Format Fisticuffs
Recording Rivalry
MiniDisc vs Dolby S

Ken C. Pohlmann
Stereo Review, March 1997

Format fights are central to the audio industry. In particular, because audio revolves around storage technology and the great diversity of media, debates about what medium is best for recording music are usually the most exciting. Although the battle of disc vs. cylinder occupied the attention of the earliest audio enthusiasts, today the major confrontation is tape vs. disc. When you combine the elements of analog vs. digital and uncompressed vs. data-reduced recording, you truly have a title bout.

The analog cassette came from humble origins. It was originally designed as a low-fi recording medium target at voice dictation, but around 1966 a few brave souls began using the cassette for music. This led it to great success in the car and home audio markets. Today the analog cassette is still a powerhouse format, but it is an aging and declining champion, an analog throwback in a digital world.

The MiniDisc was born in 1992, and it was a glitzy, high-tech invention from the start. It was the first consumer product to use magneto-optical recording technology to write, read, and erase digital audio data. To achieve a small disc size (and to sidestep the digital-copy controversies that killed DAT as a consumer music format), the MD format uses data reduction, in which a computer algorithm selectively throws out data deemed unnecessary for high-quality music reproduction.

The MD's designers hoped that it would displace the analog cassette, and the format has enjoyed great success in Japan. Yet success hasn't followed in the United States. In response, Sony -- MD's biggest booster -- recently redoubled its efforts by introducing more affordable MD components and packaging home recorders with portable players in specially priced bundles.

But is the audio world big enough for two portable, recordable formats? Can these two contenders stand up to the sonic standard set by the CD? How do they compare in terms of user features? To answer those questions, we arranged a bout between the formats, dressed them each in black trunks, and placed our bets.

The Cassette Defender

Representing the cassette format in this match was the Teac V-2030S. With a suggested retail price of $650, it sits among the upper middle class of today's consumer cassette decks and is typical of what an audiophile on a budget might choose for recording from CD's or archiving LP's. For a couple of hundred dollars less you could purchase a recorder that looks very much like the V-2030S; these decks are among the most ubiquitous of audio products. Unlike many other cassette desks, however, the V-2030S has a couple of secret weapons to enhance its performance.

First, the V-2030S features a combination three-head configuration in which separate tape heads for recording, playback, and erasing are placed in two housings. Because the recording and playback heads can be optimized for their intended purposes, a three-head design usually offers better sound quality than less expensive decks that combine the record and play functions in a single head. Having three heads is also welcome because it allows you to expertly monitor a signal from the tape as it is being recorded, instead of listening to the input signal with your fingers crossed.

Second, and perhaps more critically, the V-2030S provides Dolby S noise reduction, a system that is far superior to the older Dolby B and C systems in terms of attacking tape noise. But Dolby S doesn't come without caveats. Since the signal played back from the tape controls the noise-reduction processing, the system's success hinges on accurate recording and playback. To help insure this, Dolby S decks are designed to have a tight head-azimuth tolerance in order to limit deviations in high-end response that occur when the record and play heads aren't perfectly perpendicular to the direction of tape travel.

In addition, when making a recording, the recording circuits must be carefully matched to the type of tape being used. Thus the V-2030S provides a manually run tape-calibration system. To achieve the very best recordings, you must first turn on a built-in oscillator, then adjust the bias and record-sensitivity levels by watching meters while making a test recording. When you're finished, you can erase the test tone, set your recording levels, and make your tape.

Besides Dolby S, this deck also offers Dolby B and C noise reduction as well as Dolby HX Pro, which is a circuit that increases high-frequency headroom by automatically attenuating the bias signal during high-frequency peaks. HX Pro is strictly a recording enhancement, and tapes recorded with it can be played on any deck. That's good, because the HX Pro circuit cannot be turned off.

Other features found in the V-2030S included independent capstan and reel-drive motors, automatic tape-type selection (the deck sets standard bias and EQ levels for Types I, II and IV), automatic record mute (to place a 4-second blank section between recorded selections), an MPX filter (to attenuate the 19-kHz pilot tone from stereo FM broadcasts), an electronic tape counter with nominal or tape-time readouts, and a return-to-zero function (to fast-wind the tape to the counter's 0000 setting). There is no azimuth adjustment for the tape heads, nor is there a way to adjust high-frequency EQ in the playback signal. The rear panel contains phono jacks for right and left input and output, and the deck comes with a remote control for transport and counter functions.

The MD Challenger

Although the two fighters were stacked together in my office, everyone's attention was drawn to the Sony MDS-JE500 MiniDisc deck. MD recorders are about as rare as cassette decks are common, and everyone wanted to load and unload a few discs and to take a listen.

This MD deck carries a suggested retail price of $600, making it competitive, in theory, with the $650 TEAC V-2030S. Recently, however, Sony has been promoting the MDS-JE500 as part of its MiniDisc Bundle3, a $599 package that throws in a free portable MD player and a couple of blank discs. The MDS-JE500 is slimmer than most cassette decks, and it is immediately distinguished by a front loading slot that looks like the front of a computer's floppy-disk drive. It operates like a floppy drive too; MD's are loaded by nudging them into the slot and retrieved by touching an eject button.

As with most other MD decks, this one accepts either analog or digital input signals. Analog levels are set with a potentiometer, while digital transfers do not require any level setting at all. You can choose to record in either stereo or mono. Maximum recording time is 74 minutes in stereo or 148 minutes in mono on the same 74-minute MD. Up to 255 tracks can be recorded on a disc; the deck automatically finds blank space for the data. A counter keeps track of remaining time, and a very handy AMS (Automatic Music Sensor) knob lets you quickly select tracks. Access time is nearly immediate.

The deck offers a number of cool editing features as well. For example, you can delete and rewrite over existing tracks or parts of tracks, divide tracks into shorter tracks, join tracks together, renumber tracks, and create displayable titles for tracks and discs. If you goof, an undo function cancels your last edit command. The deck also provides track repeat, random playback, and track programming. Still another cool feature: Time Machine Recording lets you continually store 2 seconds of the most recent input signal in memory, so that your recording will contain the last 2 seconds of material that passed through the deck before you hit the record button. If you tend to miss the beginnings of programs, you'll really appreciate this feature.

The MDS-JE500's rear panel contains phono jacks for right and left analog input and output as well as two Toslink optical connectors for digital input and output. Digital inputs are automatically sample-rate-converted from 32 or 48kHz to the MD's and CD's 44.1-kHz rate. Like other MD decks, this one contains an SCMS (Serial Copy Management System) chip; recordings made through the digital input cannot be digitally copied to other MD's or DAT's. The supplied remote control handles transport functions and has buttons to control a CD player, perform CD-Sync recording, and enter or scroll through text labels.

The Early Rounds

Any contest between Dolby S encoded tape and MiniDisc will be decided according to what is most important to you in a music-storage format. Fundamentally, both formats are similar in that each allows recording, playback, and rerecording (or rewriting) on audio programs. From a purely physical standpoint, both formats use small, lightweight media enclosed in protective cases. In that respect they are both highly portable and rugged. MD holds an advantage in terms of the life of the medium since tape is inherently more fragile than a disc. Tape will wear out or simply degrade over time, and although the transports in Dolby S recorders are supposedly built to a higher mechanical standard than those of other tape decks, there is always the danger of a cassette tape's tangling or breaking. In comparison, MD's magneto-optical disc technology seems like a safer bet for durability.

The cassette format counter-punches with its universal availability of both blank and prerecorded media. In comparison, MD's of both types are scarce and much more expensive. Also, whereas MD is limited to 74 minutes of stereo playing time, cassettes can be reliably configured for 90 minutes. And tape players are about as common as light bulbs; you can always find one when you really need one. Although decks equipped with Dolby S are far less common than those with only Dolby B and C (particularly in the car market), Dolby S tapes can be replayed satisfactorily through a Dolby B or C decoder, or none at all, even if a Dolby S deck is required for the best results. In comparison, MD hardware is quite scarce in the U.S., so if you want to record and play MD's, you'll probably have to invest in wholly new equipment for your home, car, office, or portable use.

Like it or not, one of the principal selling points of any consumer product is its convenience. We generally hate to admit that we're lazy, but we are most attracted to products that are fast and simple to use. Although today's cassette decks are a far cry from their early ancestors, they are still shackled by the simple fact that tape is a very long and thin storage medium. For straightforward listening, there's no real downside to spreading the signal over a long physical distance, but it takes time to skip from one tape point to another. Random-access times are slow, and that is a real pain when you're trying to locate a particular section of music. Similarly, when recording to tape, you must always find the exact physical point where the blank medium begins. It is very difficult to record over existing material or between other recorded selections, and there always seems to be unused tape at the end of a side.

The MD overcomes all of these problems. As with other disc formats, it provides random access to all parts of its content. A user can switch from one piece of music to another without interruption because a memory buffer is used to temporarily hold data entering and leaving the disc. For example, when the playback data stream is temporarily interrupted as the laser pickup moves from one point to another, the buffer still provides a continuous data stream to the digital-to-analog (D/A) converter. When tracking resumes, because the data is read off the disc faster than the buffer feeds it to the converter, the buffer can be quickly refilled without any loss of output. (That is why MD is resistant to physical shocks that cause the pickup to skip.) This accessibility also means that MD recording is as easy as writing data to a floppy disk. Rather than having to find a section of continuous blank space on the MD, the recorder automatically fits data wherever space is available, whether that space is continuous or discontinuous.

The inherent benefits of disc media (compared with tape) and of a digital signal (compared with analog) make MD a more versatile format that cassette, and one that is far easier to use. If it is simple playback that interests you, then the formats are competitive. But if you want to make your own recordings and edit them in any way -- for example, to assemble a compilation recording and periodically rotate new tracks into it -- then MD wins the competition with a knock-out blow. In terms of user features and convenience, a random-access digital disc is simply a better way to go.

The Later Rounds

There is more to life, however, than ease of use. When it comes to audio storage, we are greatly concerned about the accuracy of the medium. That is, when a signal is recorded and stored, then replayed, how has the storage medium affected the quality of the sound? At first glance, the digital MD would seem to hold all the advantages over the analog cassette. Digital storage is inherently linear. If conversion into and out of a digital medium is handled with care, then the storage itself plays no role in sound quality. In comparison, analog tape is inherently nonlinear, altering the input signal in ways that must be accounted for and corrected. Cassette recorders use high-frequency bias and other techniques to iron out the nonlinearities of analog tapes.

An MD, however, is not simply a digital disc, and a Dolby S recording is not simply analog tape. The MD format employs data reduction to decrease the amount of data saved on disc by a factor of five. It calls on Sony's proprietary ATRAC (Adaptive TRansform Acoustic Coding) algorithm to do this. ATRAC applies high-complex digital signal processing according to a model for human hearing, coding the signal so that redundant or inaudible information can be omitted, thus reducing the number of bits needed to store it. This type of processing is called ``lossy'' data reduction, in that the digital output signal on playback lacks some (presumably) unnecessary data contained in the original.

Dolby S technology is also complex, but it operates in the analog domain, using dynamic compression and expansion as well as signal filtering to lower the noise floor inherent in analog tape. As with Dolby B and C, the signal is compressed prior to recording and expanded during playback; in the process of expansion, low-level recording noise introduced after compression gets pushed down in level relative to the music. In other words, MD is not as faithful as pure, non-lossy digital storage, and Dolby S is better than plain analog storage.

Of course, no matter what the technology is, the best way to evaluate a product's sound quality is to listen. To do this, I recorded a number of CD's to each medium. To give MD the benefit of the doubt, I used a digital fiber-optic link from CD to MD, thus avoiding D/A and A/D (analog-to-digital) conversion in that path and obviating the need to set recording levels. For Dolby S's benefit, I used metal tape to extract the best possible performance, and I carefully calibrated the deck's bias level and sensitivity using the built-in oscillator. I was also careful to adjust my recording levels to maximize dynamic range without distorting signal peaks. Although it is no real chore to watch levels, you do have to play close attention, and in several cases I had to go back and rerecord sections to get the best levels -- rewinding tape, cueing it up again, and so on. None of that was necessary with the MD; I simply hit the record button. In fact, this little recording exercise simply confirmed that MD is much easier to use, especially with a digital link from CD to MD.

Recording completed, I sat down with a number of diverse playback systems, listening to each format over an extended time and also switching quickly back and forth between formats. It is my opinion that for most listeners, with undemanding music, and over modest playback systems, both formats provide sound quality that is indistinguishable from an original CD's; this is particularly true when listening over portable and mobile systems. With repeated playback, the cassette may begin to audibly degrade whereas the MD will not, but initially they are equal. This is a tribute to both the ATRAC data-reduction algorithm and to Dolby S noise reduction.

With critical listening on a more sophisticated playback system, however, good listeners will begin to hear subtle differences. With specific musical passages -- ones that stressed the recording formats in terms of frequency response, dynamic range, and transient response -- I could reliably tell apart the original CD and each recording of it. Moreover, one of the recording formats sounded better than the other. It provided crisp transient response and extended high-frequency response that the other format lacked. Specifically, the analog Dolby S cassette format sounded better to me than the digital MD format. In particular, Dolby S was outstanding in its ability to tame the nonlinearities of an inherently nonlinear analog medium, whereas ATRAC contributed significant nonlinearities to an inherently linear digital medium. The only exception was solo piano recordings; compared with CD, both formats were relatively unsatisfactory to my ears. MD blurred transients and dulled high frequencies, whereas the cassette deck introduced a subtle bit of wow and flutter.

Hooks and Jabs

Because I am uncomfortable describing purely aural phenomena in words, I usually turn to analytical measurements to help convey what I am hearing. But traditional measurements are not helpful in this case. All data-reduction algorithms, including ATRAC, handle test tones significantly differently than they do complex music signals. For example, whereas a data-reduction algorithm may demonstrate a completely flat frequency response out to 22 kHz with discrete sine waves or noise, it will generate a psychoacoustically shaped response as it codes music -- for example, omitting high frequencies when bits are needed to code more important lower frequencies.

To help illustrate the sonic differences between the two formats, I recorded short sections of music from CD to MD and cassette. Then I fed the output from each format into a Pentium PC running Sound Forge, a commercially available recording/editing program. I then ran the three files through identical software processing to generate three sonograms, as shown below.

These sonograms ``show'' music as it is played in time, from left to right along the horizontal axis. Frequency distribution is shown along the vertical axis, going up the spectrum from bottom to top. The colors illustrate amplitudes; for example, red indicates the highest levels and deep green the lowest. Sonograms thus provide a very efficient, graphical way to ``see'' what a complex sound consists of. In this case, each figure shows about 4 seconds of music from the beginning of Track 5 of the Dire Straits album ``On Every Street.'' If you have the CD lying around, I'll wait a few seconds so you can play that section.

Okay, now listen again to that short section and look at Figure 1. It shows the reference CD file that you are listening to. Musical events can be seen; for example, the opening chord is easily seen at the start of the file, followed by periodic wide-bandwidth percussion hits appearing as transient impulses. The file ends, incidentally, just before the lead guitar attacks again. Now consider Figure 2, the sonogram of the Dolby S recording. You'll observe that the differences between the original CD and the tape recording of it are relatively slight -- mainly, the tape surrounds the crisp original recording with a slight analog haze. In addition, you'll see some level changes at various frequencies; for example, the Dolby S recording has slightly hotter high frequencies than the CD. On the other hand, a glance at Figure 3, the MD's sonogram, shows that the ATRAC algorithm has changed the signal considerably. It has removed all high-frequency content above 19 kHz and has significantly affected content as low as 16 kHz; notice the difference between CD and MD sonograms at 16 kHz just after the initial attack. Because of the complexity of this particular piece of music, the algorithm simply did not have sufficient bits to code the entire spectrum.

The sonogram showing how ATRAC has affected the audio signal must be carefully understood. In this case, the figure shows clear differences; less demanding musical selections might not show such differences. And even in this case, it should be clear that these differences are exactly as expected and in line with the entire theory of data reduction. This algorithm's model calculates that these high frequencies will not be easily audible at that moment and that a listener will not miss them -- thus, they can be removed. Whether such assumptions about audibility are accurate determines the success or failure of any data-reduction algorithm. The signal was changed -- the sonogram shows that. But if you can't hear the difference, then the changes don't matter. In this case, however, I did hear subtle differences in transient response and high-frequency response in the MD recording.

Mixed Decision

So who is the winner in this title bout between Dolby S and MiniDisc? That decision depends on how you score the punches. MiniDisc is a sexier, more convenient format, and it offers far more editing possibilities for home recordists; indeed, many musicians now use MD recorders for their home-studio work. MiniDisc is also a great portable format. So if it's convenience and features that you want, then MD is terrific.

On the other hand, in terms of sound quality, at least judging from these two midprice consumer decks, it is my opinion that a well-recorded Dolby S cassette sounds better. Its transient response is quite clean, and its high-frequency response is extended. If you listen to music critically, and your playback system is up to the job, you may be happier with Dolby S's sonics. So it turns out that the analog cassette, although an aging champion, is still impressive after all these years. Its younger digital disc rival has its own set of impressive skills and may one day beat the champion outright. Until then, the fight will continue.

Return to the MiniDisc Community Page.