High Resolution Audio
G.L. Giustiniani - Honours Bachelor of Recording Arts - Middlesex Univeristy - SAE Amsterdam
|This report intends to illustrate the different digital audio formats which have recently become available to the music consumer, Super Audio CD and Digital Versatile Disc - Audio, and to analyse and compare three articles previously published on different specialised sources, regarding the well known format debacle. SACD and DVD-A formats, introduced on the recorded music market since less than a decade respectively by Sony-Philips and Toshiba, offer a higher resolution stereo alternative to the well known traditional Compact Disc, and a chance to experience the multi channel audio in a 5.1 system. Much has been written on this matter, specially driven by the market leaders involved in an attempt to win a technological battle aimed to impose the ultimate replacement of the popular CD. The CD Digital Audio standard has been criticized for too low sampling frequency and poor bit resolution, supposedly unable to fully reproduce subtle dynamics and upper harmonics in the audio spectrum. The three articles, Bell (2004), Ladberg (2002), Robinson (2003), have been chosen for well representing the main key technical aspects and for exposing the diverging opinions of the interviewed engineers on the possible success and penetration of these new formats in the market, as opposed to the established vinyl and CD formats. Before analysing the articles it is necessary to review the different digital format specifications for a better understanding of the main argument topics. CD and DVD-A are based on the same PCM (Pulse Code Modulation) digital audio technology adopted by the most used professional hard disk recording software system, Pro Tools from Digidesign. An analogue waveform is digitally sampled at regular intervals with a frequency which must be at least twice the highest reproduced frequency in the audio spectrum (Nyquist theorem). The waveform is stored in its overall amplitude at each given instant through discrete binary values of fixed word size, the higher being the bit depth the more accurate being the represented dynamics. Once defined the sampling frequency, a steep slope low pass filter (anti aliasing) is necessary to filter out all the frequencies above the Nyquist frequency. The two formats use however substantially different sampling frequency and bit resolution. The Red Book CD standard, defined in 1982, is fixed at 44.100 KHz, 16 bit, with no data compression, allowing to store on the disc a maximum of 780 Mb for a stereo program of approximately 74'. The DVD-A standard, created in 1998, is based on a variable sampling frequency between 48 and 192 KHz, 16-24 bit, depending on the required density and length of the recorded programs on the disc, and adopts a lossless compression algorithm called MLP (Meridian Lossless Packing), for a maximum of 4.7 Gb of stereo or multi channel data. SACD technology is instead based on the Sony DSD (Direct Stream Digital) lossless compression technology, defined in 1999 by the Scarlet Book standard, initially adopted for archiving purpose, with an extremely high sampling rate of 2.8 MHz, but with 1 bit data words, representing only the instant amplitude increments and decrements thru zero and one values instead of the overall signal values. It can be observed that many of the available titles available on the new high resolution formats are reissues of notorious best sellers of the past, originally published on vinyl, thus retaining a valuable iconic charisma which becomes an important appealing tool for promoting the new technology.||Robinson (2003) provocatively opens his article for High Fidelity Review with the words of famous engineer and producer James Guthrie at the New York press event for the Pink Floyd classic album Dark Side of the Moon released on SACD: “DVD-Audio cannot be considered high resolution.” This prepares the battlefield where the more or less justifiable technical statements from the different parties will keep arguing among the rest of the article. While music itself should be the central artistic message that relies on technology to be spread out to the public, it sadly becomes sometimes the vehicle to foment the battle between manufacturer affiliates, misusing fame of respected and renowned producers and worldwide established music products as a vehicle to sponsor their own technical views. Reading on the report, the namely two specific reasons behind Guthrie’s statement are not quite clear: “My principle concern about DVD-A is in the confusion that surrounds it. I speak daily to consumers and industry professionals alike, and I can assure you that the lack of understanding is very real.” “The second issue I have with DVD-A relates to the MLP lossless encoding that apparently has not yet been optimised.” These statements, as Robinson initially reports them, are too vague and do not come to the point yet, while it is certainly agreed that an engineer of Guthrie’s calibre could easily be more extensively persuasive. Only further on in the article, more precise objections can be found specifically to the MLP compression algorithm, which is the central argument in Robinson’s article. Comparing this approach with Ladberg (2002) interview to Ing. Öhman, following another similarly shameless statement in the Swedish Audio Technical Society journal, "It is nothing less than a tragedy that Sony/Philips system SACD still is considered to be a real competitor to DVD-A, though it has lower real resolution than the CD-system in the highest octave.”, a different attitude can be noticed at least in the fact that, immediately following these lines, specific technical aspects are exposed to back up the engineer’s statement. Reading on the interview, the SACD limitations come further exposed and proven by mentioned listening tests and calculations, while Robinson’s article is all focused on the attempt of confutation of Guthrie’s statement by the DVD-A and MLP promoters, showing a more theoretical and political debate. The Meridian Lossless Packing compression adopted in DVD-A is essentially based on variable sampling frequency and bit resolution according to the complexity and length of the music program, to optimize the storage size required by the generated data stream. For example, playback time for 2 channels at 44.100 KHz, 16 bit can be up to 12 hours, 4 h at 96 KHz, 24 bit, 100 minutes for a 5.1 program at 96 KHz, 24 bit.||Robinson reports Gateway Mastering and DVD representative Bob Ludwig words to basically admit that sound engineers have plenty of room to potentially make a high resolution project become low or average resolution instead, if so dictated by the producer: “Just as in vinyl disk cutting days, if a producer wishes to have a vinyl disk or DVD-Audio disc with a longer than usual side length, compromises need to be made to stretch the physics of the system.” “...in the real world, time- is-money deadlines loom over everyone’s head. A MLP encode engineer might intentionally, on their own, choose to slightly compromise the integrity of the music of a difficult encode in order to meet a deadline, rather than taking the time necessary to go the two or three attempts in order to do it with zero loss.” As understandable, the recording studio engineers might use different word lengths and different sampling rates for each channel, it can not be assured that all of the namely high resolution channels have the full resolution that DVD-A is capable of. For example the front four channels could be recorded at 96 KHz, 24 bits, the rear channels at 48 KHz, 20 bits, and the central sub woofer LFE (Low Frequency Effects), less critically audible, at 44.100 KHz, 16 bit. The real issue becomes now more a matter of strategic choices and compromises, given that there is the technical flexibility to produce a master in a way or in another. So the point is not about which format is the winner, but probably which producer is the honest, and which is the better recording, regardless of the chosen format. It needs to be said this can be applied more in general to any different format production, but the MLP encoding seems particularly open to this risk. Back to Ladberg’s interview, the DSD 1 bit encoding technique is described to make substantial use of dithering, meaning noise shaping mixed to ultra low level signal in order to sample properly all the minimum amplitude variations. At such a high sampling rate, this results in ultrasound noise and consequent lower headroom and resolution in the highest frequencies spectrum. How much this is audible and ruining the listener experience is another question, but it is still measurable. Noise shaped at high frequencies could in fact result in a perceived sound coloration in the higher spectrum, giving the listener the illusion that the tweeters are working with a better dynamic range.||Nevertheless, on the lower frequencies range, SACD shows an evident better resolution than CD, suggesting that a SACD setup might definitely rely on good quality woofers and sub woofer. Although DSD encoding does not need to make use of linear phase anti aliasing filters (of extreme high order and steep slope), the SACD players employ a 50 KHz low pass filter to eliminate the ultrasound noise and oscillations. This is not seen positively by some purists, stating that cutting these ultra high frequencies might take out important upper harmonics in the audible spectrum. A crucial aspect here to observe is that all the DSP (Digital Signal Processing) tools in the music production and mastering world are essentially based on PCM technology. It would be necessary to convert the PCM master into the DSD data stream in order to produce a SACD, except when the master is recorded in DSD format in the first instance (of course at an additional cost for the recording artist), but even in that case there would be intermediate conversion steps due to the inevitable DSP passages in between. Needless to say obviously those DSD which are mere reissues produced from original PCM masters are basically useless production efforts. Ladberg continues exposing Öhman’s opinion that DVD-A consists of a “purer and more straightforward system”, which is understandable as referred to the adopted PCM encoding, same as the mastering and DSP world, avoiding a potentially harmful encoding conversion, but still shows a certain closure to anything else than PCM technique and completely overlooks the critical MLP compression aspect. An analysis is also done about the resolution issues, comparing a less efficient and cheaper 1 bit incremental conversion system on fast transients as opposed to a multi bit integral sample storage system, but no mention is done regarding the fact that high transient signals are negatively affected if close to the anti aliasing filter cut off frequencies in the PCM process. All these deep technical considerations, though, risk to mean nothing to the average listener, who probably does not pay enough attention to subtle differences in dynamics (often flattened out by radio broadcasting) or in the extreme boundaries of the audible spectrum. Being able to appreciate these differences means essentially having trained ears, being regularly exposed to live music experience, and, last but not least, having access to a very good set of speakers. In other words it is for educated music estimators with sufficient audiophile culture to be able to do the required comparisons. In this sense, the First Glimpse article from Bell (2004) shows a more relaxed and objective approach, far from defending anyone’s thesis, but focused on real consumer needs and market considerations: “In a frantic effort to lure consumers back to music stores, record companies are again beginning to present music listeners with a solution to a problem they're only vaguely aware they have: the limitations of the compact disc.” This leads us back to the true valid question of this discussion: is really the CD format inadequate and in need to be replaced? After illustrating all the main high resolution characteristics of the different formats, Bell points out that the real winner is probably the MP3, compressed format with lower resolution than CD and with substantial audio loss above all frequencies, but optimized for quick data transfer and internet download.||
Consumers do not really seem to care about high resolution, which would imply an
investment for an additional disc player, extra speakers, and rebuilding an entire
music collection on a different format. A cheaper and faster music media is
suddenly available over the internet, besides, the compression algorithm quality in
the MP3 audio converters is continuously improving in terms of optimization
criteria and flexibility: a 4’ song ripped from a CD in MP3 format can be of variable
size from 3 Mb to 12 Mb, depending on the desired resolution quality. Resources
like software, internet bandwidth, storage supports are becoming cheaper day by
day, all these aspects contribute in making the MP3 an irresistible format.
In order to understand what consumers actually appreciate, it is crucial to examine
some undoubtedly objective pieces of information, which are the sales reports
from the Recording Industry Association of America (RIAA).
From the latest statistics available online from RIIA (Anon 2006) for the past six
years, the following figures result in the picture below, the first value expressed in million units
shipped and the second indicating the correspondent value in Million Dollars.
While it can be stated that CD sales are gradually decreasing (though still taking more than 80% of the global music retail market) in favour of an exponential growth of digital downloads, the SACD and DVD-A sales remain small niche figures, far from being even comparable. Also observing the steadily growing DVD-V figures, important to be considered in this analysis, given the big amount of music DVD Videos released in high resolution 5.1 format, it can be surely said these new audio formats had a valid chance to hit the market for a considerable time, and, so far, it did not happen. Most SACD and more recently even DVD-A are released on hybrid support containing two distinct digital data stream layers, with different pit depths, making possible high resolution and traditional players to access them separately, because of the different laser wavelengths and filtering lenses. The bottom layer is where the high definition laser ray gets reflected reading the DSD or MLP encoded information, while the top layer is visible only by the standard CD laser, which crosses unaffected the transparent bottom layer. DVD-A discs are also readable by many of the popular DVD Video players nowadays, making the hybrid format less indispensable for the DVD consumer. This is not the case for SACD fans, who necessarily need specific players or now available universal players, but anyway will never be able to playback on a computer, for the higher copy protection restrictions in the high resolution code. While SACD promoters claim the dual layer is a distinctive plus for compatibility issues, these few questions should be formulated in order to attempt to gain a different perspective: Would SACD releases sell without the Red Book standard CD layer available? Would SACD releases sell without the universal disc players being available at a comparable price of traditional CD players? Especially looking at the sales, probably not.
|Particular attention must be given to the universal players now available on the market. During the initial diffusion of the high resolution formats, particularly SACD players were considered esoteric audiophile equipment, and, of course, very expensive, not very accessible to the ordinary music lovers. After the initial blow up and failure of the SACD to become a mass format, the only way for manufacturers to keep it alive is to let the option to the users. Here the need to create universal players at lower cost, which have to entail multiple laser systems and decoding converters of different standards, most likely using budget quality components, failing to deliver the performance required to exhibit the difference between high resolution and normal CD. How relevant is the mere psychological effect in a listener’s mind for the simple fact to possess a multi standard versatile player, perhaps not being able to really distinguish the real quality between the different formats? The performance could just be of the same quality of the CD, and sometimes of lower quality than a well maintained analogue high-end vinyl setup. For high resolution disc duplication companies, how much is the additional manufacturing cost to produce a dual layer disc? Surely high enough to require a bigger investment for the producer and artist, and to allow distributors to sell a high resolution support at a higher price of a standard CD. Talking about discs distribution, both Ladberg and Bell mention the limited availability of high resolution discs in the shops if compared to standard CD. What is worse is that, for the considerations exposed before, some SACD releases could have been produced from PCM masters, and some DVD-A releases could have been mastered with lower resolution, especially if including extra material, this causing confusion and mistrust in the consumers. Both high resolution formats require the aforementioned data compression procedures (DSD and MLP) to fit the much larger amount of data in the disc, through the encoding process. In order to be lossless, the decoding algorithms at playback time need to be able to dynamically reconstruct the audio information which has not been stored. This complex reconstruction happens in the disc player and is basically a software procedure, as such theoretically subject to possible faults, required upgrades, compatibility issues, similarly to what happens in the delicate copyright control marking in the new HD DVD and Blu Ray video players, other discussed competing formats, but out of the scope of this research. Bell illustrates equally the different peculiarities of the two audio formats, with a balanced and informative approach. Aspects like audio quality, lossless compression, disc capacity are similarly appreciable and comparable between the two formats. What seems to emerge is the major flexibility in the DVD-A both from the master production and the usability perspectives. The variable dynamic range in a DVD-A allows producers to fit bonus material in the disc, like videos, extra tracks, for example stereo and 5.1 versions or multimedia content. The DVD-Audio becomes more similar to a DVD-Video thus probably inheriting some of its popularity. All of the three articles, even with different approaches and following different paths, finally tend to put a better light in favour of the DVD-A format against the SACD. Bell, though, maintains neutrality and, what is more important, customer orientation, concluding that there is no need to be in doubt of which best format to buy, both might offer advantages depending on different releases, the best thing is to buy a (good quality) universal player so to have a larger choice to build our music collection. The unarguable advantage of high resolution formats is the 5.1 listening experience, but if used in stereo playback, advantages are marginal. Undoubtedly, the educated music consumers, who are probably also good listeners and observers, are showing they will maintain their beloved vinyl or CD collection, occasionally enjoying MP3 on their iPOD when in need of a “fast food” consumption, with no rush to find a replacement as manufacturers or patent owners would suggest.|
[Anon] (2002). The Super Audio CD Format.
[Anon] (2006). 2005 RIAA Year-End Statistics. 2005 RIAA Mid-Year Statistics. 2006 RIAA Mid-Year Statistics.
Bell, J. (2004). ‘SACD vs. DVD-A. The High-Resolution Showdown’. In: First Glimpse, vol. 2, nr. 10, pp. 29-31.
Craven, P.G., Gerzon M.A., Law, M.J., Stuart, J.R., Wilson, R.J., (2002). The MLP Lossless Compression System.
Ladberg, N. (2002). CD vs. SACD vs. DVD-A.
Robinson, S.M. (2003). DVD-Audio Meridian Lossless Packing: The Great Filter Debate.