Post a Comment
Got the creeps?
By Brian Dipert, Technical Editor -- EDN, 5/25/2000
Most of you probably know the term "creeping elegance," and some of you might have even been accused of personifying it. Engineers love to tinker. They have lots of ideas for making a product more capable (even if nobody's asking for the additional features) or faster (even if the design already meets or exceeds target performance). Unfortunately, all this extra work means that the product never gets into production. If it does, it's late, missing some important introduction date, or it costs too much.
I thought of creeping elegance a
few months ago as I listened to some cool demos at Intel's Spring Developer
Forum in Palm Springs, CA. To boost sales of its PC-audio codecs, Analog Devices
has partnered with 3-D positioning audio-software-manufacturer Sensaura Ltd and
audio-synthesis-software-provider Staccato Systems Inc (see " Sonic soft(ware)ness ,"
Sensaura's Digital Ear technology enables listeners to customize and, in effect, algorithmically scale a generic head-related transfer function for specific ear sizes, depths, and openness. The technology also adapts to a listener's head size (the distance between the ears). MacroFX technology adjusts the left-versus-right effect based on the synthesized distance between the sound source, the listener, and the resultant differential gain around the head at close distances. ZoomFX adapts the aureal characteristics to mimic real-life sound-generating lines, areas, and volumes, rather than theoretical point sources. MultiDrive creates a pleasing stereo effect—even if the listener isn't in the sweet spot directly between the speakers. Further, all of these techniques adjust themselves if you're listening through two or four speakers, or through headphones.
My enthusiasm grew as Sensaura explained each technology. The seductive power of creeping elegance had me in its grasp until I realized that combined, the company's algorithms consume nearly 9% of a 500-MHz CPU. Granted, that's not much in typical PC use, however, the resource-balancing act would be more complicated if the CPU were simultaneously running a software-based modem algorithm, decoding high-definition video, or rendering a complex, high-frame-rate 3-D graphics scene. Using fully implemented software-based 3-D audio with the less powerful processors and smaller memory arrays in embedded systems would be even more challenging.
Staccato System's software synthesizes wavetable musical-instrument digital-interface (MIDI) effects with low CPU usage. The vendor claims that each MIDI "voice" consumes only 1/8% of a 500-MHz x86 µP. For a more realistic effect (but at much higher CPU usage), Staccato Systems also supports physical modeling of sounds, analogous to the techniques at the core of advanced voice-compression algorithms. Physical modeling works by developing an arithmetic representation of the various tubes and chambers that make up a sound-generating device. It then stimulates the model with various input patterns. The company showed me a NASCAR auto-racing game that used physical modeling to create the racecar's motor sound.
The effect was realistic, even down to its real-life idle randomness. However, did it justify the MIPS needed to create it, especially if players haven't complained about games with more primitive effects? Unless you perform live A-versus-B listening tests, almost any set of speakers in your local stereo store often sounds good enough. In fairness, I should point out that both Sensaura and Staccato have restrained themselves from the extreme limits of creeping elegance. Sensaura's algorithms, for example, employ reverberation techniques for modeling the listening environment instead of the more processing-intensive, reflection-modeling approach touted by companies such as Aureal. Further, Staccato's technology dynamically scales the number of simultaneous wavetable MIDI voices to the CPU speed and available system resources.
Like Sensaura and Staccato Systems, you should remember to restrain yourself, unless you work in a research lab that encourages creative thinking. There's nothing wrong with having a passion for the products and technologies you work on. At some point, though, you have to say, "that's it," ship the product, and start generating revenue. Perhaps the biggest benefit of restrained creeping elegance is that you always have something to improve in the next go-around. So how do you know when enough is enough? I cringe to suggest it, but you might want to occasionally chat with the marketing folks down the hall. You could even (gulp) attend a focus group. However, if your life begins to resemble a Dilbert cartoon, you've gone overboard.
Author Information
Contact
Technical Editor Brian
Dipert at 1-916-454-5242, fax 1-530-937-8147,
bdipert@pacbell.net.
