[ Go to March 1997 Table of Contents ]

Sound Cards Revealed

-- by Marc Spiwak and John Woram

Sound,in nature, is an analog medium. The computer, by nature, is a digital device. So the sound card has its work cut out for it: One way or another, it must get the digital beast in the gray box to behave itself in the analog world of music and sound. Here's a quick look at what needs to be done.

CD audio processing is easy. The D/A (digital-to-analog) converter built into every CD-ROM drive converts the disc's digital data stream into analog audio before it leaves the drive electronics. A separate cable routes the now-conventional analog signal to the sound card, which functions as a simple audio amplifier.

MIDI (Musical Instrument Digital Interface) audio is a bit more complex, because such a signal is a hybrid format in which the data stream contains instructions for the synthesizer. The signal says, in effect, "Let's hear an A-natural half note that sounds like a piano. While you're at it, play a C-sharp/E-natural quarter-note sequence on a violin and at the same time an A-major chord in the bass clef." The sound card has to convert these instructions into a reasonable facsimile of the real thing, using one of the following methods.

FM synthesis manipulates various audio frequencies to synthesize the desired instruments. The resultant output may remind you of a real musical instrument, but you'll never mistake it for the real thing. By contrast, a wavetable sound system stores a collection of previously recorded (and real) musical-instrument sounds in on-board ROM. When the MIDI signal sends that request for a piano note, the sound card looks up the desired sound in a table of WAV audio signals (hence the term wavetable), runs it through a D/A converter and plays it. Because the signal is from a real piano, it sounds like one.

Some sound cards allow you to manipulate and edit ROM-stored samples, then resave them in RAM for subsequent playback. Low-end cards typically have 0.5MB of sample RAM; better ones have 1MB to 4MB available. Many cards let you add RAM in the same way you would on a PC motherboard.

At the top of the line is waveform audio, where a natural analog sound is recorded and converted to digital format for storage. The infinitely varying source amplitude is measured, or sampled, at regular intervals and converted into a binary number. If the sample size is 8 bits, each sample must be adjusted into one of only 256 (28) possible amplitudes. If you use a 16-bit sample, there are 65,536 (216) amplitudes available. Sound card sampling rates (number of samples taken per second) of 8kHz, 11.025kHz, 22.05kHz or 44.1kHz are typically available. Therefore, the size of a 1-second WAV file may vary from 8,000 bytes (8,000 samples per second by 1 byte per sample by 1 channel) to 176,400 bytes (44,100 samples per second by 2 bytes per sample by 2 channels for stereo files). Needless to say, the higher the sample size and sampling rate, the better the reproduced audio quality.

Once the sound card has done all this behind-the-scenes work, all that remains is to display the mixer on the desktop, adjust each signal's analog amplitudes and mix them into a combined analog audio output.

Copyright (c) 1997 CMP Media Inc.

Windows Magazine, March 1997, page 206.

[ Go to March 1997 Table of Contents ]