BOOM-BOOM ROOM
Although most people became familiar with digital audio through the introduction of the Compact Disc in 1982, audiophiles had been listening to analog LP discs cut from digital tape masters since sometime in the '70s. (Of course, the rabid phonophiles were against digital mastering of analog LPs right from its inception.) In those earlier days, the sampling rates of digital tape recordings varied according to the type of digital recorder employed. London/ Decca's proprietary digital tape recorder had a sampling rate of 48 kHz, while Telarc used a Soundstream recorder with a sampling rate of 50 kHz.
In 1977, I recorded Virgil Fox with an earlier Soundstream recorder operating at 48 kHz; 3M's digital recorder operated at 50 kHz. Actually, digital recording was being used by Nippon Columbia (Denon) in Japan as early as 1969! This was a 13-bit system with pre-emphasis, using 32-kHz sampling. Since only the analog output of the digital recorders was used for cutting LP lacquer masters, their different sampling rates were of no consequence. The sampling rates of current recorders still vary, but all can be converted to the 44.1-kHz standard for CD mastering.
Today many record companies have either stopped making LPs or are issuing them on a very limited basis. This is especially true for classical music recordings. Moreover, virtually all classical music is now recorded digitally, and more than 40% of pop/rock recordings are now digitally mastered.
The CD has become the highest quality medium for home reproduction of music. However, many people have found that its greatly expanded dynamic range and extended bass may impose extraordinary performance demands on their playback equipment, and that it may also require them to modify their listening habits. To those who listen mainly to pop/rock, whether on modest mass-market rack systems or elaborate and expensive audio components, the enhanced sonic qualities of the CD medium should pose no problems. This is due, of course, to the fact that such music is normally recorded with a very limited dynamic range, even if recorded digitally. It is not uncommon for pop/rock recordings to have a dynamic range of no more than 10 to 12 dB, and few exceed 20 dB. This is true in spite of all the signal-processing equipment avail able to the recording engineer, which enables him to create elaborate sound effects. In other words, pop/rock CDs usually reach very high output levels, which are sustained throughout play back. Actual playback loudness in the average home listening environment is a matter of individual preference, al though it is dependent on the output capabilities of the audio system.
A factor related to playback loudness levels of pop/rock CDs is their low-frequency content. The wide spread use of synthesizers and electronic instruments in these recordings can provide bass of great power that can extend to near-subsonic frequencies. It should also be noted that synthesizers can generate very high levels of high frequencies. The high- and low-frequency energy output of synthesizers is far beyond the capabilities of any acoustic instrument. Thus, although pop/rock CDs usually have restricted dynamic range, they afford consistently high output level and powerful, ex tended (usually exaggerated) bass response.
It is interesting to consider that a pop/rock CD, even if made from a digital tape master recorded by the most skilled engineer employing state-of the-art equipment, is still to many a contrived product in spite of its musical values. An audiophile who is willing to spend the money can duplicate the playback system that was used to monitor the recording in the studio. By reasonably approximating the studio control room's acoustics, an audiophile with this duplicate system can precisely and unambiguously repro duce the music as heard by the engineer on the edited tape used for CD production.
In contrast, the truly accurate reproduction of a classical music CD is an exercise in frustration. The goal of fully reproducing both the dynamic range and the lowest bass fundamentals of organ and orchestra is very difficult to realize. This is true even with the very best, cost-no-object audio components. Any approach to the reality of the concert hall listening experience necessitates an approximation or simulation of the recording hall's acoustics. The monitoring systems used by recording engineers on classical music sessions convey neither the full dynamic range nor the lowest frequencies of the music.
Dynamic range, sound pressure level (SPL), bass response, and signal-to noise ratio are interrelated. In spite of advertising hyperbole, relatively few loudspeakers have significant response below 40 Hz, and rare indeed is the speaker that is flat down to 30 Hz, let alone 20 Hz-or 16 Hz, the fundamental of a 32-foot organ pipe.
To give some indication of the kind of loudspeaker response needed to re produce the lowest frequencies on classical music CDs, we have invaluable data in a paper, "Subwoofer Performance for Accurate Reproduction of Music," in the June 1988 Journal of the Audio Engineering Society. Its authors, Louis Fielder of Dolby Laboratories and consultant Eric Benjamin, did a fairly exhaustive study of classical mu sic CDs containing significant low-frequency energy. They selected 13 CDs and reproduced them through a CD player whose -3 dB cutoff was at 3 Hz. The recordings were analyzed at the moment of maximum low-frequency energy, using a Hewlett-Packard 3561A spectrum analyzer with peak-hold circuits. The authors established the minimum audible low frequencies, based on the threshold of hearing and a maximum peak CD output equivalent to 110 and 120 dB SPL, for each of the 13 selected discs.
The infamous cannon shots on the Telarc CD of Tchaikovsky's "1812 Overture" registered 110 dB SPL at 12.5 Hz and 120 dB SPL at 10 Hz! Does anyone out there have a system that can achieve these figures? Telarc recorded the Dupre Organ Symphony in G Minor in Royal Albert Hall in Lon don. This hall's organ has one of the few 64-foot pipes in existence, which speaks at 8 Hz! In the initial movement, the first harmonic of this monster (16.5 Hz) is at 110 dB SPL. The sound of real thunder captured in Telarc's "Grand Canyon Suite" registered 120 dB SPL at 15 Hz! Hindemith's Organ Sonata No. 1, on an Argo CD, has a huge, sustained 18-Hz pedal note during the last few seconds of the finale, which registered 110 dB SPL. (I have the recording, so I decided to check this pedal note. I used a GenRad sound level meter set on fast C-weighting, placed 10 feet from Duntech Sovereign loudspeakers. Driving the speakers was a new FM Acoustics 811 amplifier, which delivers nearly 800 watts per channel with a continuous current output of over 80 amperes. When the 18-Hz note was reproduced, the meter registered 106 dB SPL. Believe me, the sound was LOUD and visceral but nonetheless quite clean except for a small amount of Doppler distortion.) Big Notes by Flim and The BB's, a dmp CD, was analyzed by Fielder and Benjamin as having a 22-Hz synthesizer note at 110 dB SPL. There was even a 12.5-Hz note at 120 dB SPL! The remainder of the 13 CDs had tones from 22 to 29 Hz at 110 dB SPL.
These are very sobering figures about low-frequency performance at levels which are very hard to achieve and which, needless to say, require gobs of power. Reading the Fielder and Benjamin paper explains why SPLs of 110 and 120 dB are necessary to reproduce really low frequencies.
They also point out the influence of typical home listening-room acoustics in the propagation and perception of low frequencies. This paper makes fascinating and informative reading on other aspects of harmonic and intermodulation distortion, phase distortion, and Doppler distortion. Sadly, Fielder and Benjamin conclude by stating, "None of the loudspeakers tested were able to meet [our] distortion and noise criteria over the frequency range of 12 to 100 Hz. A dozen systems were tested and found inadequate." When he was with Ampex, Louis Fielder presented an AES paper, "Dynamic Range Requirements for Subjective Noise-Free Reproduction of Music." Here again, he is informative and offers significant conclusions. His experiments resulted in a dynamic range requirement of 118 dB for non-amplified live music, 124 dB for amplified live music, and 106 dB for the high-quality consumer playback sys tem. As you can see, this correlates well with CD's dynamic range of 96 dB, once ambient noise in the playback environment is taken into account.
Now if you live way out in the country, miles from any neighbor and away from main roads, the ambient noise floor might be as low as 15 to 18 dB SPL-if all the motors in your house (pumps, refrigerators, et al.) are turned off. But if you live in a New York City apartment on a busy street, the noise floor is at least 46 dB SPL or maybe even 50 or 55 dB SPL. If you adjust system volume so that the quietest pianissimo on a CD recording is barely audible, then when the music reaches maximum fortissimo, you'll soon learn (and your neighbors will verify) just how loud the music can get! I've had the experience of having visitors drop in when I'm playing a fortissimo passage on a CD. The first comment they make is, "Good God! Does it have to be that loud?" I tell them to turn down the volume to what they consider a comfortable level. Of course, when they do that, the pianissimos disappear. At low volume set tings, you might not even realize that notes such as really pianissimo bass drum strokes were missing unless you were following the music with a score! But I'm afraid there are people who just don't like wide dynamic range.
The 20-bit digital professional re cording systems I've reported on will provide 120 dB of dynamic range. However, that must ultimately be reduced to 96 dB, 16 bits' worth, to be processed into a CD. Most audio component systems have difficulties dealing even with the dynamic range and low-frequency content of 16-bit classical music CDs. To fully reproduce those discs requires loudspeakers covering the broadest possible frequency range, an amplifier with plenty of power (and the ability to supply it instantly on fortissimo midrange and, especially, low-frequency passages), and a component system free from buzzes, hums, etc. If true 20-bit recordings were available on some sort of tape format, few listeners or systems, if any, would be able to cope with the playback demands those recordings would impose.
(adapted from Audio magazine, Jun. 1990; Bert Whyte)
= = = =