Sound Card Quality Report

by Ethan Winer

This report was posted in the MIDI forum of CompuServe on March 25, 2000.

The main purpose of my test was to assess the quality of the SoundBlaster Live's Analog to Digital converters. A SoundBlaster Live Value sound card costs about $60, and most "professional" cards cost 5-10 times that much. So more than anything I wanted to know if the SB Live sounds as good as the expensive brands and, if not, at least get a feel for how close it is. I have a CardD Plus sound card which cost me $550 a few years ago, and an AWE 64 Gold, so I tested those too. Besides the audio quality test I also measured the residual noise from each of these sound cards by wiring up an input connector with the hot and ground wires shorted together. I then recorded a few seconds of silence and measured the noise floor of each card. First I'll describe how the tests were conducted, and then in the interest of honesty and accuracy I'll explain what I feel are the limiting factors of my test method.

For the audio quality test I extracted Wave files from two audio CDs to serve as a reference, using Exact Audio Copy version 0.9. This freeware program claims to extract audio files from a CD with higher accuracy and fewer errors than other such programs. Then I played the same CD tracks on my high-end Technics CD player and recorded that through each sound card's analog line inputs. I was careful to record all of the files at the same level as the extracted references, to avoid having to normalize them later which might affect the quality. Four files were used for each of the two songs: the extracted version, plus the files recorded through each of the three sound cards.

The first track was Donald Fagan's The Nightfly, which is a well produced and great sounding pop tune. The other is a very clean recording of Lynne Arriale's acoustic jazz trio (piano, bass, and drums). The trio track is from a CD Yamaha gave away free a few years ago to show off their then-new ProMix 01 digital mixer, and the recording was made by Tom Jung. I selected this because it is a wonderfully clean recording and also has a lot of finely detailed cymbal playing.

The test consisted of playing all four Wave files, switching back and forth quickly among them, and listening carefully to hear if there was any difference. I listened to the files through a pair of high quality headphones and also through my large, biamped JBL 4430 monitor speakers. Then I brought the wave files to Doug Ferrara's professional recording studio and we listened through his Mackie HR824 monitors. Finally, I went to Ed Dzubak's studio and we listened on his Mackie HR824 monitors. Ed is a professional composer with excellent ears and audio engineering skills, and he has three Emmys and six additional nominations to prove it.

Here are the three points that could be considered as failings in my test:

1. The source was an audio CD, as opposed to a live microphone, so it has already been "reduced" to 44.1 KHz. and 16 bits.

2. Recording a stereo mix of a complete tune may not tax the Analog to Digital converters as much as a single microphone that is capturing, for example, a live cymbal. A cymbal has a high level of mostly high frequency content, and that kind of source might reveal slew rate limiting problems more readily. Still, I believe the test is essentially fair as a way to assess the overall quality of these sound cards.

3. As far as I know the SB Live Value card is identical to the "full" SB Live which I used for my tests, and is also the same as you get for $500 with the E-mu Audio Production Studio. It is possible that Creative Labs tests all of the SB Live cards manufactured, and those with the best specs are sold as the "full" SB Live and those that come up short are sold as the "Value" version. But I doubt this is the case.

AND NOW FOR THE RESULTS...

None of us could discern any difference between the extracted reference file and the files recorded through either the SB Live or the AWE 64 Gold. Initially I couldn't hear any difference with the CardD recording either, but Ed Dzubak did. Once he showed me what to listen for I was then able to hear it too. It was a subtle change in character of the cymbals on the jazz recording, and Ed was able to identify the CardD every time. When we auditioned The Nightfly, however, even Ed was unable to hear any difference between the extracted reference and any of the sound cards.

The results of the residual noise tests are as follows.

SoundBlaster Live: -76.3 (same in both channels)
CardD Plus: -63.5 (left channel), -70.3 (right channel)
AWE 64 Gold: -70.3 (left channel), -71.2 (right channel)

The noise was determined by playing the "silence" Wave files (described earlier) in Sound Forge, and reading the levels on the peak-hold meters. When I first bought the CardD and AWE 64 Gold three years ago, and they were installed in my previous computer, the CardD sounded about 10 dB. quieter than the AWE 64 Gold (though I never measured them). So it may well be that the CardD has deteriorated, or that it could be improved by moving it to another ISA slot. Or maybe the noise energy is mostly outside the sensitive midrange and thus is less audible than the measured results would indicate. I did not use weighting filters for this test, which would take into account the actual audibility of the noise.

Entire contents of this web site Copyright © 1997- by Ethan Winer. All rights reserved.