The Importance of Dynamic Range

The most generally agreed upon objective of high fidelity is to recreate the realism of live music performances in the home listening environment. Various acoustic characteristics and psychoacoustic phenomena have been identified over the years that appear to be responsible for the emotional impact and excitement that we experience at a live performance. However, past limitations in the tape recording process and software formats for traditional delivery systems, that is records and tapes, have imposed serious restrictions on one of the major factors involved dynamic range. This article will explore the significance of dynamic range in music and review the use of noise-reduction technology and dynamic range expansion to improve the realism of music reproduction for increased listening pleasure.

Since the late 1890s when Thomas Edison and Emile Berliner first began to record musical performances on cylinders and discs, there has been a persistent disparity between the quality of "master" recordings produced-with the performing artist and that of mass-produced copies made from the masters. Steady advances in recording technology, however, have improved every link in the chain of music recording and reproduction, narrowing the gap between live and recorded performances.

A major milestone in the history of recorded music was the introduction of the long-playing microgroove record in the 1940s, which increased the playing time of records so that compositions of considerable length could be recorded uninterrupted. Another major advance followed in the form of magnetic tape recording that provided a means of editing recorded performances and producing a master tape from which an unlimited number of vinyl pressings could be made. Unfortunately, magnetic tape recording of analog signals introduced its own set of problems which detracted from the fidelity of recorded music, the most notable being tape hiss, but also including wow and flutter and other forms of distortion.

The realism of recorded music was dramatically enhanced 6n the 1950s with the introduction of stereophonic sound, which brought much of the perspective of the hall or stage to the home. The "three-dimensional" character of sound produced by multi-channel signal processing touched off a great deal of research that, in the late 1960s, led to a development that promised an even greater increase in realism of recorded music quadraphonic sound. But, the audio industry experienced a marketing disaster with quad due, in large part, to its inability to agree on hardware and software standards. The potential benefits of four-channel sound were never fully realized, even though the concept had considerable technical merit and is still being explored.

Over the years, the quality of music-reproduction hardware (amplifiers, record playing equipment, and speakers) has surpassed that of available music software records, tapes, and radio broadcasts. In response to growing demands for recordings with improved sound quality, so-called "audiophile" records were introduced during the 1970s in the form of direct-to-disc records and digitally mastered records.

The major contribution made by these technical innovations was the elimination of tape hiss and the various forms of distortion associated with analog master tape recordings.

Their superiority over conventional records was immediately obvious; however, even these fine recordings are too often marred by the presence of record surface noise and restrictions on the dynamic range that could be captured on and retrieved from a vinyl disc.

We are still faced with the challenge that has been the underlying motive for the steady stream of technological advances in sound recording and reproduction to recreate the excitement and emotional impact of a live performance in the home listening environment. Our ability to meet this challenge is certainly enhanced by analyzing and understanding the acoustic and psychoacoustic factors that characterize "live performance" sound, so that appropriate consideration will be given to these characteristics when attempting to make improvements in the music recording and reproduction process.

Characterizing Live Performance Sound

Our appreciation and enjoyment of music, whether live or recorded, is strongly related to three major factors that characterize live performance sound: Tonal balance, spatial perspective, and dynamic range.

Tonal Balance

The tonal balance of music has received the greatest amount of attention over the years. It has long been appreciated that the low-frequency content of music should be kept in balance with the high-frequency content, consistent with that which occurs in live performance. Today, most recording and reproduction equipment is capable of handling frequencies that extend well beyond the audible range of 20 Hz to 20 kHz. Electronics are readily available that offer responses over this frequency range with variations smaller than those which can be detected aurally. Covering a frequency range necessary for quality music reproduction is no longer a significant technical challenge for records, phono cartridges, and tape recorders. Also, many fine speakers are now available that are capable of uniformly reproducing most all frequencies in the audio spectrum. Further improvements in frequency response characteristics are unlikely to produce greatly significant improvements in the perceived tonal balance quality of recorded music.

Spatial perspective is a somewhat elusive quality of sound involving a complex combination of geometric and temporal factors. In the geometric dimension, our perception of the spatial character of sound seems to center around a panoramic sound field in which individual instruments can be localized. Giving breadth to the musical performance, the sense of spaciousness involved is extremely important in creating the illusion of "being there" at a live performance. Stereo sound reproduction represented a major step forward relative to recreating the spatial perspective of live music. Also, special speaker designs have been developed in an attempt to produce a combination of direct and reflected sound similar to that which exists in live performances. Finally, recent developments have resulted in microphones and signal-processing techniques which are claimed to recreate the sound field geometry present during the original performance.

The temporal aspect of spatial perspective is primarily related to what is usually described as a sense of ambience.

What a listener hears at a live performance is a composite of multiple sound waves, each arriving at slightly different times because of the many different sound transmission paths that are involved. This dimension of live music has been the subject of considerable research and experimentation over the years involving the use of reverberation and delay-line devices to create auxiliary acoustic signals that are delayed in time from the master audio signal. The most recent developments that address the temporal character of sound in music reproduction are electronic "time delay" or "ambience recovery" systems that employ either analog or digital signal processing to create auxiliary signals with variable amounts of time delay. The primary function of such devices is to simulate the ambience of a wide range of acoustic environments from small, intimate rooms to large, highly reverberant halls.

There is still much to learn about the geometric and temporal characteristics of sound. Some of the approaches taken to introduce spatial perspective in music reproduction have resulted in an unnatural sound quality. Others have improved the sense of realism. More substantial advances in creating a realistic spatial perspective undoubtedly are yet to come.

Dynamic Range

Dynamic range is the difference between the sound levels during the loudest (fortissimo) and quietest (pianissimo) music passages. Giving depth to the musical performance, its existence during a live performance is as apparent to the listener as its absence in a recording. Exposure to the dynamics of live music has caused us to appreciate the extremely large amplitude differential between the whisper of a lone flute and the thunderous finale of a symphonic work. It's not too unusual to experience a 90-dB dynamic range in a live performance but, unfortunately, more than one-third of this range (and its associated effect on realism) has traditionally been lost before the music signal gets through the recording and reproduction process.

In order to store the music information on magnetic tape or vinyl discs, it traditionally has been necessary to compress or otherwise modify the amplitude of the recorded signal so that (1) the signal strength during loud passages stays below saturation and tracing distortion levels for tape recording and disc mastering, respectively, and (2) the signal strength during quiet passages stays sufficiently above magnetic tape noise levels and record surface noise levels. Basically, the problem has to do with the signal-to-noise ratio (S/N) limitations of tape and disc recording processes. The S/N ratio has to be greater than the desired music dynamic range if loud music passages are to be recorded without distortion and quiet passages are to be heard clearly above the background noise on the tape or record. Hence, an S/N ratio of 60 dB may be required to provide a clean dynamic range of 50 dB with 10 dB safety margin shared between the top and bottom ends of the dynamic range.

During tape recording, a common method for restricting the dynamic range is to "gain ride"; that is, the recording engineer manually adjusts the levels while the recording is being made reducing them during fortissimo passages and increasing them during pianissimo passages. This same result is frequently accomplished automatically by using limiters that prevent high-level signals from exceeding a preset level or by using compressors that gradually reduce level when loud passages occur and increase level during quiet passages.

Without the aid of a tape noise-reduction system and excluding allowance for signal peak "headroom," the dynamic range capability of a professional studio tape recorder is typically 60 dB at the 15 ips speed and somewhat higher at 30 ips.

The comparable figures for high-quality audiophile open-reel tape recorders operating at 7.5 ips is 50 dB, and for a good cassette recorder, about 45 dB applies. The psychoacoustic impact of such dynamic range restrictions is to make the music sound "flat" or "thin." The sharp edge of percussive attacks is blurred, the contrast between loud and quiet instruments is muddled, and the overall definition is obscured, thereby diminishing the excitement and realism of the recorded performance.

A similar problem occurs during disc mastering, since the maximum dynamic range that can be stored on a vinyl disc is about 55 dB for conventional pressings and up to about 65 dB for the very finest pressings. Again, a substantial loss in excitement, emotional impact and realism results from the dynamic range restrictions of conventional discs and they are plagued as well by annoying record surface noise, even for the best direct-to-disc and digitally mastered records.

Since dynamic range has a tremendous effect on our enjoyment of music, it should be recognized for its importance and receive more attention in the music reproduction process. Further improvements in characteristics like frequency response are not likely to significantly increase the overall quality of music reproduction. On the other hand, a modest increase in dynamic range will be immediately perceived as making substantial improvements in the realism of music reproduction. Consequently, dynamic range represents very fertile ground to be explored relative to coming closer to our goal of recreating the live-performance musical experience in the home.

Understanding Dynamic Range

The range of sound pressure that humans are capable of perceiving is extremely large. For example, the sound pressure is about one million times greater at a level that causes pain or discomfort (about 120 dB) as it is for the threshold of audibility (0 dB). This means that the human auditory process handles approximately twice the dynamic range in dB (or 1000 times in sound pressure) of a professional studio tape recorder.

======

Fig. 1 Typical sound sources or environments for a range of sound pressure levels measured in dB as well as indicated relative to 0 dB.

======

Our hearing mechanism responds to changes in sound intensity (pressure) in a roughly logarithmic manner, rather than in an absolute way. For this reason, and as a matter of convenience, the decibel scale is used to describe Sound Pressure Level (SPL), as follows:

SPL = 20 log 10(P/Po)

...where the reference pressure Po is defined as the threshold of audibility corresponding to 0.0002 N bar, and a p bar (micro bar) is the pressure of one millionth of an atmosphere.

Hence, a range of 0 dB to 120 dB covers the entire range of sound amplitude that is of interest, much like the range of 20 Hz to 20 kHz that applies to the audible frequency range of sound.

Typical sound sources or environments are presented in Fig. 1 for a 130-dB range of sound pressure levels. A doubling of sound pressure corresponds to a 6-dB increase in SPL, while 20 dB represents an order of magnitude increase. While the levels of SPL are approximate for each sound source indicated, a comparison of relative SPLs is very revealing. For example, the background noise level in a residential living room is 10 times (or 20 dB) higher than the level in a recording studio. Similarly, the noise level that exists in a general business-office environment is 10 times higher than the living room level.

Returning to music, it is possible for peak sound pressure levels to momentarily reach 120 dB during transients. If these peak levels are produced in a studio, concert hall, or home environment, the "noise floor" of 20 dB to 40 dB, depending on its spectral distribution, can reduce the perceived dynamic range of music to about 90 dB. Therefore, a music recording and reproduction system need only be required to handle a maximum dynamic range of 90 dB to properly represent the characteristic of live performance sound.

Noise Sources and Masking Effects

It is important to note that there are two separate and distinct fundamental sources of noise which detract from the fidelity of recordings those that are introduced while re cording a master tape and those that are associated with disc mastering and playback. Direct-to-disc, digital recording, and tape noise reduction applied to conventional analog recording are techniques that address the issue of eliminating (or at least reducing) the introduction of noise prior to disc mastering. Unfortunately, full benefit of these advanced recording techniques is lost to a great extent when the recording is transferred to a vinyl disc.

With the masking effect of tape hiss removed, the surface noise of the record becomes all the more objectionable. It appears that the elimination of a major noise source, such as tape hiss, serves to highlight another record surface noise generated as a result of the interaction between the stylus and the record groove. Any roughness of the vinyl surfaces of the groove walls causes extraneous stylus motion, creating the familiar record surface noise which is inseparable from the music on conventional vinyl discs. Of course, the opposite is also true. If the surface noise of discs is eliminated, any imperfections in the master tape relative to hiss or other forms of noise and distortion become more apparent. Hence, it is clear that noise in both the master tape and the recorded disc has to be eliminated, or at least dramatically reduced, to significantly improve the quality of recorded music.

==========

Fig. 2 Illustration of the various stages of the record manufacturing process.

MANUFACTURING A RECORD

Master vinyl lacquer (positive image).

Silver coating sprayed on lacquer to make it electrolytic to accept plating (positive image). Silvered lacquer is then plated with a coating of nickel.

Nickel metal master is plated with additional coating of nickel creating the-mother.

Mother is pulled from the nickel metal\ master. The mother may be played (positive image). Nickel and silver metal master is pulled from the master (negative image). Mother is plated with nickel creating the "stamper". Stamper is formed to fit into the record press. There is a stamper for side 1 and a stamper for side 2 of each record.

PVC pucks are inserted between the two stampers.

Discs are inserted into their sleeves and covers, and are then shrink-wrapped.

Stamper is pulled from the mother (negative image). Heat and steam are applied as tremendous pressure squeezes the stampers together, forming the record.

The finished discs are trimmed and set aside to cure and cool.

===========

Evolution of Recording Technology

Since the mid-1940s, there have been major technological advances in sound recording. Prior to the development of magnetic tape recording, records were made in the old 78 rpm format involving the cutting of a single-channel (monaural) groove in the surface of a master recording blank.

The fidelity and time per side (four minutes per side on a 12 in. disc) were limited, and any performance errors required recutting the master. With the advent of the long-playing microgroove record came extended playing time per side, improved pressing quality (accomplished by using vinyl instead of shellac), and increased durability all resulting in a better overall value to the record-buying public.

The Record Manufacturing Process

The beginning point in the record manufacturing process is the production of a master disc or lacquer. This is accomplished by using a master tape recording (or the real-time output of a mixing console in a direct-to-disc recording session) to provide an electrical signal that is fed to a heated cutting stylus which engraves the surface of the lacquer with traces representing the musical waveform. The diagram presented in Fig. 2 illustrates each subsequent step of the process involved. Since the quality of a lacquer may deteriorate if it is not plated shortly after being cut, the multi-stage electroplating process is generally scheduled to occur immediately after the master cutting session. The (positive image) lacquer is plated to produce a (negative image) metal master.

Then a (positive image) metal "mother" is made, from which (negative image) metal stampers are made. The stampers are used to press (positive image) vinyl discs that are a replica of the original lacquer.

From each master lacquer, only one metal master can generally be made. The number of mothers available from each metal master is limited, as are the number of stampers from each mother and the number of pressings from each stamper.

These limitations are such that only about 20,000 to 25,000 high-quality pressings can be produced from each lacquer.

Multiple lacquers, therefore, are necessary to produce larger quantities of quality pressings.

Analog Tape Recording

For the last several decades, it has been common practice in producing records to temporarily store the musical performance on magnetic tape. A typical recording session involves the recording of multiple "takes" of a musical selection in order to provide enough material to create something which approaches "the perfect performance." This is accomplished by an editing engineer who listens to the various takes and chooses the best parts of each one, then combines them together (by physically cutting and splicing the magnetic tape). This results in an edited recording of the musical selection. The practice is so common that few musical selections (notably those recorded "live" before an audience) listened to on a record today represent a continuous performance by the musicians or performing artist. Rather, we generally hear a composite of several performances where any mistake or imperfection is removed in the editing process.

Consequently, the ability to edit represents one of the major advantages of using a magnetic tape recorder in producing a record.

Another advantage of tape recording is that the master (edited) tape can be used to cut as many lacquers as are required to meet consumer demand for the record album.

Hence, there is no limitation to the quantity of records that can be produced from a master tape.

Although records produced from analog master tapes represent the best sound recordings available until recently, there have been serious technical limitations inherent in the process. Various forms of distortion and noise are introduced into the recording because of the mechanisms that are operative in the recording process. For example, analog tape recording suffers from such problems as tape hiss, wow and flutter, frequency response non-uniformity, modulation noise, multi-channel crosstalk, print-through, and transient distortion. Of particular significance is the background noise that appears as tape hiss in the recording, which ultimately gets transferred to the master disc and vinyl pressings produced from the tape.

As the magnetic tape passes by the recorder head of a tape transport during the recording mode, the electrical signal provided by the microphone pickup of the music performance causes a reorientation of the magnetic particles on the tape to produce a magnetic replica of the music waveform. In the playback mode during the cutting of a master disc, the magnetic tape passes by the recorder head but, in addition to "reading" the music signal, the head also "reads" the random distribution of magnetic particles on the tape and this is perceived as tape hiss. Consequently, during the cutting of a master disc, both music and tape hiss signals are communicated to the cutting stylus, thereby introducing background noise in the master disc and all records that are subsequently pressed. As discussed earlier, the presence of noise in analog tape recording results in dynamic range being restricted to about 60 dB for professional studio tape recorders, while for audiophile reel-to-reel and cassette tape recorders, the dynamic range is restricted to about 50 dB and 45 dB, respectively. A trade-off exists, consequently, between the advantages of tape recording (editing and unlimited production capabilities) and the undesirable restriction it imposes on dynamic range.

Tape Noise-Reduction Systems

A number of techniques have been employed since the early 1970s to improve the dynamic range capability of analog tape recorders. These "noise-reduction systems" process audio signals in various ways to increase the S/N ratio of a tape recording system, resulting in an increase of its usable dynamic range.

Two of the most successful approaches to tape noise reduction have been those developed by Dolby Laboratories and dbx, Inc. Neither one can improve the quality of an existing audio signal, that is, any noise existing in the audio signal is processed unaltered. Their function is solely to reduce the amount of noise added during the recording process. Hence, they might appropriately be called noise-prevention systems.

Both Dolby and dbx operate effectively as companders a signal processing system involving encoding of the audio signal during recording and decoding the signal during playback. However, their operation is based on completely different principles.

The Dolby Type A ("professional") noise-reduction system is actually a dynamic equalization system that operates over four separate bands of the audible frequency range. During recording, increasing amounts of pre-emphasis are applied to the audio signal in these frequency bands as the signal level approaches predetermined reference levels. During playback, complementary amounts of de-emphasis are applied to the signal in the four frequency bands. Recognizing that the behavior of the Dolby system is nonlinear relative to amplitude, the identical reference or "threshold" levels must be employed during recording (encoding) and playback (decoding) to avoid mistracking distortion. The Dolby Type B ("con sumer") noise-reduction system operates over a single frequency band in the high-frequency region above 1 kHz where tape hiss is normally encountered. During recording, all signals having high-frequency amplitudes below a reference level are subject to pre-emphasis, with subsequent complementary de-emphasis applied during playback. In the course of attenuating the high-frequency content of the audio signal during playback, tape hiss is reduced.

The dbx Type I ("professional") and dbx Type II ("consumer") noise-reduction systems operate on all frequencies and all amplitudes of the audio signal. The dbx system functions as a linear decibel compressor/expander that is neither frequency selective nor level sensitive, and thus it does not involve the use of reference levels or calibration test tones.

During recording, the amplitude of the audio signal is compressed while, during playback, the amplitude of the signal is expanded in a complementary fashion. The basic idea behind the dbx tape noise-reduction system is to keep the music signal level sufficiently higher than the offending noise level (e.g., tape hiss) during recording (encoding) so that, in the course of expanding the signal during playback (decoding), tape hiss is reduced, thus extending the S/N ratio of the tape recording system.

By processing all the frequencies in the audio signal, the dbx system eliminates any anomalies that may be introduced by limiting the companding action to a limited frequency band. Precision level-sensing circuits are used to control both the encode (compression) and decode (expansion) modes of operation during recording and playback, respectively. Employing a voltage-controlled amplifier (VCA) and a level detector in each stereo channel, the compression/expansion gain instructions given to the encode and decode VCAs are equal and opposite (mirror image) as long as the two level detectors track accurately.

All recording systems have some frequency-dependent phase shift which can change waveforms considerably even though the sound is not audibly degraded. This could introduce an error signal that is added to the music signal prior to decoding. The dbx system circumvents this problem by employing wide-range rms (root-mean-square) level-sensing circuits that respond to the signal energies, regardless of their phase relationships, in creating a command signal to drive the VCAs. Using a 2:1 compression during recording and a 1:2 expansion during playback, this linear decibel compander provides accurate results, even with extremely sharp music transients, over a wide (100 dB) range of signal level.

The decode process in tape noise-reduction systems should be a mirror image of the encode process, and the accuracy of the system in providing precise mirror imaging determines the fidelity of the signal processing. Any extraneous signal introduced into the system subsequent to encoding but prior to decoding is decoded along with the original audio signal. This, as well as any mistracking errors between the encode and decode modes of operation, can introduce distortion or artifacts in the form of unnatural sounds. The nature of the artifact is different for each type of companding system, but its character is frequently described as a "pumping" sound or "noise modulation," Obviously, one would like the companding system to be totally free of any forms of distortion, but reality is such that a trade-off is frequently involved between the elimination of tape noise and distortion of the companded signal. Fortunately, psychoacoustic masking provided by the music signal itself is generally adequate to make the artifact inaudible. The masking effect of a given sound is greatest upon offending sounds that are of somewhat higher frequency. Hence, a music signal having frequency content that encompasses the frequency range of the artifact will generally be processed with complete fidelity. The enhanced music experience that is provided by increased dynamic range and reduced background noise makes the trade-off heavily in favor of using these types of tape noise-reduction systems.

To illustrate the effect of tape noise reduction on the performance of a range of tape recording systems, background noise spectra were measured with and without Dolby and dbx noise-reduction systems in the circuit. The results are presented in Figs. 3 to 5 for a professional studio, audiophile open-reel, and high-quality cassette recorder, respectively.

The frequency spectrum of noise was measured using an automatic spectrum analyzer for a constant 125-Hz bandwidth.

The noise spectra in the figures indicate the noise level relative to 0 VU for frequencies ranging between 250 Hz and 25 kHz. The total wide-band noise was also measured in each case for all frequencies in the audible frequency range (20 Hz to 20 kHz). Since Dolby noise-reduction systems comparable to dbx Type II systems are not readily available, only noise spectra with and without dbx Type II noise reduction are shown in Fig. 4 (as well as later in Figs. 6 and 7). A summary of the total wide-band noise levels for professional studio, audiophile open reel, and cassette tape recorders, with and without tape noise reduction, is presented in Fig. 6. The chart also indicates the approximate increase in dynamic range provided by Dolby and dbx tape noise-reduction systems for each type of tape recorder. The information on this chart may be combined with that previously given to tabulate the approximate usable dynamic range capabilities of various types of tape recorders, with and without noise reduction, as presented in Fig. 7. Usable dynamic ranges in excess of 80 dB are possible for each type of tape recorder using dbx noise reduction. Professional studio recorders equipped with dbx noise reduction meet the objective of 90 dB dynamic range and, therefore, they can properly represent the music dynamics of live performance sound.

It is interesting to note that, with its wide-band companding action, the dbx noise-reduction system makes the dynamic range capability of open-reel tape recorders approximately equal to that of 14-bit and 16-bit digital recording systems (discussed later). This is a particularly important observation in view of the cost penalty associated with digital recording equipment.

===============

Above: (left to right) Figs. 3-5

Fig. 3 Constant 125-Hz bandwidth frequency spectrum of Ampex 456 tape using a Studer/Levinson A-80 professional studio recorder operating at 15 ips.

Noise level is shown relative to 0 VU for no noise reduction (upper trace), Dolby Type A noise reduction (middle trace), and dbx Type I noise reduction (lower trace). The total wide-band (20 Hz to 20 kHz) noise level of-65 dB is reduced to-76 dB by the Dolby noise-reduction system and to less than -100 dB (at the limit of the measurement instrumentation) by the dbx noise-reduction system.

Fig. 4 Constant 125-Hz bandwidth frequency spectrum of Scotch 206 tape using a Pioneer RT-707 audiophile open-reel recorder operating at 7.5 ips.

Noise level is shown relative to 0 VU for no noise reduction (upper trace) and for dbx Type II noise reduction (lower trace). The total wide-band (20 Hz to 20 kHz) noise level of-59 dB is reduced to-96 dB by the dbx noise-reduction system.

Fig. 5 Constant 125-Hz bandwidth frequency spectrum of Maxell UD XL-I tape using a Tandberg TCD-340A cassette recorder operating at 17/8 ips.

Noise level is shown relative to 0 VU for no noise reduction (upper trace), Dolby Type B noise reduction (middle trace), and dbx Type II noise reduction (lower trace). The total wide-band (20 Hz to 20 kHz) noise level of-48 dB is reduced to-56 dB and-84 dB by the Dolby and dbx noise reduction systems, respectively.

================

Direct-to-Disc Recording

During the 1970s, a "new" recording approach was introduced for disc mastering. It involved cutting a master disc in real time the same way that 78-rpm discs were produced from about 1898 until their demise in the early 1950s. Instead of storing the musical performance on magnetic tape, the electrical signal from the mixing console is amplified and fed directly to the cutting stylus of a mastering lathe hence, a "direct-to-disc" recording.

In the many years since 78-rpm discs were cut direct to disc, advances in the design of disc-cutting equipment have allowed a return to the direct-cutting method to produce recordings that sonically are superior to conventional records produced from master analog tapes. The performance advantages of direct-to-disc records result from the elimination of problems associated with magnetic tape recording, such as tape hiss, channel crosstalk, distortion due to magnetic oxide saturation, print-through, wow and flutter, etc. Records produced in this fashion have greatly increased clarity of complex musical passages, particularly those that involve loud bursts of percussion.

While their sonic superiority over conventionally produced records is immediately apparent, direct-to-disc records must still contend with the dynamic range limitations and surface noise problems of vinyl discs. Furthermore, direct-to-disc records have other potential disadvantages, including (1) inability to edit, since the master disc is cut in real time during the musical performance; (2) decreased playing time per side since the cutting lathe must be controlled manually, rather than by a computer as is normally done, and (3) limited quantity of records that can be produced (since there is no master tape to cut additional master discs as required), which results in a premium price. The trade-offs involved with records produced in this fashion may not be worth it for those who are bothered by the dynamic range limitations of vinyl discs and the annoying record surface noise that may be still present.

Digital Recording

Digital recording involves the storage and retrieval of a musical performance on magnetic tape with the continuous analog audio signal replaced by a series of binary numbers.

Basically, when music is recorded digitally, its waveform is sampled electronically many times a second (such as 50,000 times/sec) to produce a sequence of numbers that, as a function of time, represents a piece-wise replica of the original analog music waveform. This sequence of binary numbers (which are comprised solely of a combination of "zeros" and "ones" the language of the digital computer) are generated by an analog-to-digital converter at the input to the tape recorder.

When the tape is played back (for the purpose of cutting a master disc, for example), the binary numbers are "read" by the recorder head, and the sequence is translated into an analog electrical signal by use of a digital-to-analog converter. One of the most significant aspects of digital recording is that the only information the playback system can recognize is the sequence of binary numbers recorded on the tape. As the magnetic tape passes by the recorder head, the head ignores the random distribution of magnetic particles on the tape which, for an analog recording, is perceived as tape hiss.

Only information about the pure music signal is passed on to the master disc, and noise comparable to tape hiss on analog recording is non-existent. Furthermore, digital recording is not plagued by other forms of distortion inherent in the analog recording process, such as wow and flutter, crosstalk, or print through, although it is susceptible to a type of distortion known as quantization noise.

==========

Fig. 6 Comparison of total wide-band noise levels of tape recording systems relative to 0 VU, with and without Dolby and dbx noise reduction, for professional studio, audiophile open-reel, and cassette tape recorders. The approximate dynamic range increase provided by Dolby and dbx tape noise-reduction systems is indicated for each type of tape recorder.

============

Fig. 7 Approximate usable dynamic range capabilities of professional studio, audiophile open reel, and high-quality cassette tape recorders, with and without Dolby or dbx tape noise-reduction systems.

=========

A 16-bit digital system is capable of recording music with a 90-dB dynamic range over the complete audio frequency range when a sampling rate of about 50,000 samples per second is used. Similarly, an 85-dB dynamic range is available from a 14-bit digital system. Editing is accomplished electronically rather than by physically cutting and splicing the tape.

Digital tapes are immune to degradation caused by long-term storage of magnetic tape, and they can be duplicated through many generations without loss of sound quality. The only drawback to digital recording at its present state of development appears to be the relatively high equipment cost involved and potential problems caused by lack of standardization. Nevertheless, from a purely technical point of view, the performance capabilities of digital recording are extremely impressive and suggest the degree of sound quality that ultimately may be made available to the listening public when technical standards of delivery systems are agreed upon and digital playback systems of reasonable cost are developed.

Dynamic Range Expanders

Putting aside for the moment the concept of producing recordings with full dynamic range and inaudible background noise, it is appropriate to explore the possibility of enhancing the value of existing record and tape collections by restoring at least a portion of the dynamic range that existed in the original live performance. Since recordings played in the home, or those played in broadcast studios and transmitted to the home, have their dynamic range limited to something generally less than 60 dB (representing a loss of at least one-third of the potential 90-dB dynamic range), it seems logical to introduce some form of dynamic range expansion to counteract the compression that exists in tapes, records, and broadcasts. The general function of such devices, known as dynamic range expanders, is to make loud passages louder and/or make quiet passages quieter, resulting in a spreading out or expansion of the dynamic range of the music signal.

Unlike the situation which exists with companders, where the expansion process is a mirror image of the compression process, dynamic range expanders operate in a "single-ended" fashion, processing the music signal according to its particular design or user concept, rather than providing expansion that is the exact converse of the compression process.

Since the amount and nature of the compression that existed during recording and/or broadcast is not generally known, dynamic range expanders typically offer a variable range of expansion capability. The user selects the degree of expansion that provides an overall pleasing effect, avoiding excessive expansion that can introduce undesirable artifacts.

Dynamic range expanders based on somewhat different operating concepts are available from companies like dbx and MXR. Most of these devices process the wide-band audio signal by sensing the overall average level of the music signal and increasing the level (making the music louder) when a preset threshold level has been exceeded. More sophisticated expander designs separate the music signal into multiple frequency bands so that the degree of expansion that occurs in a given frequency range depends on the level of signal in that range, thus preserving the integrity of tonal balance and the timbre of individual instruments during complex musical passages.

The amount of expansion that is appropriate will vary according to degree of compression that resides in the audio signal as well as on individual music tastes. Excessive expansion can lead to artifacts frequently described as "pumping" or "breathing" noises. To avoid this situation, expansion should be limited to less than a factor of 1.5 for most popular or rock music, while factors of 1.3 or less are generally preferred for classical music. Properly utilized, dynamic range expanders can restore a significant portion of the original dynamic range that is lost in the recording process, dramatically increasing the excitement, realism, and enjoyment of conventional recordings and broadcasts.

dbx Encoded Discs

The two major problems with vinyl discs that have stayed with us over the years are restricted dynamic range and record surface noise. In cutting a master disc, a signal level that is too "hot" can create a condition that will cause tracing distortion by the cutting and/or playback stylus. If the level of the signal gets too low, it may be obscured by the record surface noise. These two conditions place upper and lower bounds on music signal levels that can be stored on a vinyl disc, resulting in a maximum dynamic range of 50 dB for conventional pressings and up to 65 dB for the very best pressings.

==========

Fig. 8 Diagram depicting the combination of signal compression during encoding and expansion during decoding that results in surface noise reduction and dynamic range retention on dbx Encoded Discs.

==========

Record surface noise is generated as a result of the interaction between the playback stylus and the record groove.

Modulations in the groove cause the stylus to undergo complex motions that are translated by the phono cartridge into an electrical signal representing the musical waveform. The stylus tip is less than one-thousandth of an inch in diameter, yet it must travel up, down, and sideways thousands of times a second to follow the undulations of the groove. Any roughness of the vinyl surfaces of the groove walls cause extraneous stylus motion, creating the familiar record surface noise which is inseparable from the music on a conventional disc.

One solution to the problem of restricted dynamic range and surface noise of vinyl discs is available through the application of dbx noise-reduction technology. The dbx Type II noise-reduction system, previously described relative to tape noise reduction, can be employed in record mastering and playback to render record surface noise virtually inaudible, while dramatically increasing the dynamic range of the reproduced music signal. The operation of this noise-reduction process applied to discs is illustrated in Fig. 8 where, for purposes of convenience, a master tape signal having a 100-dB dynamic range is assumed to exist.

The dbx linear decibel compression/expansion (companding) system operates as follows. The music signal from the master tape (or directly from a studio console) is encoded (compressed) during the cutting of the master disc and decoded (expanded) during playback. The dynamic range of the music signal is linearly (in dB) compressed by a 2:1 factor when cutting the master disc, which means a music signal having a 90-dB dynamic range is reduced to 45 dB. This fits comfortably within the maximum dynamic range storage capability of vinyl discs. During playback through a decoder, the signal picked up by the phono cartridge is linearly expanded by a 1:2 factor so that the dynamic range of the original music signal is completely restored. And, as a result of downward expansion during decoding, the surface noise on dbx Encoded Discs is approximately 30 dB lower than on conventionally-recorded discs.

The frequency spectrum of record surface noise for a conventional disc and a dbx Encoded Disc is shown in Fig. 9 for a constant 125-Hz bandwidth analysis. Measurements were made on a disc that was cut with an unmodulated groove (no signal). About 30 dB of noise reduction is provided by the dbx Encoded Disc for frequencies below 10 kHz, which encompasses the frequency band of greatest concern relative to record surface noise. The total wide-band noise of-57 dB is reduced to-85 dB, in this particular case, for an overall reduction in surface noise of 28 dB. There are a number of side benefits to dbx Encoded Discs.

During cutting of an encoded master disc, the compressed signal reduces the demands on the cutting stylus, resulting in [...henceforth, missing content til end of article--Sorry!]

Article by Jerome E. Ruzicka Vice President, dbx, Inc., Newton, Mass. 02195 (adapted from Audio magazine, Jan. 1980)

= = = =