By Craig Anderton
Computers can be pretty smart, but when it comes to sound, they need a translator that can convert sound waves into the digital language of ones and zeroes. Audio interfaces provide this translation, and you'll need one if you want to get into computer-based recording. But if you bought an interface several years ago, it might be time to upgrade. Like everything else, interfaces continue to evolve, and they can definitely affect overall sound quality.
Most stock computers include sound capabilities, but these tend to be consumer-grade (Macs with digital ins and outs are the exception). Also, built-in I/O is typically limited to 16-bit operation at sample rates of 48 kHz or less. If you're serious about your audio, forget about any built-in sound capabilities, and go for a pro-level solution.
However, the growing sophistication of audio interfaces has led to an option overload. Choices are nice, but it becomes hard to separate what manufacturers want to sell you as opposed to what you really need. So, we'll explain the concepts involved in audio interfaces so that you can become educated enough to make an informed choice.
There are three main ways for audio interface to connect to your computer:
PCI card. This plugs into a PCI slot on your computer's motherboard. It may place all its connectors on the backplate, a second backplate that installs next to the board but doesn't take up a slot, or have a cable that goes to a separate (usually rack-mountable) breakout box that's festooned with connectors.
USB. This type of interface is a separate box that plugs into a computer's USB port. Early USB interfaces got a bad rap because of problems in computer operating systems; however, starting with Windows 98SE and MacOS 9.1, USB functionality was decent and starting with Windows XP and Mac OS X, USB became fully integrated into the operating system.
This is the type of USB connector you find on a USB host.
FireWire. This is similar conceptually to a USB interface as it plugs into a built-in computer port, but runs much faster than USB 1.1 interfaces. Compared to USB 2.0, FireWire runs at about the same speed, but puts far less overhead on your computer's CDPU because it offloads a lot of tasks to its own dedicated hardware.
This picture shows 6-pin FireWire ports, which supply power to peripherals. Many laptops have 4-pin connectors, which are identical except that they don't provide power - any peripheral wil need its own AC adapter.
A new type of interface, which pretty much started with the TASCAM US-428, adds more functionality (e.g., mic preamps, a mixer with faders, signal routing, etc.) to a basic interface. These connect via USB or FireWire, and are generally intended for smaller studios and portable applications. However, as these units have become more sophisticated, they've blurred the line between audio interface, digital mixer, and human interface. Digidesign's FireWire-compatible 002, which incorporates audio/MIDI I/O and is optimized for use with Pro Tools LE software, was an early example of this trend.
The big issue here is dedicated vs. general-purpose interfaces. Dedicated types fit specific software like a glove. General-purpose systems need templates for the software you want to use, but should you switch programs or platforms, your investment may still be protected.
The other issue is modularity vs. a self-contained solution. If a unit contains I/O and control, will it accommodate an upgrade to higher sample rates or bit resolution? Will the human interface section also control other programs, such as soft synths? A modular system made up of several different units (one for I/O, one for human interfacing, etc.) is more flexible, but requires more savvy to put together, and there is no guarantee that the various elements will work together seamlessly. Self-contained solutions are more foolproof, but usually less expandable and compatible with fewer products.
Perhaps more than any other feature, this will drive your selection. The first question is whether you need analog, digital, or both types of I/O, and if digital, which flavors of digital.
If your studio setup uses a digital mixer or has outboard A/D converters, then you don't need analog I/O: you can feed analog signals into these outboard units, whose digital outs go into a digital-only card. This has two advantages:
Analog I/O choices. I used to avoid sound cards with onboard analog I/O, but reviewing the Lynx Studio series of cards changed my mind -- it is possible to get good audio performance from a card sitting inside a computer, although this requires rigorous board layout and shielding. Most pro-level cards with analog I/O use external breakout boxes that keep the audio circuitry well away from the computer's EMI (electro-magnetic interference).
Analog I/O can be either balanced or unbalanced. Balanced inputs can minimize electrical interference induced into a cable, which is important with "noisy" digital studios that are prone to interference from monitors and computers. Balanced ins are popular in pro installations that use long cables, which are more prone to picking up interference than short cables. Unbalanced inputs are usually fine if cables connected to the card are under 6-12 feet or so.
Balanced line systems use either XLR or TRS (tip-ring-sleeve) 1/4" stereo connectors. (Even though TRS types use stereo connectors, they still carry only a mono signal. The tip carries the signal's "hot" line, and the ring carries the out-of-phase version required for balanced line operation.) However, plugging into a TRS jack with a mono plug will convert it to unbalanced operation. Unbalanced lines use either standard 1/4" mono phone jacks or, less commonly, RCA phono jacks.
Also check if the analog system includes mic preamps, and whether there's phantom power available (required by some microphones) should you need this. For convenience, having built-in mic preamps is useful, but for ultimate performance, you will likely want to hand-pick the mic pres you like best, and feed them into line-level analog ins (or if they have digital outs, into the digital inputs).
Digital I/O. There are five popular types of digital I/O protocols:
The ADAT interface uses optical connectors that look the same as optical S/PDIF connectors.
Of course, what you'll need depends on the other gear in your system. In general, I like ADAT I/O because so much gear is compatible with this standard, and putting 8 channels on a thin, fiber-optic cable is a plus. But you'll probably also need SPDIF and AES/EBU to handle the occasional piece of outboard gear.
Your software determines which drivers are essential; for example, Steinberg Cubase is optimized for use with ASIO drivers on Windows and Core Audio on the Mac, while Cakewalk Sonar is compatible with ASIO and kernel-streaming WDM drivers (note that not all WDM-compatible programs offer true high-speed kerner streaming; some just use a WDM "wrapper" around an older standard for compatibility). Windows systems also usually include "old school" drivers which are compatible with just about anything, but offer lower performance than newer options.
ASIO, developed by Steinberg but embraced by dozens of manufacturers, provided a pro audio driver solution before Apple and Microsoft realized that low-latency audio drivers might be a good idea. Although some feel that ASIO will be eventually eclipsed by Apple's and Microsoft's standard protocols, ASIO is well-entrenched and is also a true cross-platform solution. It's not going anywhere for quite a while.
As drivers are often updated, check the sound card manufacturer's web site for details, as updates can offer significant improvements. Also note that some companies do clever variations on a theme, such as developing high-performance MME-compatible drivers that retain the compatibility of MME, but offer lower latency. Finally, not all drivers are created equal: a well-written driver with a well-designed sound card will outperform combinations that aren't as meticulously designed.
Also, web sites for software manufacturers may include lists of recommended cards that are known to work with their software, so this is a good place to start your research. For example, it's no accident that people who use MOTU's Digital Performer often use MOTU interfaces; it's a no-brainer that they work perfectly together.
The ability to handle high bit resolutions and sampling rates is a function of both hardware and software, but even if the software can handle it, that means nothing unless the hardware can too. Do you need 96kHz or 192kHz sampling? Higher sample rates will cost you more, but you may need them to remain commercially competitive, or if the rest of your signal chain is of sufficient quality that using lower sample rates will degrade the overall fidelity.
As to bit resolution, the general consensus is that it's well worth running at 24-bit resolution, even if your final delivery medium is a standard 16-bit Red Book CD. As a result, most modern analog interfaces use 24-bit converters, and digital interfaces can pass audio with 24-bit resolution.
There are other features to consider in professional contexts. SMPTE sync is built in to some interfaces, which is important if you need to sync with other SMPTE-compatible devices (not a given in audio-only studios, but pretty much a necessity for those involved in post-production).
Another option that may be important is word clock input and output. Word clock is a sync signal, identical to the sample rate, that insures digital audio signals are all clocked at the same rate. When connecting two pieces of gear together, one device's input simply follows the clock signal generated by the other piece of gear. However, with multiple pieces of equipment, you want all digital audio to sync to a common clock signal. A word clock out can provide this common signal, while a word clock input allows synching to a word clock source.
Systems that are based exclusively on the ADAT optical interface don't require word clock, as clocking is part of the ADAT interface. Still, it's good to be prepared.
Also check out any "special sauce" features. For example, some manufacturers include onboard DSP that raises the price, but can pay for itself in convenience. For example, CreamWare's Pulsar boards pioneered using DSP to run virtual instruments and processors, independent of the host processor; MOTU's 2408 Mk III uses DSP to provide the functionality of a digital mixer, thus allowing for zero-latency monitoring (otherwise, monitoring through the sound card/computer chain results in some amount of delay).
In any event, audio interfaces have made dramatic progress over the years, from purveyors of sounds for games to full-blown, high fidelity devices. The bottom line is this: An audio interface is as crucial a component as a microphone, as it links the outside world to your computer -- choose it with care.
Craig Anderton is Editor Emeritus of Harmony Central. He has played on, mixed, or produced over 20 major label releases (as well as mastered over a hundred tracks for various musicians), and written over a thousand articles for magazines like Guitar Player, Keyboard, Sound on Sound (UK), and Sound + Recording (Germany). He has also lectured on technology and the arts in 38 states, 10 countries, and three languages.