MP3THE MEANING OF A FORMAT
DUKE UNIVERSITY PRESSCopyright © 2012 Duke University Press
All right reserved.ISBN: 978-0-8223-5287-7
Acknowledgments.........................................................ixFormat Theory...........................................................11. Perceptual Technics..................................................322. Nature Builds No Telephones..........................................613. Perceptual Coding and the Domestication of Noise.....................924. Making a Standard....................................................1285. Of MPEG, Measurement, and Men........................................1486. Is Music a Thing?....................................................184The End of MP3..........................................................227Notes...................................................................247List of Interviews......................................................295Bibliography............................................................299Index...................................................................331
Chapter One PERCEPTUAL TECHNICS
In 1938 the unfortunately named Professor Edwin G. Boring wrote that "really not so very much happened in the sixty years after" Hermann von Helmholtz published On the Sensations of Tone in 1863. By this, he meant that Helmholtz inaugurated a new paradigm for understanding hearing—one grounded in the methods and ideas of physiology and physics—that other researchers played out for six decades. The judgment may have been a bit skewed by Boring's interests. But something did change in hearing research—especially American hearing research—in the 1910s and 1920s. Researchers and money poured into the field, and hearing studies began to shift their touchstones from physiological and physical models to a particular set of psychological models. During this period, psychologists and engineers devised ways of thinking about sound, hearing, and technology that shaped media for the rest of the twentieth century.
The shadow of a listener imprinted in every MP3 has its origins in the history of psychoacoustics as an academic subject and the history of a particular kind of imagined human subject within that field. Starting about one hundred years ago, the needs and interests of telephone research increasingly conditioned the problems, materials, and methods of hearing research, transforming the field and providing the foundations for modern psychoacoustics (as well as speech therapy and several other related fields). The models of the hearing subject, and the ideas of information that make possible the MP3 format and digital audio, were rooted in specific problems faced by AT&T as it sought to increase its profits as an industrial monopoly. This chapter focuses on the origins of modern psychoacoustics, while the next chapter considers the development of information theory. The Bell System and a burgeoning electrical components industry provided both fields with plenty of sunshine and fertilizer to nurture their growth.
This chapter and the next explore how the nature and tenor of the connections and analogies between ears and sound technologies shifted in the early twentieth century. In part, this was a result of the change from the dominance of physiological acoustics to psychoacoustics in hearing research. Not only did each field construct a different model of the ear, but the two fields also posed hearing as different kinds of problems and in different kinds of spaces. In the history of aural-electrical thought, the 1910s and 1920s marked the beginning of a shift from the middle ear as the site of inquiry toward the inner ear and the mind. Similarly, conceptions of what was transmitted by ears and media changed. Telegraphic electricity signified a message (as did the messages in the nervous system), but telephonic electricity was transduced from sound and back into sound. Between these moments, sound would come to exist as information. An electrics of a resonant middle ear gave way to an electronics and informatics of the inner ear. Bell's drive to conserve bandwidth led the company to participate in the development of a new way of thinking about perception, its limits, and the limits of communication technologies that still echoes in the design of new media a hundred years later. Perceptual coding descends from Bell's initial quest to squeeze more profit out of its infrastructure, as do many of the electronic and digital communications technologies that surround us. Riffing on Friedrich Nietzsche's discussion of aesthetics in Nietzsche Contra Wagner, John Durham Peters writes that early sound media were forms of "applied physiology." As research into the telephone and research into hearing collided, the converse also came to be true. Psychoacoustics and information theory were theoretical extrapolations of communication technologies.
Audiometry and the Emergence of Modern Psychoacoustics
Standard stories about the history of hearing research mark two major shifts in the twentieth century: an orientational shift from physiological acoustics to psychoacoustics, and a series of major technological changes that transformed the field. Boring's history of hearing research listed specific technologies as turning points in ideas about the ear. Technology also plays a starring role in Audrey B. Davis's and Uta Merzbach's history of hearing research. They note that American research generally followed European work before the First World War in substance, method, and equipment. 3 The authors lament that early twentieth-century university-based psychological researchers could not follow "the path of technology" unless they aligned themselves with engineers in industrial laboratories and clinicians in state laboratories: "These were no longer the days when one could rely on the munificence of a King Maximillian of Bavaria to underwrite an apparatus for vowel construction, as had Helmholtz, or make history as a Bell or an Edison on the basis of invention and entrepreneurship alone. In the industrial society of the twentieth century, scientific advances based on technological breakthroughs have been made with the support of either government or industry." Though their view of Bell and Edison is somewhat romantic (in fact, both benefited greatly from an association with Western Union and private investors), Davis and Merzbach rightly identified a new set of principles guiding scientific research into hearing in the early twentieth century.
Davis and Merzbach write that in the approximate half-century between the publication of the final edition of Hermann Helmholtz's On the Sensations of Tone and 1930, "audition in psychology passed through a well-defined phase." By this they mean that psychological studies of hearing began the period as a mélange of laboratory psychology, theoretical exploration, and physiological experimentation but ended with the dominance of electrophysiology in the study of hearing. The employment of hearing history by Boring, Davis, and Merzbach echoed an important paper published by the Bell Labs researchers Harvey Fletcher and R. L. Wegel in 1922 titled "The Frequency-Sensitivity of Normal Ears." They list prior attempts to determine absolute thresholds of human hearing. In the process, they directly connect knowledge of hearing to available sound technologies. Consider the technologies used and dates in the list:
1870: organ pipe
1883: tuning fork
1889: telephone receiver
1904: modified "phone"
1905: telephone receiver
1921: thermal receiver
The list notes a shift from musical-mechanical technologies to electrical apparatus connected with the telephone and suggests a story of progress that ends with the authors: "The development of the vacuum tube amplifier and oscillator, condenser transmitter and thermal receiver has given us precision apparatus which has made it possible to make accurate measurements of ear sensitivity." Georg von Békésy's and Walter Rosenblith's history of hearing research, published in 1948, also marks a break between a fourth phase characterized by microscopic observation in physiological research and a fifth phase characterized by experiments with living animals and electrical effects. For them, the shift was technological.
The psychoacoustician Hallowell Davis's reflections on his career, published in 1977, place the period from about 1910 to about 1930 as the key moment (and especially around 1930) when a device called the audiometer came into general use, displacing "tuning forks, ticking watches, and the whispered voice as a clinical tool for diagnosis of hearing loss." Davis approvingly cites Fletcher's preference for phonograph records, sound film, condenser microphones, vacuum tube amplifiers, and high-speed mirror oscillographs for the study of human speech and hearing. In making those choices, Fletcher privileged the trappings of modern sound media over earlier tools of acoustic research.
The audiometer encapsulates this shift. An audiometer is a meter that electrically generates a continuous tone that can vary in pitch and loudness. It is important because it allows the standardization of responses and measurements in experiments, and thus enables researchers to create and compare aggregate data and to reproduce experiments and results. Psychoacoustics emerged at a moment when psychology as a whole aspired to the condition (and social status) of a physical science, when psychology labs were first being developed and experimental methods were crucial to the field's self-understanding and to the knowledge it aimed to create. The problem was that psychology dealt with the intensely private, interior dimensions of human experience, an issue that had a longer history in empirical thought. According to Kurt Danziger, "In ordinary scientific observation the experiences that are described are private too—no two scientists see exactly the same phenomenon—but this problem is overcome by the use of special media of communication that make agreement on crucial aspects of the experience possible. These media are partly physical ... and partly linguistic." The audiometer represented just such a special medium—one that combined a physical instrument with a new, more precise and more measurable way of talking about people's perceptions of sound. It took a central place in psychoacousticians' own accounts of their field's history because it represented the moment when the psychology of hearing found a possible solution to the problem of how to walk and talk like a physical science.
The earliest published references to audiometers appeared in 1879, shortly after the invention of the telephone, and telephone technology would be important to their development. A biography of Alexander Graham Bell attributes the term to him, although at least two other researchers were calling their devices for measuring hearing "audiometers" by 1879. Audiometers appear to have first been used to test for partial hearing in the Deaf. Bell, for instance, reported in his experiments that many children originally classified as totally deaf actually had some partial hearing. For Bell, even a little hearing could be useful in teaching Deaf children to speak. But despite these early reports, the technology itself was not in wide circulation for decades. As late as 1896, we can still see psychological researchers struggling with how to measure the perceived intensity of sound and how to talk about it. A Science article from 10 April of that year describes the two then-reigning methods for determining loudness: (1) "dropping a ball successively from two different heights and recording the minimum difference in height necessary to enable the observer to determine which fall gives rise to the louder sound" and (2) "moving an object producing a constant sound, such as a ticking watch, or a tuning fork, uniformly towards or away from the ear." From the point of view of the emerging scientific mind-set in psychology, neither approach was sufficient, since they lacked reproducibility, standardization, and calibration, making both methods impressionistic at best. The author, Joseph Jastrow (founder of the University of Wisconsin's psychology department), makes reference to an electrical audiometer as a possible solution but deems it unfit for experimental use. Instead, Jastrow constructed a gas-powered apparatus to continuously vary the volume of a steady pitch. With it, he was able to mark the points at which a sound became just audible or just inaudible to an experimental subject. Jastrow's device was, however, limited in that it was difficult to standardize and reproduce.
The first widely used audiometer in North America was developed at another Midwestern land-grant university by Carl Seashore (see figure 4). Seashore was an experimental psychologist and a student of Edward Wheeler Scripture who had trained in Germany with Wilhelm Wundt and was thus aware of the emerging German psychophysical tradition. There is a possibly apocryphal story that Yale's psychology laboratory was founded the very day that Seashore enrolled as a student. Seashore's interests were steered toward the then-new field of applied psychology, and especially apparatuses for testing subjects. As an assistant professor at the University of Iowa in 1898, Seashore developed an audiometer in collaboration with a physics instructor named Charles Bowman. Their unit used a telephone receiver attached to a battery-powered electrical circuit. The circuit produced forty gradations of a clicking sound, which the operator could vary to measure the points at which a test subject could hear differences. The Stoelting Company sold the Seashore audiometer commercially for decades, but it did not initially catch on as a device for psychoacousticians. In part, its reliance on the electrical and telephonic technology of the late nineteenth century limited its applications for research, and in part, its potential user base was quite small.
The device that did catch on, and that in some ways still defines audiometry, was developed by Edmund Fowler, Harvey Fletcher, and Robert Wegel. It was a vacuum tube–based audiometer that amplified pitched tones generated by a wave filter. The amplifying vacuum tube (or audion, as the inventor Lee de Forest called it) allowed for the amplification of small electrical signals. Until transistors were developed in the late 1940s, vacuum tubes were the best way to amplify electrical signals. Tubes made it possible for experimental psychologists and physiologists to fine-tune their measurements, to generate sounds for study, and to measure tiny electrical signals in the brain and elsewhere. For AT&T, tubes would also be essential for long-distance service and other amplification applications.
As a social field and as an intellectual practice, the psychology of hearing underwent many changes that made the tube-based audiometer possible and its use meaningful—and these changes are registered in the device itself (see figures 5 and 6). Western Electric's audiometers differed from Seashore's model in more than just their tubes: they combined many strains of emergent audio technology, from telephony, sound recording, radio, and other fields in order to make precise measurements of hearers' sensitivities to different frequencies and different volumes. M. D. Fagen's history of the Bell System describes it thus: "The new tools making these more accurate measurements of audition possible were, of course, the vacuum tube oscillator and amplifier, the condenser transmitter, and the thermal receiver, together with a special attenuator whose range of variation in current output was more than three-millionfold." Fagen's "of course" comes from the fact that the components in this new audiometer all emerged from other strands of Western Electric's research. During the First World War, Western Electric engaged in research on a string galvanometer for amplification of small signals on telephone lines, extremely sensitive microphones for the study of sounds emitted by large guns, portable transmitters with vacuum tubes that could withstand battle conditions, and secret-signaling methods. Immediately before the war, they worked on telephony, radio, and public-address systems as well as other devices useful for building and testing such technologies. Western Electric hired a physicist named Harold Arnold in 1911, and by 1913 Arnold had convinced them that the best way to improve the telephone system was through basic research into sound and hearing. They hoped to outline the smallest perceptible thresholds for changes in sounds. With this knowledge, it would be possible to design a telephone system that transmitted sound that could be heard well but was maximally efficient in its use of current and bandwidth. Thus, Western Electric's audiometer combined a new organizational ethos with the fruits of its ongoing research programs for military and corporate interests.