 'mult' signals to different devices, or multiple mixer channels, in a much easier way than I can in my DAW. With such software, you can save different configurations, which can be handy for different projects. If you're not into this sort of parallel processing, and having variable setups, this approach would probably add an unnecessary layer of complexity.
If you have hardware permanently wired in to your interface's I/O, it's a simple case of labelling each plug‑in with the name of the device in question: if you have, say, a Golden Age Project Pre 73 patched in to audio input and output channel 1, you could simply label it "GAP Pre73” or some such. However, if you're forever tweaking your setup, as I am, you'll always have to remember to go in and update the labelling of the hardware plug‑ins, so it's not always ideal!
When you're running a setup with more devices than you can accommodate with your analogue I/O count, and you're using a patchbay or a mixer to organise things outside the box, or simply where you want to change effects chains in the analogue domain, it can help to specify 'Patchbay out 1+2' (or mixer out 1+2', for example) instead of 'analogue out 5+6', as it can save you searching for the label on the patchbay.During the course of writing this article, I decided to sell my mixing desk (It's not the first time I've reached that conclusion), and to plug everything — interface, converters, mic preamps, outboard, synths — into a couple of bantam patchbays and label it all according to the patchbay numbering. Ahhh... blissful simplicity! Keeping organised is important if you want to avoid routing confusion in the middle of a busy mix. You need to try to find a system that works for you, that helps take the pain out of routing, and stick to it.
So what about more complex hardware setups, with many different hardware devices? We've discussed patchbays many times in these pages in the past, so I won't dwell on them here, other than to say what a godsend they can be to anyone running a hybrid setup with more device I/O needed than they have I/O available on their computer, or to anyone wanting the option to build different chains of analogue processors and effects for use with their DAW. If you need a primer, I'd recommend a thorough read of the article Hugh Robjohns wrote back in SOS December 1999 (/sos/dec99/articles/patchbay.htm). Essentially, all you need do is have your DAW take the place of the mixer in the setups described there.
An alternative means of patching in the analogue domain is to use a mixing desk, with your processors or effects permanently plumbed into inserts and sends. Your DAW remains the hub of the studio, where all the serious editing work is done, but you're able to send any channel from your DAW out to any channel of your desk, either for processing with the channel EQ, inserts or compression, or any aux sends you have set up. There are a few ways to connect your computer and desk to each other (let's ignore the patchbay for now), and which is best will depend largely on how much I/O you have and the configuration of the desk.
With a decent large desk, you're likely to have tape sends and returns on each channel, and the normal approach is to connect the patchbay or interface I/O to these. You can then mix on the desk, using channel EQ if desired, and record the results back to the DAW. I you want to use any aux sends under this simple setup, you patch the returns into spare channels and record those. Unlike analogue tape, of course, you have the option of keeping previous takes and then selecting between them all later.
If you plan to do any actual mixing on your desk, though — recording the summed result — you'll need to save some of your interface's inputs to record the signal from the Master bus, as you will if you want summed stems, some for the group buses, any dedicated effects return channels and so on. Before long you'll probably end up building a full‑on analogue studio, with a computer pretty much taking the place of a tape machine! As you'll see when I discuss ergonomics, below, this may or may not be the best approach for you.
On a smaller desk, such as one of the Mackie VLZ3 series, the routing options are more limited, as you tend only to get a stereo tape send/return for two‑track recording. In such cases, you have to look at what options the mixer presents. With the VLZ3 series, you'd be able to plug your interface outputs into each channel's line inputs and take a feed from either the channels' direct outputs, or the group bus outputs (assuming it's a large enough model to have those). If you're planning to acquire such a mixer, it's probably a good plan to make sure you have a few spare channels on the mixer, so you don't have to keep plugging and unplugging the jacks that feed the line inputs.
It's also possible to take a feed from the insert points on a desk, but that's not really advisable, for a couple of reasons. Firstly, on smaller mixers these are very often unbalanced send and return signals combined on a single stereo jacks; and secondly, if you're planning to use a mixer to link to other hardware devices, you'll probably want to keep the insert points available for patching to those.
If you're looking to acquire a desk to give you some 'analogue flavour', there are plenty of options. Popular recent models include the Toft ATB, TL Audio M1 and, slightly lower in price, the Mackie Onyx range. However, none of these is an investment to be taken lightly. Prices on the second‑hand mixer market have fallen through the floor in the last few years, and you can easily pick up an old desk with characterful EQ, such as a TAC Scorpion or a Soundcraft 6000, for what appears to be a bargain price. Beware, though, that the quality and state of repair of old consoles varies massively — and even if you find one in perfect working order, you may still have to go through the rigmarole and expense of replacing many of the capacitors.
A promising development in the last few years has been the emergence of dedicated hybrid mixing devices, such as SSL's flagship Duality console, which is a fully‑featured analogue console whose faders can also be used to control most DAW software. Their Matrix does the same thing on a smaller scale, and dispenses with the expensive channels of preamps, analogue EQ and compression, leaving you with an excellent line‑level summing mixer and a very convenient patchbay system. Audient have also made inroads with their Zen mixer, which uses the DAW to record fader movements — whether for DAW control or analogue‑mix recall — and recently the much more affordable ASP 2802, launched at this year's Frankfurt MusikMesse. Digital mixers, of course, have been capable of doing some of these tasks for several years, with products as old as the Mackie D8B and DXB offering a means of DAW control as well as a means of digital mixing and routing. More recent digital devices such as the Yamaha N‑series mixers have made DAW/hardware integration at the more affordable end of the market much easier.
It's only recently, though, that we've seen analogue devices offering anything like this degree of integration reach anything near a project‑studio budgets. The Allen & Heath ZED R16 is an analogue mixer that transmits MIDI data from its faders — but the faders are static, so there's no mix recall. Mackie's latest Onyx mixers (the 1640i is reviewed in this issue) don't offer you control over the DAW's faders, but do give good, easy options for routing the signal between the DAW, the mixer and the outside world — and the footprint is such that you could conceivably use it alongside a dedicated control surface. So there are some great hybrid mixers out there, but there's still room in the market for something that really does offer you the same degree of control over an analogue mix and an ITB mix (I imagine anyone who can run with this design idea and manage to do so to a reasonable price will find an eager market in the education sector).
Similar developments have been occurring in the humble world of the patchbay, and again SSL seem to be leading the drive to further this particular frontier. Their X‑Patch is an eight‑way patchbay that's controlled by computer software, routing any input to any output without the need for patch cables, and multiple units can be linked to increase the I/O count. Other companies have developed patching systems for mastering applications, but which could also prove useful (Manley Labs' Backbone, for example). Excellent as these products may be in delivering a flexible but top‑notch analogue signal chain, they remain an expensive option compared with a £30$50 Behringer patchbay and a few decent patch cables, or even a more commercially priced Neutrik bantam patchbay.
Your computer can replace the recording role of the tape machine, but the sound of analogue tape remains a tough act to model and there are very few digital products that come close to giving a realistic tape sound (arguably the best emulation, the Anamod ATS1, is an analogue outboard device!). So, if you decide to use real tape, how do you hook the machine up to a computer‑based system without a mixing desk?
At its most basic, the tape machine has analogue outputs and analogue inputs, just like any other device, and you can connect it to a patchbay, or your interface's I/O just like anything else. Most interfaces now offer you the option of setting their inputs and outputs to the ‑10dBV consumer level or the +4dBu professional level, and it obviously makes sense to set these to the right one for the machine in question.
You can, of course, record directly to multitrack tape and subsequently transfer the recordings to your DAW for editing and mixing. It's not strictly within the remit of this article — but it's worth using a few lines to describe some tips and tricks for speeding things up a little. Assuming you have your tape outputs connected to a patchbay, or to your audio interface's analogue inputs, you're able to track simultaneously to the tape machine and to your DAW, complete with tape effect printed. You'll need a DAW that features automatic delay compensation (pretty much anything other than Pro Tools LE or M‑Powered). You also need a latency delay plug‑in of some sort, such as Voxengo's free Latency Delay, and you place instances of this on the input channels that are connected to the tape deck outputs.
If you send the same outputs from a desk both directly to your DAW inputs, and to your DAW via the tape machine, you'll be able to determine exactly what the delay is due to the physical space separating the record head and the repro(duce) head. If you enter that number in your plug‑in, you should be able to record tracks through your tape machine, with the machine's output set to monitor the repro head, while simultaneously recording other sources straight to your DAW. Your DAW's automatic delay compensation will automatically put all your tape tracks in the project at the right time.
There's also no reason you can't try bouncing a stereo mix in real time to a tape machine and back into your project in this way, without having to go through the chore of transferring the tape recordings manually.
If you've ever moved from using a desk to working inside the box, although you may have waved goodbye to the experience of hands‑on tactile control, you'll probably also have been delighted by the lack of clutter: mixers can occupy a significant chunk of space without doing much a lot of the time, and freeing up that space can be liberating.
If you have no console, and don't have several banks of motorised faders and rotary encoders in front of you, you have an opportunity to set up your studio in a configuration much more akin to a mastering studio than a conventional mix room, with your choicest pieces of gear arranged in such a way that you can reach out and patch them in, tweak the controls and see all the meters and settings from the listening position, or near to it, at any rate: no more walking to the end of the desk to twiddle a compressor control and walk back again to find that it doesn't sounds as good as you hoped (and then repeat ad nauseam).
It's not just about ergonomics, though: many people find that when they get rid of their desk, the stereo imaging improves, as the sound from the nearfields no longer bounces off the console to cause comb filtering. So when you (almost inevitably) start to augment your setup again with choice bits of hardware, it seems a shame to replace a desk directly with a large control surface (or a flat office-desk surface for that matter) that will do exactly the same thing to the sound.
I've tried a few control surfaces over the years, ranging from products like the old Keyfax Phatboy through the Behringer BCF2000 and Mackie Control to the Euphonix devices. Good as they all are in some respects, I've found them all wanting. I'm sure things will improve in the near future, particularly with the recent leaps in multi‑touch screen technology and OS support for this from Apple and Microsoft, which should make control surfaces much more personal. But I've reached the conclusion that for the time being I'm better off using a mouse (or better still a trackpad) a keyboard and dedicated numeric keypad with DAW shortcut key commands assigned to it. This takes up a lot less physical space and leaves me freer to get on with the job in hand, without the sonic sacrifices of placing large surfaces between the speakers.
I'm simply saying that you have many different options when configuring a hybrid setup, and it pays to consider what best suits the way you work. If I have anything in front of me now other than the obligatory computer input devices, I'd personally rather it was my monitor controller, bus compressor(s) or other key pieces of hardware than a bank of faders I rarely use.
Of course, a control surface may be the right choice for you, simply because you find the hands‑on mixing experience more intuitive. What do you need to consider when choosing your means of control? Again, it depends very much on your way of working. Moving faders may look impressive, but are you the sort of person who spends lots of time tweaking plug‑in parameters that are more suited to rotary encoders, or even a mouse? If it's just transport controls and a single fader you require, dedicated products such as the Frontier Designs Alphatrack offer an elegant solution in a small footprint — but, again, you might find that you can do all you need with a computer keyboard and a mouse. A keyboard, in particular, can be made much more useful when embellished with a set of Editors Keys stickers (which provide shortcut labels), or something similar.
Don't forget that some hardware devices are also controllable via MIDI. I have two such effects units in my current setup. If you wish, you can map control surfaces to control those devices — to select patches, set delay tempos and so on — and if you can do this, of course, you can also record the automation data into your DAW for recall or tweaking.
One of the biggest difficulties presented by analogue devices relates to project archiving and recall. With a digital device, such as an algorithmic reverb box, it's easy enough to save a patch, take a note of it and load it back when you boot up your DAW project, but with most analogue devices, you'll need to set them up again manually, which poses two questions: how do you keep a record of what settings were used on a session? And how do you ensure that those settings are repeatable?
Any devices with digitally controlled analogue circuitry can recall settings automatically and some manufacturers, including AMS Neve, have also developed digital snapshot facilities, whereby the user captures a record of the device's settings digitally in software, and then uses the snapshot to make manual setting of the controls more accurate — rather like tuning a guitar string while watching a chromatic tuner. The company's 8816 summing mixer, 8803 EQ and 2254/R compressor and most other Neve products use this method of recall.
The other approach to capturing settings is the good old-fashioned analogue method: keep a visual record. In days gone by a log book might have been used to note down the settings by hand, but we now have the luxury of digital cameras, and most of us have a sufficiently good one in our mobile phone. Though popular, this approach does have its down sides: you need to go through and organise your pictures, store them somewhere handy and figure out how to link them with your DAW project. It doesn't take a huge leap of the imagination to see DAWs allowing you to import images to store hardware settings alongside projects or individual tracks (wouldn't it be nice to have everything stored in a single project!), but I'm unaware of any such integrated system at present.
One alternative is the Tea Boy software, which includes virtual panels for a wide range of hardware devices used in recording studios, and allows you to record settings by hand or visually, by turning dials. It's a good idea in principle, though unlike the digital camera it still requires you to enter all the data manually. Also, there's no guarantee there'll be a suitable virtual device to capture the settings, particularly if you're creating chains of effects, or using non‑standard studio gear (modified units, for example).
If you don't have access to the Neve‑style facilities, recalling settings accurately can be a pain. As with an automated analogue mixer, you can often find that the tiniest differences in settings can combine to give you a significantly different sound, which might be magic, but will more probably be frustrating. This is why mastering processors usually have detented dials, where you can change parameters in fixed increments, so you have a precise setting that can easily be recalled.
An obvious way in which plug‑ins and hardware differ is that you only get a single instance of hardware: if you want to run two of these devices simultaneously, you'll need to buy another. Fortunately, the DAW, and a few tools designed to run inside it, can make life easier, offering you the means to create virtual 'instances' of some hardware, both freeing up the real thing to process other tracks and keeping a record of the sound to use later.
The most commonly used tool is probably convolution, which can capture a reverb sound pretty accurately — so long as there are no long modulations in the algorithm; and even then, it's likely to be a passable imitation. Similarly, Waves' Q‑Clone plug‑in can send an analysis signal through a hardware EQ and 'sample' the settings. This does more than just bouncing your channel through the device, as you're able to apply the same setting to another channel (perhaps you were happy with the tone of a guitar part, but found that it needed re‑tracking due to the performance, or a requested edit). Since then, other plug‑ins, including the TC Electronic Assimilator for the Powercore DSP platform, Voxengo's Curve EQ and Apple Logic's Match EQ, have been able to offer EQ‑matching facilities, although none is as automatic or convenient as Q-Clone.
What no such user-configurable device can yet do well, though, is to capture anything in the time‑domain, so modulation in reverbs, and some of the distinctive distortion artifacts of compression or tape saturation are less easy to 'sample'. Focusrite and Sintefex have created the Liquid Mix DSP system and Acustica Audio the Nebula plug‑in, both of which use multiple convolution impulse responses to mimic the dynamic response of hardware such as EQs and compressors. In fact, it's possible to capture such impulses yourself for Nebula — but it's far from an automated process at the moment, and certainly not something you'd want to do while you're busy with the creative side of your brain on a mix. Perhaps, with time, this technology will mature and the GUIs improve to make things easier for the end user. As things stand, the only real option is to bounce your audio through the hardware device and capture the result as audio — or, in other words, 'print' the effect. Thankfully, the unlimited (or at least massive) track counts of modern DAWs mean you have the space to keep the dry version and bounce as many times as you like.
That pretty much completes this guide to setting up a 'hybrid' studio, but I'll leave you with a little cat‑among‑the‑pigeons thought: computers are getting faster, and plug‑ins are getting better at mimicking pretty everything that analogue can do... so perhaps it's simply a matter of time before it really is impossible to tell the difference in sound between 'real' gear and the plug‑in equivalents. I'd also hope to see similarly promising leaps in the design of control surfaces that enable you to interact with these models. So before you decide that you need to invest a heap of cash in esoteric hardware, think hard about what it's worth spending your money on. What hardware really can't yet be 'done' in software? The revered Manley Massive Passive has been emulated to a quality that Manley Labs are happy to put their name to; and Lexicon have released their most celebrated reverb algorithms as a native plug‑in. But it's as much about what has or hasn't been emulated as about what can or can't be. There are plenty of desirable devices that don't yet appear to have been been modelled successfully: I can't think of a great plug‑in equivalent to some of the earlier Lexicon reverbs, like the PCM60, or similarly unique‑sounding units like the Klark Teknik DN780; and there are plenty of quirky guitar effects pedals out there too. I, for one, will be pleased when we can do all the processing we want in the box, and spend all that hard‑earned cash on nicer spaces in which to record and listen — but that time isn't here yet. Spending your money wisely on good hardware gives you something unique, and it gives you an asset that, unlike most software, keeps at least some of its value. For the time being, at least, hybrid systems really can offer the best of both worlds!
Here's a quick guide to setting up hardware inserts in some of the best‑selling DAWs. Typically, you'd insert processors such as compressors and EQ, and distortion and some modulation effects, as inserts, and time‑based effects such as delays and reverbs on send, effects, or aux channels (the name will depend on the DAW you use). But in a DAW, the principle is the same, whether using inserts or sends: you place the 'hardware effect' plug‑in as an insert on an audio or group channel, or on an FX/aux channel and route the signal from your audio channel to the send effect in the normal way.
Logic employs an audio I/O plug‑in to send signals out to hardware and back. It's a simple matter of selecting the outputs and inputs you want to send to and record back from (or monitor, if not recording) and setting the levels.
Digidesign Pro Tools
Pro Tools includes a dedicated window for setting up and labelling audio connections. In this window, you can select the 'insert' tab to create new hardware inserts. You assign inputs and outputs to the inserts and label them as you desire. Then, back in the mixer window, you click on any insert point and you can select the relevant hardware insert. Renaming the inserts in this way doesn't relabel the physical outputs of your audio interface, which is a nice touch.
There are some limitations in both of the 'lite' versions of Pro Tools, LE and M‑Powered. Currently, you can only route audio out and back on the corresponding hardware I/O: if you send out on channels 1+2, you have to come back on inputs 1+2. It's not an insurmountable problem, but as many people tend by default to monitor on outputs 1+2 it's worth being aware of.
In Cubase you'd configure 'external plug‑ins' in the Connections window. It's a similar approach to that used in Pro Tools, in that you assign I/O to a dedicated hardware insert plug‑in and then open that plug‑in in the desired insert slot on a channel in Cubase — although you can also name hardware synths and sound modules and can assign both audio and MIDI tracks to these. Again, you're free to leave your physical outputs' labels alone. This is arguably the most elegant implementation of the idea.
No surprises for guessing that Sonar takes a similar approach to most other DAWs. However, there's one rather frustrating quirk in Sonar (from version 8 onwards), which means that if you use the External Insert plug‑in on a mono channel and set it to use one analogue output of a stereo pair, the other output of the same pair becomes unavailable for use with another outboard mono processor. The workaround is to use a stereo External Insert plug‑in on a stereo bus, and route mono channels to that bus, panned either hard left or hard right, depending on which mono effect you'd like those channels to be processed by. This should work fine for analogue gear, where the latency incurred by the processing itself is practically non‑existent, but pairing an analogue and a digital processor in this way is likely to cause problems with Sonar's delay calculations.
Reaper, again, includes a hardware routing insert plug‑in called ReaInsert. It's similar to the others discussed above and, like Cubase's, enables you to automatically calculate latency but, in common with Reaper's other plug‑ins, it includes wet/dry mix control, which makes it more flexible than the equivalent plug-in in most DAWs.
If you have a DAW that doesn't include a dedicated I/O plug‑in, such as MOTU Digital Performer for example, it's still usually a fairly painless process to patch in external hardware: you simply need to create an additional track each time you want to bring a return signal back into the DAW.
I like analogue consoles, and there's something refreshing about having to make mix decisions in real time, just as you would when mixing the front-of-house sound at a gig. When you combine an analogue console and DAW, you have the opportunity to get some of this feeling back, while keeping the convenience of recall.
Recently, I started to take editing and mixing to a certain degree of completeness inside the box, using just the mouse, and only then routing all the channels out to the mixer. It wasn't analogue summing I was looking for. Rather, I wanted to exploit the playback‑looping and overdub/comping functionality of modern DAWs to enable me to mix in real time on an analogue console and capture the multitrack product. I've been pleasantly surprised by the results...
It's easy to record several passes of a mix through your desk by looping the whole mix in this way, and recording each channel as a separate take on the same DAW track, just as you would when overdubbing vocals. But as you've already taken care of most of the hard work in the DAW, you're left with a less hectic task on the board; and because you're recording multitrack rather than summing, you'll still be able to correct any mistakes (though there's no reason why you can't capture any effects returns or bus outputs at the same time if you have sufficient I/O.)
If I perform the same mix five times in a row, I never get the same result. One mix will usually 'just work'. A trick I've discovered is to change the monitoring level for different passes: I tend to react differently to the difference in level, and thus mix slightly differently. It might affect overall levels, how I choose to pan a guitar part, how much reverb or delay I want to hear, or how much I want to raise the wet part of a parallel compressor group, for example. By recording several passes (a few at normal listening volume to get into the groove, and a couple more with the monitoring level higher or lower), I get the opportunity to pick my favourite mix, or indeed to comp the best one. This approach really can give you the best of both worlds: you get the joy and buzz of the performance side of mixing, and you get the convenience and recall (up to a point), of your DAW.
I enjoy working in this way very much, but there's a significant — and potentially show‑stopping — downside. If the analogue mix is using group or mix-bus compression, then several tracks may be feeding into these devices and collectively dictating how they operate on the program material. This is one aspect of the mix you really can't recall, whether you have the multitracks or not. So it's great if mixing for yourself; less good when you need to address a client's request for "a few final tweaks”.
The simplest way of introducing some hardware into your system, which is use a processor at the record stage, patched in between your source (mic preamp, guitar or other instrument) and the input to your audio interface. If you're using an outboard preamp, then there's little more to say — you take a line output from that to your processor of choice, and another from the processor to your audio interface input.If the mic amps are built into your audio interface, then you'll need to connect the hardware via the interface insert points (on those models that have them — and most don't) or you'll need a separate mic amp feeding the hardware unit which in turn links to a line input on the interface. Where your audio interface has additional inputs and outputs, you may be able to route audio through your external hardware box using these connection combined with the routing ability of your DAW software, but this is unnecessarily complicated at this stage and may throw up monitoring latency issues — and, as we shall see later, this type of routing arrangement is arguably better suited to mixing.
The kind of hardware you might connect in this way would typically be a compressor, equaliser or guitar recording preamp, but with so many software alternatives, why might you want to do this? Back in the days of analogue tape where the noise floor was something to be feared, compressing your audio as you recorded it made the best use of the available headroom by keeping the average signal level sensibly high. If you compressed after recording to tape, any tape noise would be emphasised by the compression as every dB of gain reduction equates to a dB increase in any noise due to the audio source.
Now that we have 24‑bit recording, we can record without compression, leaving 12dB or so of headroom, and still end up with a very quiet recording so input compression is no longer needed as a means of keeping noise at bay. However, if you have a hardware compressor that has a sonic character that just can't quite be replicated using a software plug‑in, then there's no reason not to record through it to capture that character. The only caveat is that as compression is much easier to apply than to undo, it is best to err on the side of using less compression than you need rather than more. You can always even up the level when you mix using either additional compression, mix volume automation or a combination of the two.
A similar argument applies to equalisation — if you have a great‑sounding equaliser or have a front end with EQ built in, then by all means record through it as long as you bear in mind that your equalised sound may sound different when heard in the context of the rest of the mix. For this reason, some recording experience is beneficial so that you can second‑guess what type and amount of EQ will work in the finished track. As with compression, EQ is harder to undo than to apply and you can always add more as you mix, so don't overdo it.
The case for recording via a modelling guitar preamp or similar is more straightforward as guitarists need to hear something close to their final sound in order to be able to play appropriately. Again there are software alternatives so ultimately it comes down to artistic choice. If it sounds right and you feel comfortable playing through it, then it is right. Depending on your audio interface, you may be able to set up zero latency monitoring, and though some software such as the Line 6 Pod Farm also offers an essentially delay‑free monitoring mode, most is subject to the latency determined by the buffer size set up in your audio driver. Even a small amount of monitoring latency can upset some players so this is definitely a factor to take into account when deciding how to work. Paul White
Although latency can make life hard when you're recording, or playing virtual instruments, it's normally much less of an issue when mixing: you're able to set your audio interface's buffer size to maximum to reduce the burden on your computer's processor — enabling you to run more plug‑ins in the mix — at the expense of increased latency.
Most modern DAW programs now include ADC, or Automatic Delay Compensation, whereby, in essence, the software calculates how much time a plug‑in takes to process the audio that passes through it and adds that amount of time to the other channels. Thus even if you're using a convolution plug‑in, or something running on a DSP platform such as UAD, with inherent latency, the computer nudges all the other tracks accordingly. (The notable exceptions here are Pro Tools LE and Pro Tools M‑Powered, neither of which feature ADC — you have to calculate and make adjustments manually).
In software-only setups, this usually works flawlessly: the plug‑in 'declares' its latency to the host software, and the host makes calculations based on that. When mixing using automation, this setup poses no problems: the lag in tweaking some plug‑in controls and hearing the results can be a pain if the buffer value is set too high — but it's still nothing like the show‑stopping problem it can be when tracking or trying to play virtual instruments in real time.
"What has all this to do with hybrid systems?”, I hear you ask. Well, although generation loss from A‑D/D‑A conversion is not such an issue these days, each round trip from the digital to the analogue domain adds latency. But all your software knows is that it's sending audio out and back in: it has no way of knowing automatically how long the round trip takes, and therefore the prospect of automatic delay compensation becomes more challenging. Again, thankfully, most DAW software now includes a means of measuring the time it takes for the audio to make a round trip out of the DAW, through a piece of gear and back. All you need to is set up the 'hardware plug‑in' (see box earlier in this article) and 'ping' a signal through it: your DAW should calculate the latency automatically.
As I've mentioned in the main text, most audio interfaces that offer multi‑channel analogue audio I/O also provide several channels of digital I/O, most frequently via ADAT optical ports, S/PDIF phono connectors, or AES/EBU on XLRs. For example, my RME Fireface 800 gives me 10 analogue ins and outs, up to 16 channels of ADAT I/O (over two physical sets of ADAT connectors), and a S/PDIF phono port. There's no dedicated AES/EBU port, but I've found it perfectly possible to send a signal out of the S/PDIF port over a short phono‑to‑XLR adaptor to feed an AES/EBU monitor controller. If you use a PCI- or PCIe‑based system that enables you to add multiple breakout boxes, you might have many such digital expansion options available.
So how do you make use of all this digital connectivity? There are a now a number of devices you can buy that take advantage of it, allowing you to increase the analogue I/O count significantly. However, some of these are better suited to recording applications than to mixing using line‑level signals, so it pays to think about which option you go for.
A standard, modern ADAT port will transmit eight channels at up to 24‑bit, 48kHz, or half that number at double the sample rate. The S/MUX protocol allows you to split eight channels at 96kHz (or four channels of 192kHz) across two sets of ADAT connectors if your interface offers them. It's probably the cheapest and most convenient way of adding multiple analogue I/O. There are many products that offer a bank of mic preamps with an ADAT output — the Focusrite Octopre MkII, the Behringer ADA8000, the Presonus Digimax range, and the Audient ASP008, to name but a few — but while these might appeal, don't forget that the aim here is to add both inputs and outputs, so you may not need all those preamps. Moreover, if you're connecting to a patchbay, you'll probably prefer to have all the line I/O on the rear panel, to keep things neat, tidy and manageable in your studio.
Not all of the ADAT expanders offer both in and out: the Octopre MkII, for example, only gives you additional inputs, whereas the MkII dynamic gives you both in and out. Interestingly, there aren't actually that many line‑level converters around without added features like mic preamps. In fact, the only dedicated line I/O devices that seem to me to fit a project-studio budget are the Sonic Core A16 Ultra, which gives you 16 line‑level ins and outs in a 1U rack (this is the unit I chose to expand my own setup), and various eight‑channel devices made by Alesis.
Further up the quality mountain, you could consider something like the RME ADI‑8 DS, or perhaps the Lynx Aurora, which uses the AES/EBU protocol by default, but also has an ADAT card as a cost option. Given the price of some of these devices, though, you might even consider purchasing a 24‑track digital hard disk recorder such as the Alesis HD24 (or its better‑quality HD24 XR variant) and similar models by Fostex, Tascam and others: as well as offering you a means of tracking 24 analogue channels to a hard drive, these can double up neatly as a set of 24 ADAT-format A‑D/D‑A converters. One of these recorders paired with something like the M-Audio ProFire LightBridge or PreSonus Firestudio LighPipe ADAT interfaces could give you all the I/O you need.
There are fewer stereo S/PDIF converters, although plenty of hardware processors offer a digital output via S/PDIF. Although I opted to use my S/PDIF connection to give me a stereo feed from my DAW to a monitor controller, I've also used it in the past to feed a digital effects unit, and on occasion have even pressed that unit into service as a stereo converter, with the effects bypassed, where I needed to add one more pair of ins and outs.
If you prefer to sum your mix in the analogue domain, there's a huge range of analogue mix solutions available, from full, large‑format traditional consoles down to passive summing boxes, and everything in between. What you use depends on how you like to work, as well as the budget. Technically, dedicated summing boxes have a significant advantage over most traditional consoles, in that the actual physical summing buses are very short and therefore less prone to introducing noise. Some devices, like the Thermionic Culture Fat Bustard, major on their use of transformers, valves and so on to impart 'traditional' colour, while others, like the Audient Sumo, major on how 'clean' they are... and then there are entirely passive devices, my favourite of which is the Rolls passive mixer, which is just a bunch of resistors and switches in a box. It needs around 40dB of external preamp gain at the end to restore line levels (but it can cope easily with maxed‑out D‑A outputs, which is more than many active units can do). The passive mix idea is a modern version of the passive mixing that was standard in the 1950s and '60s before virtual‑earth mix buses became practical with transistor and IC circuitry. But you can at least choose the flavour of preamp gain for the desired effect — valve or solid‑state, coloured or squeaky clean!
But mixing a dynamic track on the small rotary controls of a rackmount summing mixer isn't much fun, and many people then resort to programming the mixing into the DAW automation... which seems to me largely to defeat the object.
Comparing the relative merits of different analogue summing devices is pretty easy — you can compare the technical specs and use your ears to judge what coloration is being added. And, arguably, if you have to spend ages deciding what sounds better, it's not going to matter what choice you make! However, the arguments over the relative advantages of analogue and digital summing have raged for years, and no doubt will continue to do so for many more, probably regardless of the actual technical truths involved. Theoretically, digital summing should be absolutely perfect, and we already know that analogue summing is inherently flawed with noise and distortions being endemic to the process. For example, the more channels you sum in the analogue domain, the more noise is present in the end result. But maybe some people like those analogue flaws; perhaps they're an element of what we've come to regard as a musical sound.
"But digital summing isn't perfect, either,” you say? Let's consider peoples' differing perceptions of digital summing. Certainly, it has not always been performed perfectly. The earliest DAWs were little more than audio recorders and it just made sense to mix on an analogue console. Early efforts at keeping everything, including the mixing, in the box also suffered at the hands of some dubious DSP code writing and poor recording technique — for example, the still‑prevalent idea that individual tracks must be banging up against 0dBFS to maximise 'digital resolution', which simply isn't true. Mix 24 maxed‑out signals together and the result is way over 0dBFS, and while floating‑point processing should be able to deal with that, in practice not all systems have behaved quite as well as desired, and especially not all plug‑ins and other signal-processing apps. Although rare these days, a decade or more ago these issues undoubtedly caused problems in some cases, resulting in audibly inferior results.
However, there's now plenty of evidence that, when done right, digital summing works and is acceptable: every feature film is mixed on digital consoles, often combining several hundred individual sound elements in several mix passes; and most major TV shows are now mixed on digital consoles, as are many big live concerts and stage shows. Yes, there are practical and convenience benefits to the use of digital consoles in these situations, but if the quality wasn't up to the job it wouldn't get used — especially in the film world, which is arguably the last bastion of dynamic range and ultimate audio quality throughout the complete chain.
For me, mixing ITB works perfectly well provided it is approached rationally, with source tracks having sensible headroom margins, and with modern, competently written DAW software. But I won't deny that the mathematical precision of ITB mixing can leave something to be desired in some situations: sometimes a little familiar analogue colour makes all the difference. To this end, I've often found that routing the final ITB mix through a nice analogue outboard device — something with well-designed analogue stages and transformers, even with processing stages bypassed — can add just the right amount of colour needed to create a certain character in the end result. This approach has the practical advantage that it requires only a stereo D‑A and A‑D converter.
What bothers me is all the shouting that analogue summing is 'better', because technically that's just not the case: perhaps "psychologically more pleasing sometimes” would be a more accurate form of words. There's no denying that some people find mixing on a traditional console easier (which in turn makes them happier with the results).
But let's consider why some people prefer the sound of analogue summing, and some of the implications of going down that route. Full analogue mixing requires a D‑A converter for every source channel in the outboard mix, as well as a stereo A‑D to re‑digitise the resulting mix — plus the analogue mixing system itself, of course. Clearly, though, this is a far more expensive option than the stereo treatment described above, and is also impractical for many people when working on big mixes.
That's why a hybrid solution is often employed, whereby source tracks are mixed to stems inside the box, and then combined outside in an analogue mixer or summing box. If the aim is purely to sum those stems, this makes little sense to me: either digital mixing is horrid, in which case I can't think why you'd compromise the mix in this way, or it is acceptable — in which case don't bother with the analogue mix. All you need is to use a stereo analogue processor to add final analogue flavour, as described above.
Perhaps the people who advocate analogue summing aren't actually talking about summing at all, but adding coloration. Most converters these days are designed to be clean‑sounding or 'transparent', but some are deliberately designed to impart a subtle flavour each time audio is passed through them. The same can be said of any analogue gear with transformers, valves or other physical circuitry — and there's certainly a potential argument in favour of treating different stems with such devices, which is what makes products like the Neve 8816 and the Thermionic Culture Fat Bustard appeal. However, I don't see that there's really much difference between summing the resulting signals in the analogue or digital domains. Hugh Robjohns