How to choose a sound card - under construction
 
Introduction
It used to be that sound wasn't really considered a crucial part of putting together a
PC. How times change! Today, playing a game without sound would be
unthinkable, and even businesses can justify the use of sound cards for applications like
conferencing and voice recognition. There are a lot of features and issues to be
aware of when purchasing a sound card. Hopefully this guide will help you figure out
what you, as a user, need out of your sound card.
Compatibility
The meaning and connotations of compatibility change quickly in the world of computing.
Back when AdLib dominated the sound card market, you needed to be compatible with
the AdLib standard to sell your sound card. Then, for a very long time, by extending
the music-making capabilities of the AdLib card with a DAC for digital voice and output,
Creative Labs moved in and set their own standard. Almost every card you see today
must provide Sound Blaster compatibility in one form or another.
Now, however, as Windows continues to work its way into the marketplace, we are seeing
more and more that Windows is insulating users from these compatibility problems.
Usually, when a card has drivers for Windows, that means that any card will work for
programs written for Windows (at a basic level). You may still have to wait for your
card or chipset manufacturer to write drivers that enable all the features of your
soundcard or fix compatibility problems, however.
The other problem is OS support. If you are looking to buy any relatively new
soundcard and plan on running something besides Windows 95, you must make sure that the
card has compatible drivers that work and are up and running NOW.
In my experience, most companies promise drivers and then may or may not deliver
them. It's up to you, the consumer, to avoid these problems. If you are
serious about avoiding compatibility problems, go with a brand name like Sound Blaster.
The sheer number of people who have these cards almost guarantees you that somebody
will find the time to make a driver for it if one doesn't exist.
Features
Here is where things can get complicated.
MIDI playback
There are basically two dominant ways of playing back MIDI's in today's PC market.
The first involves the little synthesizer that used to be on the AdLib and Sound
Blaster cards called the OPL2 and OPL3. This chip created sounds via a method called
FM synthesis (frequency modulation). It really doesn't sound all that good (kind of
like an old Gameboy!), but it's all that was possible back in the early development of
sound and music on the PC.
The second method is usually referred to as wavetable synthesis. A card using
wavetable playback actually has recorded samples of real instruments which it uses to
playback MIDI's. These instruments sound fuller and more realistic than FM synthesis
by far.
The quality of the music generated by a wavetable synthesizer is dependant upon the
size and the quality of the samples of these recorded instruments. With more space,
you can store higher quality samples for more realistic sounding instruments. These
samples, or patches, often come built into the card as ROM's. Some cards have the
capability to store patches in RAM, which lets you customize your own instruments for
better sounding music or original pieces.
The average user doesn't need great sound MIDI playback anymore. It used to be
important back when MIDI was the only way for games to playback music, but now most games
have purely digital soundtracks. The only reason to get better than average MIDI
playback now is if you happen to like MIDI's a lot of if you are a would-be composer.
Full-duplex capability
Full duplex refers to the ability of a sound card to record and playback audio at the
same time.
Why would you want to do this? Most frequently, this kind of capability is
desirable in video/audio conferencing programs like Internet
Phone or Microsoft Netmeeting.
Without full-duplex capability, you will not be able to hear what the other person
is saying while you are talking, nor will he be able to break through to talk to you.
As a result, the conversation sometimes resembles truck drivers talking over a CB
radio, where you have to signal the other person that you are done talking so he can
start. Having full-duplex allows you to carry on your conversation naturally, as if
you were talking over the phone or face-to-face.
Many people refer to the difference between full-duplex and enhanced full-duplex.
Since the most popular line of sound cards, the Creative Labs Sound Blaster
16/32/AWE64, is full-duplex but not enhanced full-duplex, this is a good tidbit of
information you might want to learn about. Enhanced full-duplex simply refers to the
ability to truly record and playback audio at the same time, with different sampling
rates, bit resolution, etc. Sound Blaster's apparently suffer quite a bit of
degradation when recording and playing back audio at the same time, plus they can only
record and playback simultaneously at identical settings (same sampling rate and bit
resolution).
Positional 3D audio and acceleration
This is a much hyped feature that hasn't quite caught on yet, but holds some potential.
Using special processing techniques, the sounds coming from your speakers can be
made to sound as if they were coming from your side or even behind you. Some cards
even support the use of 4 speakers so that the sounds really WILL be coming from behind
you.
The application of these special effects require a good deal of processing power.
With the right sound card...that is, one with an onboard DSP, you can offload these
processing requirements to the sound card and get nice effects and lower CPU usage.
The problem is that very few, and I do mean very few, games or applications support
this kind of immersive audio environment.
ISA/PCI
A lot of expansion cards have been moving away from the ISA bus to the PCI standard.
Sound cards have been fairly late to make this transition, because of the
traditionally non-demanding requirements of audio on the system's bus. After all, a
CD quality audio stream only takes 150KB/sec of bandwidth, and ISA has a 5MB/sec limit,
right? But the shift to PCI is finally happening.
How are manufacturers taking advantage of the higher bandwidth that the PCI interface
offers? The most common change being made right now is removing the RAM onboard the
soundcard used for storing MIDI instruments and using system RAM to store instrument
samples instead. PCI soundcards can also begin to playback multiple simultaneous
streams of digital audio without taxing the system too much.
S/P DIF
This connector physically looks like your standard RCA connector cable, but is
completely different. S/P DIF stands for Sony/Philips
Digital InterFace. SPDIF
connectors are the standard for exchanging digital audio signals between consumer
equipment. The kind of cables you are probably used to using carry an analog signal,
which can be distorted by the digital-to-analog conversion circuitry in your sound card,
as well as noise and amplification. If you have a nice receiver that you can hook
your sound card up to, or if you want to record music to your MiniDisc player, having an
SPDIF output on your sound card can be nice. Alternatively, having an SPDIF input
can be useful for recording music to your computer as well.
Basically, if you feel quality is very important, you might want to see if your sound
card is going to have one of these. Remember, you need something to hook the SPDIF
up to as well.
Sonic fidelity
Here is a category where opinions abound. Audiophiles stare down specifications
of SNR and THD, while many consumers can't even tell the difference between 8-bit and
16-bit sampled audio. The fact is, your PC is an incredibly hostile environment for
a sound card to be in. All that electromagnetic noise is bad news for a sound card
just trying to get a clean signal in or out of your system.
Basically, what this all means is that in order for a sound card to be true to the
original sound, it not only has to have good hardware, it also has to be properly
shielded. Aside from that, many lower quality cards tend to deviate from a flat
response curve around the lower and higher ends of the audible spectrum.
A nice link to go to for tests is http://www.rockpark.com/soundcards.
Unfortunately, I can't personally vouch for the scientific accuracy of these tests,
but they do look to be on the level.
A common way to measure the "cleanliness" of the output of a card is the SNR,
or Signal-to-Noise Ratio. SNR is measured in decibels (dB) and the higher the
number, the cleaner or better the signal is. Above 90 dB is generally quite good and
below 70 dB is getting down there. SNR measurements are usually only useful if you
can make sure that all the measurements were done identically. Unfortunately, that's
rarely the case with sound cards. SNR figures from the manufacturer should probably
be considered suspect. Your best bet is to figure out what the popular opinion about
the sound card you are considering is.
Performance
Performance isn't usually a big issue with soundcards. Processing audio usually
takes only a small percentage of your total CPU time. Still, however, it is possible
for a card with features like the ability to mix audio streams in hardware to reduce the
load placed on your CPU by these tasks. When games taking advantage of features like
positional audio become more popular, then performance may become a more meaningful thing
in the world of sound cards.
Summary
Basically, a consumer needs to know what he wants before he buys a card. Big
surprise there, huh? The items typically most important to a consumer are
compatibility and feature support. As you move into the higher-end, you begin get
more concerned with the quality and noise level of the inputs and outputs.
Techniques to Measure and Maximize
the Performance of a 120 dB, 24-bit, 96 kHz A-D Converter Integrated Circuit. This paper
was presented at the 103rd Convention of the Audio Engineering Society, New York,
September 1997, Preprint 4530.
Richard Kuo,
rkuo@eniac.seas.upenn.edu
Last updated 06/02/98 08:33 PM -0400
|
