RAM | Reseller Advocate Magazine: Issue 62

By Chris Angelini

THERE ONCE WAS A TIME WHEN AN ADD-IN audio board meant the difference between a system capable of distinct "voices" and the mere beeps and buzzes of the PC speaker. Since then, we've seen audio evolve from drop-in ISA cards capable of playing back nine simultaneous sounds to today's PCI cards with highly programmable special effects and near-audiophile quality. But even though we're way past the days when discrete audio processors either made or broke a machine's capacity for multimedia functionality, there are still plenty of customers under the impression that a drop-in board is the only way to go for great sound.

The wild world of audio is so confusing in part because it's distinctly subjective. You can't offer side-by-side comparisons pitting sound systems against each other as you would graphics cards. There are no compelling performance charts to show how much better one solution runs versus another. Moreover, the rise of integration has displaced the popularity of discrete sound cards. So what gives? When is a motherboard-based solution better than a dedicated sound processor? And when does it make sense to upsell one of the few discrete cards still available?

Fortunately, understanding fundamental audio attributes and the features included with today's most popular components gives you at least a bit of insight as to where each solution excels. Don't think that your customer's questions about sound are all answerable by hardware, either. Especially with the most popular integrated audio products, software plays a defining role in establishing certain features. Remember that Vista is right around the corner, too. Microsoft's next-generation operating system will change the way audio hardware works in a big way.

Getting Those Specs Right

Depending on your familiarity with audio hardware, making sense out of sound card specs can be extremely confusing. Not only are the numbers themselves overwhelming, but vendors use different scales to measure the performance of their products.

Flip over the nearest sound card box and you're likely to see line after line of specifications that read something along the lines of: 24-bit conversion of analog sources at 96 kHz sampling or 24-bit conversion of digital stereo sources at 192 kHz and signal-to-noise ratio (SNR) of 109 dB plus .004% total harmonic distortion. To the uninitiated, it's line after line of nonsensical gibberish.

But each of those ratings helps define the measured internal precision and performance of a given audio solution. Of course, none of that might matter to the business desktop user playing back an occasional MP3 or the home user ripping and burning CDs. Those folks are probably more concerned about inputs and outputs. Gamers can tell a difference, though. So can the audio enthusiasts who pay extra for dedicated sound processors.

Size Doesn't Matter
Codecs such as VIA's VT1708 connect to an HD Audio-compliant southbridge facilitating conversions between analog and digital inputs and outputs. The VT1708 meets the HD Audio standard with DACs that operate at a 100 dB SNR.

In sound, the bit depth numbers you see represent the potential accuracy a piece of hardware or software is capable of achieving, similar to graphics. The spec comes into play every time a digital source is converted to analog and vice versa. Playing music, recording, and gaming are all applications subject to a given sound solution's bit depth. A processor capable of 16-bit conversion is able to represent a given audio stream using 65,536 levels. That sounds like a lot until you step up to 24-bit processing, represented by 16,777,216 levels, delivering a much richer result. Whenever possible, opting for products with higher bit-depth is better.

You'll usually see a sampling rate figure trailing bit-rate—something along the lines of 44.1 kHz, 96 kHz, or 192 kHz. The two are interrelated and consequently represented in the same specification. That rate indicates the number of times per second a given waveform is translated into digital audio. You're probably already familiar with some of the more common sampling rates. Telephones sample at 8 kHz, for instance. CDs are sampled at 44.1 kHz. The latest high-def audio technologies, such as DVD-Audio, HD DVD, and Blu-ray Discs, are all sampled at either 96 kHz or 192 kHz. Once again, shooting for PC audio solutions capable of 96/192 kHz sampling will give your customers the best possible quality, at least in theory. Depending on the exact implementation, your mileage may vary.

As all of this audio is being sampled, converted, and played back, an SNR defines the ratio of desired sound to background noise, or hiss. The higher the number, conveyed in decibels, the better the ratio. One reason that discrete sound cards are said to be so much better than motherboard-down sound chips is that they plug into a PCI slot, saving them from the noise a motherboard circuit might encounter. When a customer auditions a good source on a great soundcard and compares it to a poorly engineered one, much of the difference will likely come from the respective SNRs.

Two customers might have completely different opinions after listening to "Hotel California" on your average business PC. And that's what makes audio so difficult to evaluate—everyone's hearing is different depending on age, background, and so on. The audible range of vibrations reproducible by a given piece of hardware is referred to as its frequency range. Humans without any ear damage generally pick up sound between 20 Hz and 20,000 Hz, so a respectable sound system will at least cover that spectrum. High-end cards, such as Creative Labs' X-Fi, can do between 10 Hz and 46 kHz. The real key when building systems around such capable audio hardware is selling comparably high-end playback equipment. Picture a pipeline that starts by double-clicking on a DVD-Audio disc in your drive and ends with speakers reproducing music. The weakest stage in that pipeline will diminish overall fidelity.

Beyond the basics of sound quality, integrated and discrete solutions have their individual abilities and special features. Fully-fledged DSPs like the X-Fi's apply custom environmental sound effects to games or create positional audio through ordinary stereo headphones. Integrated solutions, such as those based on Intel's HD Audio standard, pave the way for up to 15 simultaneous input and output streams and quality specs once found only on pricey drop-in cards. Once you get a general idea of how each differs, recommending the right solution gets much easier.

Simple for Simplicity's Sake
Some motherboards sport built-on audio controllers such as VIA's Envy24 as an added value. However, moving the controller to a discrete card enables more connectivity and, in many cases, better sound quality.

The Benefits of Integrated Sound

In years past, integrated sound solutions were shunned for their sub-par connectivity options, sloppy software, and questionable quality. Through the efforts of a few hardware vendors, those stereotypes aren't necessarily accurate anymore. Integrated audio quality is still largely affected by the motherboard manufacturer's implementation, but at least you have the potential for high-quality sound and a stable software suite at your disposal.

The HD Audio specification, championed by Intel, goes a long way to normalize what your customers can expect from built-in sound processing. Out the window goes AC'97, yesteryear's eight-year-old standard that tarnished the efforts of codec vendors such as Analog Devices, SigmaTel, and C-Media. HD Audio introduces the concept of a single, stable software driver (Microsoft's Unified Audio Architecture, or UAA), employable by any compliant hardware configuration. It enables multiple, simultaneous streams and recognizes codecs independently, even on the same motherboard. Support for array microphones and auto-sensing jacks are two other features you won't find on any high-end X-Fi card. The HD Audio initiative provides for up to eight-channel output at 192 kHz/32-bit quality, which is audiophile-class by any measurement. Given the right complementary hardware, simple integrated audio based on the HD spec can deliver fidelity comparable to expensive add-in boards.

Much of the functionality inherent to an HD Audio solution resides in the motherboard's southbridge, be it Intel's ICH7/ICH8, NVIDIA's latest nForce, or AMD's 580X CrossFire. The southbridge links to an external codec (coder/decoder), such as Analog Devices' AD1988B, SigmaTel's STAC9220, or VIA's VT1708. Those codecs handle the conversions to and from analog and digital sources at up to 32-bits of resolution. Depending on the codec, many HD Audio implementations facilitate 7.1-channel analog output. Optionally, motherboard manufacturers that choose to license Dolby technology may apply to one of three tiers: Dolby Sound Room, requiring at least an SNR of 75 dB and two-channel output, Dolby Home Theater, calling for 85 dB and six-channel audio, or Dolby Master Studio, necessitating an SNR of at least 95 dB and eight-channel sound for Dolby Digital Live. Tagging your whitebox machine with Dolby's stamp of approval might be a great way to attract attention from multimedia enthusiasts.

Higher-End Sound
Although it doesn't offer any sort of hardware acceleration, M-Audio's Revolution 7.1 is decked out with inputs and outputs. The card also boasts an incredible 107 dB SNR and yields greatplayback quality at a price point under $100.

HD Audio goes a long way in cleaning up a lot of the negativity once associated with integrated audio. But each motherboard vendor's HD Audio implementation differs, meaning sound quality does too. Moreover, all of the great features enabled by the specification still require host processing cycles, dinging system performance in demanding situations. Controllers from C-Media and VIA, which you'll often find on motherboards, do come with hardware mixing engines, offloading CPU load to an extent as an alternative to full software processing. More commonly, though, they constitute the backbone of inexpensive discrete sound cards.

When Discrete Makes Sense

HD Audio is doing wonders for the credibility of integrated audio, and there's no doubt that a built-in sound processor makes the most sense financially. Some customers would rather not depend on their motherboard manufacturer's layout decisions for noiseless playback, though. There are those who'd prefer a bit of hardware acceleration and perhaps more S/PDIF options. For connectivity, higher-end processing, and that extra level of fidelity and depth, stepping up to an add-in card makes sense.

The same VIA and C-Media controllers used to upgrade motherboard audio often become the centerpieces of inexpensive PCI boards, giving you an affordable segue from integrated audio to higher-end sound. Take M-Audio's Revolution 7.1 as an example. Built around VIA's Envy24 controller and priced under $100, the card enables 7.1-channel analog output, a separate mic input, an analog input, and coaxial digital output. M-Audio rates the Revolution's DACs at a straight 107 dB SNR and its ADCs at 100 dB with .003% total harmonic distortion. Getting such clarity from integrated audio would be much more difficult.

Of course, the upper echelon of audio processing belongs to Creative and its latest X-Fi lineup. The X-Fi cards are driven by a powerful DSP, which accelerates all of those functions integrated audio chips normally offload to your customer's CPU. Granted, Intel's dual- and quad-core processors leverage so much computing horsepower that audio acceleration isn't as pertinent as it once was. Nevertheless, Creative is still the only vendor enhancing games with the latest EAX advances and up to 128 simultaneous voices using the X-Fi's DSP. Armed with more mature, robust software, the X-Fi still holds a quality advantage over integrated audio that's obvious to any discerning ear...under Windows XP, at least.

Microsoft's Vista Curve Ball

The talk about hardware accelerated audio may turn out to be much ado about nothing. Windows Vista alters the audio stack currently in play, removing support for acceleration in games by preventing communication between applications and hardware. Creative saw the changes being made to Vista early on and threw its efforts into advocating OpenAL, a cross-platform 3D audio API capable of giving developers a different path to audio hardware.

Quite a few games currently support OpenAL, and we're sure more will emerge after the launch of Vista. However, today's games are still looking for a Windows XP driver and certainly won't find one on a Vista platform. The result is likely ordinary stereo output instead of positional surround sound. Acceleration once made possible under DirectSound3D is now purposely being handled in software by Vista. Naturally, Creative is working on a wrapper that'll intercept DirectSound3D calls, convert them to OpenAL, and re-enable hardware-assisted processing. But for the reseller interested in solid platforms, beta drivers, wrappers, and work-arounds hardly hold the same appeal as Microsoft's UAA, which was specifically designed around HD Audio.

If things looked bleak for companies selling discrete sound cards before Vista, the situation's even worse in a world forced to live with Vista. Except for sound quality and connectivity, there's less of a reason to spend $100 or more on an add-in audio controller. Stabilizing hardware acceleration on an operating system not designed for it sounds like a shaky proposition at best. And while cards such as the Revolution 7.1 facilitate impeccable analog quality, home theater buffs are far more likely to take the simple route and use an S/PDIF optical output.

The value winner for 2007 would appear to be a trio of technologies. Windows Vista and its UAA is one, laying the foundation for a homogeneous audio architecture. HD Audio is second with its potential for exceptional quality and ample connectivity. Finally, multi-core CPUs open the door to intensive audio processing in software without worrying about performance losses.