Testing...1, 2, CD

When the Compact Disc was introduced 10 years ago, most audiophiles were not familiar with the arcane world of digital technology. Even when they eventually understood the fundamentals of digital recording and the complexities of CD players and digital processing, many audiophiles were frustrated by their inability to "tweak" the new devices. Of course, in the analog era, the tweaking of audio equipment was a time-honored tradition, the mark of a true audiophile. In spite of the relatively immutable parameters of CD playback, many audiophiles have embraced a mind-boggling array of devices that purport to improve the sound quality of CD. In my opinion, many of the devices being offered are of dubious scientific or sonic validity, and some clearly cross over into the realm of audio fantasy.

There are, however, some very useful things that, used intelligently, can at the very least provide a solid frame of sonic reference for CD performance values; at best, they can significantly improve the CD sound in your listening room. I am referring here to the CD test recordings that have been issued by a number of companies since the launch of the CD format. CD test recordings vary in content. Some have musical excerpts, along with test signals, to demonstrate certain technical aspects. Some have test signals that can differentiate performance parameters of the CD player, and still others have signals useful for checking overall system performance.

Some recent test CDs place emphasis on signals that can provide information on the interaction of room acoustics and loudspeaker placement. Most of these test CDs are useful even if the only measuring instruments are your ears, but they can provide far greater information with the addition of some fairly basic and inexpensive measuring devices.

The most useful device to accompany test CDs is a sound level meter, which can range from the elaborate and expensive Brüel & Kjaer units, widely used in professional audio, to good-quality meters from GenRad and other companies. At rock bottom and of particular value is a simple sound level meter from Radio Shack. It has an SPL range of 60 to 126 dB, plus A and C weighting scales, and costs only $31.95. To properly plot SPL readings of the sound level meter from CD test signals, you need Keufel and Esser frequency response graph paper.

In surveying various test discs, keep in mind that similar or duplicate test signals appear on these CDs. There is no essential difference between third octave test tones on one disc as compared to the same tones on another. In choosing a disc, see what special-purpose signals or tests are offered along with the ubiquitous signals to check frequency response, and decide if they fit your needs. Actually, to utilize the full gamut of tests, it is necessary to own several test CDs.

One of the earliest CD test recordings was The Digital Domain (Elektra 960303-2). This is a sort of showcase for digital technology in the form of elaborate synthesizer works, done at Stanford University's Center for Computer Research in Music and Acoustics. The very first track is very unusual-and potentially dangerous! It opens with 15 seconds of silence (all bits set to zero), which can be used to check the noise floor of a system, followed by faint sounds of insects, light breezes, and a burbling creek. If you happen to turn up the volume to increase the audibility of these sounds, you may blow your speakers with the tremendous roar of a jet plane take-off that suddenly follows the pastoral scene! Then another jet takes off, and one lands with a huge output of low frequency rumbles. PZM mikes were used to record the nature sounds, and B & K mikes were used to record the jet planes. Track 16 has another dangerous high-level sound-that of a helicopter, recorded with B & K high intensity mikes. The test signals are for the usual channel identification, followed by pink noise with and without emphasis, and then 1-kHz sine signals down to -60 dB. Digital Audio Check (Denon 33C39 7441) has the expected complement of balance and phase signals plus a frequency signal swept from 5 Hz to 22.05 kHz. The extended high frequency is exactly half of the 44.1-kHz CD sampling rate, which allows examination of the attenuation characteristics of any low-pass filter in a CD player's circuitry. A level sweep and a phase sweep follow, and then there are short intervals of repeated orchestral music and solo piano at levels of -60 dB, -40 dB, -20 dB, and at (maximum) 0 dB. Playing the highest, 0-dB level first and then raising your volume control for each lower level will provide a good test of low-level linearity and noise. The test signals are followed by a dozen musical excerpts that demonstrate a variety of recording techniques used in various recording venues.

One of the most comprehensive arrays of tests is on a two-CD set from Pierre Verany (PV. 788031/788032). No less than 106 digital tests and demonstration tracks are provided. The first CD starts with music examples of voice, solo instrument, chamber music, organ, and orchestral music. The test section starts with a maximum-level, 0-dB, 1-kHz calibration tone. This is followed by tests for channel identity, frequency response, harmonic distortion, signal-to-noise ratio, de-emphasis, transient response on square wave signals, and response to tone bursts. Tracks 43 to 46 are both fascinating and revealing. An excerpt of Dvorák's Ninth Symphony is heard on track 43 with normal 16-bit quantization, on track 44 at 15 bits, on track 45 at 14 bits, and finally on track 46 at eight bits! Even on a high-resolution system, it takes very keen ears to pinpoint differences between the 16- and 15-bit tracks. The 14-bit track adds noise, is not as clean, and has less accurate timbre. The eight-bit track is quite noisy, sounds compressed, is grainy, and has grungy artifacts. Next are tracks for intermodulation distortion, while tracks 52 to 56 are used to demonstrate D/A converter overload on both pure tones and music. When I listened to this, it was obvious that the ear is more sensitive to overload on tones than it is to music.

The second CD from Pierre Verany is mainly concerned with digital, mechanical, and electronic performance parameters of the CD and the CD player. Tracks 1 to 24 provide information on linear cutting velocity and track pitch of a CD. Tracks 25 to 50 are among the most interesting on this disc and may cause consternation among some audiophiles. Deliberately produced "dropouts" (lack of data) were generated by the laser cutter. These range from 0.05 mm on track 25 to a horrendous 4.0 mm on track 38. The CD Red Book standard states that CD players should be able to correct dropouts of 0.20 mm. Beyond the standard, the Reed-Solomon code's error correction should be able to regenerate a lack of information of 2.47 mm, corresponding to an interruption of 1.9 mS. Most good CD players can do this, and some can correct even larger defects.

The remaining test tracks concern dropout correction on minimum track pitch and the ability to correct successive dropouts. It should be understood that today's CD manufacturing plants produce CDs with an amazingly small number of defects.

Actually, most dropouts are a result of careless consumer handling, which produces scratches, stains, etc. Of course, if there are gross dropouts beyond the error-correction capabilities of the CD player, it will go into "mute" mode.

The first Stereophile Test CD (STPH-002-2) begins with a reference tone of 1 kHz, includes tests for channel identity and phasing, and includes some pink noise. This is followed by a dissertation on hi-fi by Stereophile founder J. Gordon Holt, recorded in mono. As Holt speaks, the mike used to record each of his statements is changed (and identified in the CD booklet). Most of the professional mikes used for today's recordings, as well as several more unusual types, are utilized, and it is easy to discern the tonal balances and flatness of response between the various mikes. There is also a music section that permits comparison of ND converters. Finally, a series of third octave warble tones, starting with a 1 kHz reference and then descending from 200 to 20 Hz, can be quite useful for checking loudspeaker bass performance and room interaction. (This CD is available from Stereophile, P.O. Box 5529, Santa Fe, N.M. 87502, or call 800/358-6274.) The British publication Hi-Fi News & Record Review on its test CD #2 (HFN 015) has perhaps the most comprehensive group of test signals. In addition to signals for checking channel identification, balance, and phasing, it has signals for calibrating level down to-100 dB (dithered), checking spot frequencies from 2 Hz to 20 kHz, making frequency sweeps, ascertaining S/N ratio and IM distortion, checking pre-emphasis, and checking response to impulse noise, pink and white noise, and square waves. The disc contains an Ambisonic walk-around, a number of Ambisonic music excerpts, and music for microphone comparison. (This CD is available from Music & Sound Imports, 2381 Philmont Ave., Suite 117, Huntingdon Valley, Pa. 19006, or call 800/487-7686.) One of the more recent test CDs is from Chesky Records (JD37). The first part is a jazz sampler of Chesky recordings, followed by the obligatory test for channel identification. Then follows one of the most fascinating tests available on any CD: Listening Environment Diagnostic Recording (LEDR). The test is for stereo imaging and uses a computer-generated signal developed by Doug Jones of Northwestern University. It really is a means of detecting reflecting surfaces in the loudspeaker/listening room interface, which can degrade imaging. In the first test, the left-channel signal should appear to begin at the midpoint of your loudspeaker and then move straight up toward the ceiling! The same should be true for the right channel.

After the signal moves straight up toward the ceiling, it should arc across the ceiling and then move down to the opposite channel's speaker. Finally, the signal should move laterally between the speakers. If the signal does not have the motion as described, it indicates the presence of an interfering reflective surface. For example, if the signal does not rise straight up from the speaker, suspect a reflection from the ceiling, which might be alleviated by using some absorption o diffusion.

Repositioning the speakers might be in order, and running the LEDR test again will reveal if the results are any better--or any worse! Other useful tests on this CD are for image depth, absolute polarity, low-level linearity, and ND converter comparison.

For all of the aforementioned test CDs, you can plot the test signals by using a sound level meter and frequency response graph paper, or a third octave analyzer. This will give you at least some idea of major peaks or notches in frequency response as heard from your listening position.

Whatever test CD you use, it is reassuring to know that its signals, unlike those on the LP test discs of yore, are as accurate as those from the test instruments themselves!

(adapted from Audio magazine, Nov. 1992; Bert Whyte)

= = = =