How does the quality of todays game boards measure (well, listen) up?.
By Tomlinson Holman
I guess everybody knows that digital audio is practiced today at a huge range of quality levels. Some years ago when I was chairman of the IEC Audio-Video committee and we held a meeting in Athens (join the standardization effort and see the world!), the question came up before the assembled group as to whether by definition if having digital audio in a product meant having high quality. No one present would agree with the proposition. Yet the general public probably had the analog-bad/digital-good equation already set in their heads based on the performance of the compact disc vs. the phonograph record it was simply a case of no more ticks and pops. While the original digital audio slogan perfect sound forever came under fire even in consumer circles, nonetheless the pervasive view is that all digital equipment is the same, and good. And it is not.
Enter the computer industry. First they gave us sound boards that plugged into the motherboard as an accessory, and then the audio functions migrated onto the motherboard as an essential feature. Check out the Intel Web site, and you will find audio as an item you can specify in your search for a motherboard, and a great many have it.
As the computer has come to dominate professional editing situations, both digital audio workstations and picture/sound editors such as Avid and Final Cut Pro, these game boards or motherboard audio functions are pressed into service as the input/output method to import and listen to audio, some of which is going to end up in the finished goods. So the quality of these devices comes into question.
So how good are these things? Now usually I am the kind of guy who goes around measuring everything to characterize it Ive even been known to measure such items as video camera sound systems, from lowly consumer models through industrial grade ones to the highest quality ones, high-definition cameras. (Perhaps a bit surprisingly, the $15,000 camera beats the $100,000 on audio dynamic range by about 34 dB, and semi-pro cameras come within shooting distance of the $15k breed, but thats another story.)
Measurements arent everything, as we all know, because for reasons Ive written about in the past, when you make a big technology change, like from analog to digital for instance, your whole measurement scheme has to change, because the new system is subject to a whole range of possible distortions that the older system is just not susceptible to. In the interim, before formal testing standards are developed that catch all the problems, people judge the quality of new technology subjectively. And it is here, between the period of introduction of a new technology and having the formal standards to measure it, that the reputation of a new technology is made or broken, largely on a subjective basis. The problem with this is the notorious unreliability of subjective judgments.
So Ive made the case for measuring things, but did I go out and measure a bunch of game- and motherboards for this article? Nope. Who has time to take on every battle, in every field? I just plugged em in and listened. And I did so because I was trying to characterize what the best way to use them was in order to employ them as digital audio editors.
Heres what I found: Audio on computers is a mess. The first problem is that there are multiple interlocking hardware and software level controls such that it would take an expert to do a proper job at gain staging. Its a little like the big trouble with NTSC video. If you dont like what you see, a tech will tweak the monitor to make the pictures pretty. But then you might not like the next pictures down the pike, so you tweak the monitor again. Ugly. The discipline that is required to set the monitor to a standard, and then tweak the pictures so that they look good on a standard monitor, only happens in the most disciplined facilities, and sometimes not even there. And its understandable what happened in the development of computer audio: each provider of an audio function decided theyd better provide a level-setting feature. So with an editing system having audio output to a game board all running on an operating system, you wind up with three interlocking sets of level controls. Each one is providing the functionality expected of it, especially because, who knows, in a particular hardware-software package, they might be the only game in town.
Of course the problem with gain staging is that you need test signals to check out the headroom vs. noise floor for each condition, and maximize getting the most range through the system. You have to have test signals that stress both ends of the dynamic range for this, and (shameless plug) they are provided by use of the Hollywood Edge Test Disc series and its 90 pages of documentation, a whole course in digital audio. Almost all of the tests available can be evaluated by ear, but that, too, is another story. Suffice it to say I have all the test materials I could need at hand.
Then I discovered that when you turn the volume up and down in software and listen over headphones that you hear something tragic: something I hadnt heard in many a year zipper noise. I guess I have to define that for the younger generation of audio engineers, because here I thought it was something buried deeply in the past, dead and gone. Well I was wrong. When you change gain on even the noise floor coming out of these boards, you hear a kind of click for each gain step that when summed up sound like a zipper being pulled down.
The third problem is even crazier: it seems that hardware and/or software loses track of left-right channelization. Thats right folks, what we work extensively to get right in stereo winds up arbitrarily in the left or right ear. It is stereo after all, eh? Ugh. Imagine the problem extended to 5.1 channels. We dont have a completely clean slate in pro audio on this, having just about every channel assignment to track number possible coming into mastering houses like Bob Ludwigs Gateway Mastering, but, although everything is every which way coming in, I doubt that a place like Bobs have ever shipped anything out with the channels scrambled.
When the computer industry came to the Audio Engineering Society to set a standard for computer audio, there was a great deal of back and forth between those guarding the traditional digital audio measurement quality standards and the new crowd of computer guys saying well these werent the standards of their industry. The struggle went on for some time, and Im not sure which side won out, but I can guess. AES had a duty when a group of audio engineers came to it with a standards effort to adopt a standard practical for an industry I guess, but maybe by lowering the digital audio standards to the level of game boards something gets lost in the prestige of the society.
Multichannel audio may well be the point at which audio on computers could get straightened out. Yet I hear a tale from a highly credible source that theres a 5.1 channel board on the market that wakes up each time it starts with a new channel assignment! Anyone for untangling spaghetti?