Shadowplay: Archival and Cultural Perspectives of Delivering Music Online
Contrary to popular perceptions of the Wild West, modern historians of the American thfrontier note that many westering pioneers in the 19 century wanted desperately to retain
the trappings of the eastern, Victorian age civilization they were escaping. This was difficult to achieve in any real sense – many had thrown the better portion of their
belongings out the back of the wagon in order to negotiate rugged terrain and save the oxen, while the rigors of frontier life taxed Victorian niceties. Nevertheless, although artifacts of respectability and civilization littered the migrant trail west, re-creating the east in the new western context proved an abiding goal, and was achieved with varying degrees of success.
So appear the ironies of delivering digital archival media online. Even as we throw bytes out the back of our virtual wagons we want what‟s left at the end of the trip to give us a fair resemblance of what we started out with. The prosperity of the digital frontier lies in freeing information from physical constraint, but its price is the continuing challenge of better representation of the original. A digital copy will never be more than a shadow of its physical analogue, regardless of technological advance. How good, then, does this shadow have to be, to accurately, reliably, authentically represent the artifact that is our object? The answer, echoing around the next corner up ahead, seems to be: as good as it can be.
Certainly this is the approach for digitally preserving physical objects such as photographs or sound tapes. But for less apparently esoteric purposes, such as the listening copy of a recording, the answer is, initially, not as clear, and approaches vary. One rarely hears debate over reference copy quality especially regarding sound files, and seldom is the question pondered, “What is the critical mass at which a song delivered online attains authenticity for purposes of study or appreciation?” Although most sound
and music archivists are in all probability taxing their resources just attending to preservation and migration of their recordings, this question becomes more compelling as use of online archives grows. If the original sound artifact remains locked in the archive cupboard, and if a master copy is rarely pulled except for migration or duplication purposes, the digital reference or listening copy becomes the de facto cultural artifact as
music repositories with online collections make the claim, “This is what the music sounds 1like.”
This presents an obvious problem as sound and music repositories increasingly provide online access to both non-profit archival and non-unique commercial collections: If listening copy quality is not rigorously attended to, the classes taught, papers written, and cultural interpretations made from these primary sources are jeopardized. How much time and how many resources, then, should archives and repositories give to creating listening copies? Again, the answer appears to be something along the lines of: as much as they can give. However, unlike digitizing for preservation, where the digital sound object often remains within the archive and may be independent of popular listening formats, archivists must take into account limitations of online technology, and even client
patience, when devising their methods of efficiently delivering listening copies to users via the spectrum of bandwidths, CPU speeds, and browsers. The result is a digital copy made to be intentionally different from but still mirror its source. That is, while shedding massive quantities of bytes so listener technology can bear its weight, the slimmed-down copy needs to be good enough to convey the cultural message inherent in the master file.
This paper analyzes the “listening copies” provided by several types of online music repositories, to address the question, What challenges do online music providers face in
delivering their particular type of cultural material, and how are they meeting these challenges? The discussion and case studies will be informed by archival concepts of quality, including selection and uniqueness of materials, authenticity, and patron trust and expectations, as these vary across systems and especially between academic repositories and their commercial counterparts.
TECHNICAL HISTORY OF A LISTENING REVOLUTION
In discussing the technical standards of sound delivered online, it may be helpful to review some of the basics of digital sound technology and especially its evolution in online environments over the last several years.
Many of the same principles that apply to digital imaging also apply to digital sound. For both, a measure of resolution is accompanied by bit-depth, also known as word length. Where in imaging pixels per inch define resolution, in digital “sampling” of sound the
measure of resolution is Kilohertz (or sound wave frequency) per second. Where greater
bit depth in imaging allows for a larger color palette and therefore greater color precision in the digital scan, so in sampling greater bit depth allows for a larger spectrum of tones and therefore greater tonal precision in the digital sample.
The advent of the compact disc in the early 1980s set the first widely held standard in digital audio, with a sampling rate of 44.1Khz and a bit depth of 16 bits (or two bytes). The CD offered extended playtime, greater clarity than LPs or tapes when played on most consumer equipment, and a seemingly unlimited number of plays. CDs changed the way music was both recorded and listened to, and according to some not for the better. Audiophiles accustomed to high-end analogue setups in fact called foul early on, as the digital spectrum allowable by CDs clipped out what they considered hear-able chunks of sound, harmonics and nuances not picked up at 44.1/16. Sound archivists, already hard-edged audiophiles with a sense of historical mission, continued to preserve to tape.
Fifteen years on, digital technology capable of rendering analogue sound into digital had grown considerably stronger and become tremendously cheaper. A basic consumer computer with a merely decent soundcard could generate better-than-CD quality copies of analogue sources (typically 48Khz/16-bit), while with some modest investment in a meatier soundcard and extra RAM a home computer could make a .WAV file at 96Khz/24-bit (even if this file couldn‟t be rendered onto a playable CD without dumbing it down to a 44.1/16 AIFF file). The .WAV file, the audio equivalent of the .BMP raster 2image, had become the common currency of digital audio recording.
The drawback of the .WAV file was its size; with a CD-quality 3-minute file taking up around 30,000 megabytes, the format did not lend itself to easy desktop processing or web transfer (especially on a dial-up modem) and thus creators of .WAVs could not easily share them using the internet. MPEG technology, particularly .mp3 technology, mitigated this problem. At work on compression standards for audio visual materials since 1988, by 1992 the Motion Picture Experts Group developed a means of dramatically reducing sound file size. Based on the psychoacoustic principle that humans best hear tones of 2Khz-4Khz (within an overall hearing spectrum of 20Hz to 20Khz), an .mp3 file rendered from an uncompressed source at 128 kilobytes per second (Kbps) reduced an uncompressed audio file by a factor of ten, getting ride of the tonal ranges humans don‟t hear so well. This “lossy” compression scheme sliced out audio information from the extreme portions of the audio spectrum, much as the JPEG image rendering system did with pictures, leaving behind what to the human senses might 3constitute a fair mirror of the original.
The passable audio quality provided by .mp3, especially when compared to other portable sound packages with inherent limitations (e.g., cassette tape), sparked a revolution in the way people listened to recorded music; or, put another way, changed the face of delivering cultural material to audiences. In retrospect the success of .mp3 appears to be serendipity: By the time .mp3 emerged, consumer technology had developed at a pace where it could capably handle the format, and users recognized that digital music could finally be effectively processed on a desktop and online. Audio quality in this context was secondary; convenience was king in the .mp3, much as it had been in the format‟s cassette tape predecessors.
Although interests of the recording industry and issues of copyright will not be addressed in this paper, it is significant that .mp3 technology is non-proprietary and that it is the first audio medium that was neither created nor controlled by the recording or 4broadcasting industry. It is difficult to overstate the significance of the cultural shift that has occurred because of this. In the space of four years audio “file-sharing” via .mp3 has
decimated the recording industry, redefined popular perceptions of copyright, and given rise to a whole new business model: online music vendors who believe they can offer both an alternative to the dubious business practices of their traditional recording industry rivals, as well as a convenient way to legally own cheap tunes.
However, the emerging online .mp3 giants are haunted by the populist stance of the lamented Napster mach 1, a file-sharing software of and by the people whose free music swapping service succumbed in 2001 to the legal entanglements of copyright. Vendors like iTunes, legally picking up Napster‟s dropped torch in pay-to-download services, are
faced with a singular problem. As long as .mp3 was the currency of grassroots users, it could avoid close scrutiny by critics regarding quality standards or best practices. Because users of the service also provided its content, and because Napster was so free it 5was libertarian, the concept of “standards” was anathema. Less than self-governing, it
was little more than controlled chaos. New services delivering online music for pay 6cannot afford such freedom, when customers demand value.
Held to even higher standards, as is perhaps appropriate, some music libraries and archives have seen fit to use .mp3, as well as other formats, to deliver audio files to academic audiences online. And again, convenience rather than audio quality, the desire to cater to an increasingly online audience, appears to be a major motivating factor (although not always). Here the quality stakes may be a little higher, but as we will see, in the context of online music providers, quality is a matter of perspective, and can be measured in many ways. In assaying quality in the delivery of online music, to see how providers are meeting the challenges of their audiences for worthwhile cultural material, we must take a thoughtful approach that goes beyond simple measurement of the sound technology itself, and even beyond the more complicated measurement of convenience. As important as these factors are, this study of quality extends to selection and uniqueness of the materials in the archive or repository, authenticity of the material as delivered, and the trust and expectations of the client population.
RARE FINDS? SELECTION AND UNIQUENESS
“Archival records are thought above all to be unique, and much of their value is seen as a 7consequence of this inescapable circumstance.”
It would be folly to summarily treat online music vendors such as Apple‟s iTunes
(http://www.apple.com/itunes/) as one would the online presence of an academic or
archivally-minded music repository like e-Tree‟s Live Music Archive (accessed through
http://www.archive.org/). Marketing dollars, profit-motives, and the Internet Archive,
philosophical differences aside, however, the case for distinction between the two when measuring quality of the “online deliverable” begins to break down, especially once we
start speaking of standards such as selection of material and that material‟s uniqueness.
Can online music vendors and archives lay claim to uniqueness as traditional archives do, in their promise to provide access to unique materials? It would appear for purposes of discussing what are in effect online versions of listening copies that the answer is no, but
the question merits further investigation, since uniqueness of holdings in traditional archives has implied some expertise on the part of archival staff in the content areas of those holdings. For archives, providing unique information to a greater or lesser extent could be a measure of its ability to ably access and/or represent cultural content, as the 8established “expert” regarding the material. If uniqueness does in fact imply this sort of
expertise, then it could have an impact on assessing online music sources that cannot claim uniqueness.
The traditional assumption regarding non-archival sources of cultural information (whether one considers these bookstores, record stores, print shops, galleries, or libraries) is that the materials they hold are not unique (i.e., copies are held elsewhere). Archives on the other hand have often been defined in terms of their uniqueness, their value to the community perceived as a function of preserving and providing access to rare or one-of-a-kind records, selected for their uniqueness, that can speak as surrogates for the people 9who created them. Uniqueness in an archives is a reassuring affirmation of its
documents‟ authority and, by association with those documents, the authority of the archive itself. But actually defining the qualities of unique documents was never easy, 10and according to some is getting harder to do every digital day. For instance, James
O‟Toole has identified four different meanings of uniqueness and in the process concluded uniqueness may be the straw man of archiving. There is, he says,
a) Physical uniqueness, or the individuality of the artifact containing the information. b) Uniqueness of information, or the individuality of the content of the artifact. c) Uniqueness of process & functions, or the individuality of how the information was created.
d) Uniqueness of aggregations of records, or the individuality in which a collection of 11information or documents is put together.
The trouble with the first two meanings of uniqueness is that identifying the unique object and the unique information in many cases can be quite difficult, and assigning value to them can be quite problematic. For example, in terms of photographic images, is the cultural object of importance the photographic negative or the print? Isn‟t it the copy 12that carries the cultural message, rather than its source? Can something that is
inherently not an original be as or more important than that unique object? Sound recordings carry similar issues with some important differences. Many modern recordings, especially after the development of tape and the ability to manipulate it, are often several steps removed from source material. With the exception of live performances recorded with no subsequent editing, modern recordings, often mixed down from multiple takes (not always performed with chronological linearity) and then glued together, cannot be considered mirrors of real events. “Such a recording may well
represent a unique assemblage of scattered bits of aural information but…this kind of uniqueness is unconnected to any pre-existing reality. A new, artificial reality has been 13constructed instead.” The unique cultural object cannot be considered the multi-track tape, or its mixdown, or the master, or the CD copy sitting on the shelf in the record store. These may all have varying degrees of uniqueness, but none of them can claim to have captured an actual performance, and as far as conveying cultural material is concerned, it is the last link in the chain, the CD copy, that may be the most likely candidate for achieving that goal.
The last two aspects of uniqueness are interesting, and may help our interpretation of the concept as applicable to online music providers, if only to an extent. Archives are interested not only in the uniqueness of objects and the information they contain, but how that information was generated, and in what kind of organization the information was then placed. From O‟Toole‟s perspective, both of these types of uniqueness are doomed because it would be difficult to locate a process or aggregation that did not admit of some unique quality. “In the end, everything differs from everything else, and the presence or 14absence of uniqueness thus permits us to draw no meaningful distinctions.”
However, for this paper, O‟Toole‟s suggestion of unique aggregations of records, and the way in which they were created, is meaningful, with a slight adjustment of perspective. There are three major functions or actors that define an archive or repository: 1)
document creators or authors; 2) document keepers or archivists; and 3) document audiences, or patrons. If in the realm of online music delivery and downloading we shift the boundaries and consider the provider the document creator, and the user the archive, then there are unique aspects about providers that users may well be interested in. The way in which the providers render their music files, organize them, present them, and restrict them could all be considered unique aspects of music providers and among criteria for their ability to deliver sound and music.
As demonstrated, uniqueness in the context of online music archives and commercial sites doesn‟t work as a measurement of quality, since there are few, if any, sites providing 15sound files not derived from source tapes or otherwise copied. If uniqueness plays a
role at all, it is in the way providers bring records together and organize them, which, if we shift perspectives and make the user the archive, might be described as uniqueness of aggregation. Related to this is the principle of selection, a critical aspect of archival quality that speaks to the same kind of expertise implied by a collection‟s uniqueness.
Selection in academic and commercial online music archives tends towards one of two sides of a spectrum of choice. At the specialty end, offering fewer choices but greater depth of story, are most of the academic or non-profit archives. The Live Music Archive, part of the Internet Archive and run by eTree, is one example of this type. The spirit behind the Live Music Archive is similar to the one inhabiting the mini-industry that built itself around audience taping of Grateful Dead shows (which the band always allowed, recognizing its value to their reputation). Bands included on the site are selected by recordists who choose to upload concert recordings of performances by bands who themselves very much follow the ethic of the Grateful Dead, both musically and culturally. There are strict rules regarding formats on eTree, the primary directive being that no lossy compression schemes be used in uploading files to the site. This again is an expected outgrowth of the tape trading ethic begun by “jam” band audiences of the 1960s and 70s. Sound quality, as a means of achieving proximity to the performance, has always been of critical importance to these audiences. Files are therefore placed on the archive as .SHN (“shorten”) or .FLAC (Free Lossless Audio Codec) files, and downloading them requires software that can translate these non-lossy formats back into .WAV files playable on a PC. It also requires considerable time – lossless
compression files, while smaller than their source .WAVs, are still many times the size of .mp3 equivalents. Because tape traders of this caliber have always been tireless with regard to description, eTree also requires that uploaders record in metadata fields not only pertinent facts of the performance, but the chain of transformations that rendered the sound files to their downloadable state. Selection has therefore been defined by the culture itself, both bands and audience, who follow a shared ethic with regard to musical taste and use. In this model, the archivists are also often the users.
At Southwest Missouri State University, the Max Hunter Folk Song Collection (http://www.smsu.edu/folksong/maxhunter/) archives traditional folk ballads in compressed RealAudio (a proprietary format) and CD-quality AIFF format (the format played by commercial CD players). Selection of songs is not an ongoing process – Max
Hunter, a salesman with a tape recorder, collected the songs from 1956 to 1976 as he traveled through the Ozark Mountains, so the collection is discrete and will not continue to grow. Hunter‟s decision on what to record became the default selection, and because he recorded traditional songs, many of which are variations on the same tune sung by different singers, the collection conveniently serves scholars studying the development of folk song and the malleability of the form. Its rarity as a web resource has less to do with the uniqueness of its songs than its mere presence online – there simply are not many, if
any, comparable web collections that also offer cross-referencing to other collections or folk song transcriptions (such as the Child Ballads). With some songs represented as many as eight times (“Barbara Allen”), the mission of the archive appears implicit in its selection. In this model, the authors, archive, and users are separate, but the specialized nature of the selection of the collection‟s songs shows that it is closely aligned with specific needs of folksong scholars.
At the other end of the choice spectrum, offering greater choice but thinner collateral, or context, for those resources, is Apple‟s iTunes. Currently the champion of online music vendors (10 million downloads in its first four months alone, before it released its 16interface for PC), iTunes‟ selection policy appears motivated by the pop culture that
supports it and its ability to talk artists and record companies into allowing songs to be placed online in their proprietary twist on the .mp3 format. Like a traditional record store chain, iTunes sees its strength in allowing users to select from the broadest possible collection they can put together. In this case, the authors, archive, and users are separate entities, and selection of culturally relevant material is controlled partly by the archive, and partly by the authors or rights holders. Customers do influence selection insofar as they (theoretically) decide what to buy, and therefore influence what is selected and sold, but it could also be argued that as the content manager iTunes ultimately controls user choice.
At the same end of the spectrum with iTunes, although a little more towards the middle, is eMusic, which uses VBR, or variable bit rate .mp3 technology, to provide music to customers. Unlike iTunes, eMusic actually has a stated selection policy (of sorts), which is decidedly mass market but focused on a specific audience: eMusic is the digital music
service for discerning music fans, having paved the way for groundbreaking independent artists and labels to digitally distribute and market their music online. Boasting more than 275,000 songs in every genre from 900 of the world's best independent music labels, 17eMusic is the premier subscription service for independent music. A selection policy
aimed to draw listeners of independent music (i.e., music independent of big record labels) distinguishes eMusic from many other online commercial services and in an archival sense adds value to its collection, as eTree does, by letting users know the nature of its collection.
Selection can be revealing in estimating the ability of online music providers to effectively deliver cultural material to audiences. Ideally, selection should function as it does at the Max Hunter Collection, where the mission of the archive is implied in the content it accessions. This is not to say that the other archives mentioned do not achieve their goal, but only at the Max Hunter Collection is the actual cultural content the star of
the show, and the focus of the archive‟s expertise. At eTree, while the jam band genre naturally fills its vaults, given its community‟s tape trading ethic, the story told is less about that genre than it is about the importance of a lossless compression scheme –
beyond that requirement, there are no restrictions (outside of copyright) limiting selection for the archives. Many may count this as a strength, but it is possible to read into it a certain lack of control over the direction of the collection. iTunes and eMusic may achieve commercial success, much as a discount department store might, on the strength of policies lacking clear selection standards. Their success as repositories of commercial music may, ironically, depend on an actual disregard for the cultural content of the material they vend.
Uniqueness and selection is a first step in considering the archival qualities of online music providers, and as we have seen, uniqueness is difficult to realize, particularly in digital environments (and may not even be possible, or desirable), while selection can be effectively wielded or ignored to differing effect. The next step, consistently ensuring the authenticity of the material presented, that the sound files users access represent the real deal of (pop) cultural heritage, is a challenge that archives must meet if their reputation is to remain intact.
AUTHENTICITY: WHEN IS ELVIS NOT ELVIS?
In the United Kingdom for the past 25 years, [Elvis] Presley and the Beatles have enjoyed the joint honour of having achieved 17 No. 1 hits. Now, nearly 25 years after his death, and 33 years since The Beatles last reached the top of the British charts, Elvis is 18the unchallenged king. He has 18 No. 1s. – The Age, June 2002.
thElvis‟s 18 hit in the UK was a remix made in 2002 by Tom Holkenborg who, with the Elvis estate‟s blessing and using the name JXL, added drum samples, basslines, and more
than a minute to the original runtime of Presley‟s “A Little Less Conversation.” The song (written by Billy Strange and Scott Davis) was a standout among the typically throwaway tracks he recorded for his movies in the late 1960s, and came from the 1968 film Live a
Little, Love a Little. As remixes go, Holkenborg‟s treatment of the song was fun, lean and tastefully done. That Elvis should be given some credit for the song along with JXL appears reasonable (and vice versa), and although it may be stretching matters to award Elvis credit for achieving his 18th No. 1 hit, that designation is a measurement of sales and is a reflection of the popularity of the performance or performer, so Elvis might be considered more than a little responsible for the song‟s success. But there can be no doubt the song was substantially transformed, and to some criticism (Elvis fans and others noted that the song was already one of his funkiest, and the last thing it needed was a modern 120-beat-per-minute disco interpretation). Furthermore, like remixes of published recordings in general, there are concerns of authenticity regarding original performance or mixes versus performance or mixes added (often many years) after the fact, and of who should be credited for the result. There are also questions of archival and cultural authenticity: what is being presented to audiences as Elvis, versus what is being presented as JXL? What series of transformations did the material undergo? Is there a risk that I
could confuse the two if not made aware of their differences, thereby making reasoned assumptions about cultural material that are not valid?
A short tour through iTunes proves that critical questions such as these do in fact test the authenticity, and ultimately the trust, that music archives, whether they are commercial are not, need to cultivate to be successful. iTunes carries the two versions of “A Little Less Conversation,” both of which come from published compilations of Presley‟s music. Recognizing the need for metadata of some type to help users access the music and also to act as a surrogate for traditional album liner notes, iTunes does in fact provide some minimal information about the songs they sell, but only as related to the albums from which they came (meaning that songs taken from compilations and rereleases don‟t give dates of original song release, and since many albums rereleased on CD don‟t provide year of original publication, this can become a real problem). Because JXL‟s version of
“A Little Less Conversation” was included on one of the most recent compilations of Elvis‟s hits (of which there are dozens), Elvis alone is listed as the artist, and there is no 19mention anywhere of JXL or Tom Holkenborg. Although a nominally sophisticated
listener would probably be able to detect the modern sound of JXL‟s treatment upon hearing it, the lack of collateral information regarding the song‟s context is damaging to iTunes archival reputation and, even if unintentional, deceives the listener. They are selling a song that appears to be something it is not. If it weren‟t for the difference in runtime between JXL‟s version and the original from the movie that is also available, the user would be unaware on the face of things that any differences existed between the two versions offered for sale.
Figure 1. Screen shot of iTunes song information menu for the JXL version of “A Little Less Conversation.”