Continuing his series on setting up a recording environment based around a Mac or PC computer, Paul White goes through the hardware and software options for recording audio alongside your MIDI tracks. This is the second article in a three‑part series.
Last month I covered basic MIDI sequencing systems, which don't place heavy demands on your computer's CPU or require much in the way of additional hardware. Life gets more complicated, however, when you want to add audio. Not only do you have to budget for a faster computer, you also have to make decisions concerning the audio interface or soundcard that you'll be using. For example, should it be 16‑bit or 24‑bit, do you need multiple ins and outs or will stereo suffice, do you need digital I/O, and if so what kind?
Many systems are based around a simple audio interface that effectively adds A‑D and D‑A converters, but no additional processing hardware. Such setups (which are often referred to as host‑based or native systems) leave the computer to do all the hard work involved with audio routing, recording, playback and mixing. Powerful host computers may also be able to run software‑based 'virtual' effects and 'virtual' instruments simultaneously. Though modern computers are fast and powerful, they have their limits, so at some stage you're going to run out of processing power. In practice, this may limit the number of audio tracks you can record or play back simultaneously and/or limit the number of virtual effects you can use at one time. If you're using virtual instruments, the general rule is that the more voices you play at once, the more processor power is used. If you're realistic with your expectations, a native system can function very effectively, but there's another route if you need more power.
The alternative is to use audio hardware that includes its own DSPs (Digital Signal Processors) for handling things like digital effects and mixing. The most famous of these systems on the Mac is probably Digidesign's Pro Tools, while Soundscape offers similar functionality on the PC. At a lower price level, the Yamaha SW1000XG card can provide up to 12 tracks of audio recording and mixing, multi‑effects, and a very respectable GM/XG synth without taxing the host processor. Other DSP‑based systems include Creamware's Pulsar and Powersampler cards and Lexicon's Studio/Core II package, with its optional MP100 expansion effects board. One point to bear in mind early on, however, is that none of the systems that use their own DSP to power effect plug‑ins currently allow you to use VST‑format plug‑ins — they tend to have their own proprietary format of plug‑in designed to work on their specific DSP architecture.
On the plus side, the more serious DSP‑assisted audio systems are generally free of latency (the delay between feeding in a sound and hearing the same sound come out over the monitors after being routed via the computer's CPU), and they don't tax your host computer in the same way that native systems do, leaving more of its processing power free for other purposes. High‑end systems such as Soundscape and Pro Tools also tend to be more stable than native systems.
Driving Hard
The EIDE hard drives used in standard Macs and PCs are becoming fast enough to record and play back a good number of audio tracks — see, for instance, Martin Walker's review of Millennium Music's custom‑built PC in this issue of SOS — but it's important to understand that they are controlled via the host processor, so they tend to slow down if there are a lot of edits or if the audio is fragmented. For more serious work, a SCSI drive is recommended, and with the new Macs and any PC, this means fitting a SCSI interface card. There are several incarnations of SCSI that have different speed capabilities, but if you're setting up a system specifically for audio work, you need a separate Ultra Wide SCSI‑2 drive with a spindle speed of at least 7200 rpm. This will ensure that you have an adequate number of tracks, even when the going gets tough, and because SCSI has its own sub‑processor to handle data transfer to and from the disk, you don't waste any of the power of your host processor.
RAM & Resolution
At one time, RAM memory cost more than the computer it was being plugged into, but these days the prices are a little more sensible. Most serious audio applications need at least 128Mb of RAM memory to run efficiently and, if you're thinking of using a software sampler that uses the host computer's RAM, you'll need as much as you can possibly afford. As a rule of thumb, see what your sequencer software manufacturer recommends as a sensible amount of RAM, then double it!
Most audio hardware now works at both 16‑ and 24‑bit resolution, so you can choose which one is best for your projects. However, some hardware is limited to a sampling rate of 48kHz, while other gear supports 96kHz sample rates. My own view is that 96kHz sample rates only provide a sonic advantage in a world‑class studio environment, so in a home or project studio any sonic gain will not be worth the bother of having your processor power and disk space eaten up exactly twice as quickly as with 48kHz audio. Working at 24‑bit/44.1 or 48kHz seems a sensible compromise when working with acoustic instruments or music that has a wide dynamic range, while 16‑bit/44.1kHz is perfectly adequate for most pop or dance music, which has a limited dynamic range. There's more on this subject in the Making A Difference feature.
Basic System
Figure 1: A basic setup based around a computer with simple stereo audio I/O.The most basic systems you can use for audio recording employ the stereo analogue ins and outs of a games‑style soundcard, or the built‑in Mac audio hardware, as shown in Figure 1. The sound quality depends on the spec of the card, but you don't have to spend a fortune to get something that sounds respectable. The audio output will be mixed with the sound of any MIDI instruments generated by the computer soundcard, so if you don't have any external synths, you can get by without a mixer. Nevertheless, it's unlikely that you'd want to record via the microphone input of a typical soundcard as the quality of these is invariably poor, and the level control is usually handled by software after the input has been digitised. If your mic signal is too low, boosting it via a software gain control won't improve your signal‑to‑noise ratio.
Essentially, you need an external piece of kit that can amplify your mic signal to line level before it reaches the soundcard. You could use a mixer or a stand‑alone preamp, but the best choice is often a dedicated voice channel — a piece of outboard gear that combines a mic preamp with a compressor and perhaps an equaliser too. Using a voice channel with one of the excellent low‑cost capacitor mics currently available, you should have no problem making clean recordings. Most voice channels also provide line‑ and instrument‑level inputs for recording other sources. The system shown in Figure 1 includes a voice channel and microphone.
A system of this type is clearly restricted to recording no more than two tracks at a time, so it's best suited to the person who works with both MIDI and audio and who builds up the audio tracks one or two at a time. Because the output is stereo, any mixing must take place within the computer, which means effects must either be added during recording or created within the computer using plug‑in effects (or hardware effects if your soundcard provides them). If you add effects to the signal as it comes out of the soundcard, the same effect will be added to everything in the mix, which isn't usually what you want.
Note that if you're working without a mixer, you won't be able to use the anti‑latency dodge of monitoring via the mixer while you overdub (see SOS August '99) rather than monitoring everything at the soundcard output. This could be a problem for those using cheap consumer games‑style soundcards, because these often don't have ASIO or other low‑latency drivers.
Upgrade Path
Providing you have a suitably fast computer, you can upgrade a system like this by adding extra VST plug‑ins to enhance your virtual mixing environment, and you may also want to load in some virtual instruments or samplers. You can do all these things without upgrading your audio I/O providing your computer is powerful enough to do all the things you want it to. Additionally, you may wish to add external hardware synths, which means you'll also need a mixer as described in last month's section on MIDI sequencers.
If you do have problems with latency, you could either add an external mixer and monitor overdubs via the mixer while muting them on the soundcard, or upgrade your soundcard to one with low‑latency ASIO or other drivers. If both your card and software support ASIO II drivers, you can use the option of 'through monitoring' where the signal being overdubbed is sent directly to the soundcard output rather than being routed via the CPU. This is just like monitoring overdubs via an external mixer. However, one drawback of both approaches is that you don't have the opportunity to add native effects to the monitoring signal as you record (hearing your voice with reverb can, for example, sometimes help you sing better), but at least you get to hear your overdubs in time.
Multiple‑Output Systems
Figure 2: A more complex setup using a multiple‑output soundcard and an external mixer, allowing audio submixes to be processed using hardware effects units and processors.You'll also need an external mixer if you wish to use a soundcard that has multiple analogue outputs, as illustrated in Figure 2. Note that although the example shows a soundcard‑based system, the diagram applies equally to a system that uses an external interface box. The same connection regime is also applicable to systems that have their own DSP‑powered effects. Although many users would be using external MIDI modules such as synths, samplers and drum machines, these have been omitted to avoid over-complicating the diagram. The MIDI setup diagrams shown last month in Part 1 illustrate how these MIDI instruments can be connected.
Though you can do all your mixing inside the computer, it can be advantageous to have multiple outputs on your soundcard and thereby keep some of the signals separate. This allows you to use the hardware mixer EQ rather than the digital EQ within your software (presuming the analogue EQ sounds better) and, perhaps more importantly, it allows you to use external effects and other outboard processing. Good‑quality native reverb plug‑ins tend to be very power‑hungry, yet you can buy a dedicated hardware reverb unit for little more than the cost of a software plug‑in and it won't use up any of your CPU power. Similarly, if you already own multi‑effects units, compressors, gates or enhancers, it will provide a means of using these effectively. Some users find it much easier to mix using the physical controls of an external mixer rather than making all adjustments with a computer mouse, though the advantage of doing at least some of the mixing inside the computer is that most software packages allow the mix levels and other parameters to be automated.
Most multiple‑output soundcards offer no more than eight outputs (often with two inputs plus S/PDIF digital I/O), though some such cards can be used in pairs to double up on the I/O capability of a system. Practically, this means that unless you limit your number of audio tracks, you still won't be able to allocate one physical output to each track, so some internal mixing will still be required. In most instances, this won't be a problem as most sounds fall into logical subgroups that can be mixed inside the computer and then treated together outside it; and when different sounds within a subgroup do require different treatments, you can still use your plug‑ins at the virtual mixing stage. For example, suppose you have a percussion submix that requires a general short reverb, but more reverb on one particular sound. You might treat this sound with a basic (less power‑hungry) reverb plug‑in within the computer, then apply a better quality reverb from a hardware unit to the overall percussion mix. You may decide to allocate your main vocal and main lead instrument outputs of their own to give you complete freedom over their treatments, but things like backing vocals, pad keyboards and percussion can often be submixed without sacrificing flexibility.
A system like the one just described is still best suited to those who build up their compositions one track at a time, but who demand more flexibility when mixing. These users may already have a mixer and outboard effects which they wish to continue using after upgrading to a desktop computer‑based system. If the requirements are more ambitious and you need either to record multiple inputs at the same time, or to integrate a digital mixer into the system, the choice of audio interface will be quite different, and next month I'll be looking at some of these more sophisticated systems.
Noise Annoys
Some computer components, especially fans and drives, are physically noisy, while conventional monitors also generate a significant amount of electromagnetic radiation that will interfere with guitar pickups if you get too close. Ideally, you should record in a separate room to your computer, but unless your system is particularly noisy, you'll probably get away with working at the opposite end of the room. For critical work, consider a sound‑deadening computer cabinet (see the feature on reducing computer noise in SOS January 2000), and if you do a lot of work with electric guitars, either use humbucking pickups or budget for a flat‑screen LCD monitor. Most MIDI + Audio sequencers need a 17‑inch monitor or larger, though a 15‑inch flat‑screen monitor offers almost as much viewing area.
Digital I/O
In addition to analogue inputs and outputs, some soundcards also provide digital connections. At the budget end, this may simply be a co‑axial S/PDIF stereo output, but for a little more money you can buy a card that has an S/PDIF input as well. S/PDIF I/O is particularly useful if you're mastering directly from your soundcard to a DAT machine, or if you have an external digital mixer with an S/PDIF input. By the same token, an S/PDIF input will enable you to load in material from DAT, MD or CD‑R for further editing, providing your source machine has an S/PDIF output. S/PDIF digital audio can be sent down either co‑axial or optical cable, but inexpensive converters are available for changing optical S/PDIF into co‑axial S/PDIF and vice versa.
Choosing A Computer Platform
If you haven't yet decided on which computer system to use, you'll need to choose between Macintosh or PC as they're the only two options really worth considering for serious audio work. But even before making that choice, you need to look at what software you want to run. If you need to do a specific task for which only one piece of software is appropriate, and if that software runs on only one of the two major platforms, then your choice is made for you. For example, if you've just got to have the PC sequencer Cakewalk, then you're going to have to go the PC route, and conversely, those who can't do without Digital Performer will need a Mac to run it on.
The remaining top sequencing packages — Logic Audio, Cubase VST, and Studio Vision — are all available in Mac and PC versions. However, cross‑platform availability tends to be less common with other audio software, such as audio editing packages: Wavelab, Sound Forge and Cool Edit Pro are all PC‑only, while Peak and Spark are currently Mac‑only.
Macs tend to be easier to set up and manage than PCs and audio latency seems less of a problem, all other factors being equal. Macs also have basic 16‑bit stereo audio ins and outs as part of the basic hardware configuration whereas a PC requires a soundcard or other audio interface. The down side to Macs is that they are more expensive than their PC counterparts (but bear in mind that Mac processors are much more powerful than Pentium chips of the same clock speed) and they have fewer expansion slots for adding things like audio cards, SCSI interfaces, second monitors and so on. Most new models have only three available slots while the iMac doesn't have any at all.
New Macs are also less easy to upgrade than PCs. SCSI, which used to be standard on all beige Macs, is no longer fitted as standard and the serial ports, which were once used for connecting printers, MIDI interfaces and modems are now replaced by a USB (Universal Serial bus) connector. Third‑party adaptors are available to provide old‑style serial ports on a new Mac, but unless you have a lot of legacy hardware to accommodate, you're better off buying USB‑compatible hardware and software with USB dongles where applicable.
PCs are relatively cheap, they can be pulled apart and upgraded as necessary and peripherals tend to be cheaper than for Macs. They also tend to have more free slots than Macs and most new models have USB as well as legacy PC connectors. You need to buy a separate soundcard to get audio in and out of the computer (though most bundles come with a basic audio card) and because of the countless hardware permutations that can be used in the average PC, you may experience problems when using audio. That's one reason we always recommend that you buy a ready‑configured system from a specialist retailer, even if it costs a little more.