Hearing pastness and presence: the myth of perfect fidelity and the temporality of recorded sound

ABSTRACT

This article rethinks the experience of listening to technologically reproduced sound and music by dispelling the myth of perfect fidelity or the ideal of complete similitude between originals and copies. It does so, on the basis of a media archaeological analysis of the symbolic idealisations used in mathematical acoustics and the physical processes that turn these idealisations into media technological operations. Contrasting Friedrich Kittler’s media theoretical take on the idealised sonic purity of the sine wave with Jacques Derrida’s epistemological concept of the temporal “delay” that defines all mediatic operations, the article argues that the inevitable introduction of transient elements – noise, distortion and randomness – shapes the listener’s experience of the multi-layered temporality of recorded sound and music. By no longer focussing on the input and output of a recording chain but on the transmission channels in between, it is argued that recorded sound and music simultaneously marks pastness and presence. Pastness in the sense that sound recordings resonate with the transience, temporal irreversibility and finitude of all physical phenomena; and presence in the sense that they also produce the experience of the constant flow of time through the here and now.

KEYWORDS:

• Sound recording
• noise
• media archaeology
• acoustics
• uncertainty principle
• Kittler
• Derrida

“The laws of physics are like musical notation, things that are real and important provided that we do not take them too seriously.”

- Wiener (1976, 545)

In March 2017, during a performance at the University of Cambridge, I first listened to a phonograph recording.¹ Three musicians played a short excerpt of Wagner’s Ride of the Valkyries into the recording horn, while a wax role was spinning and the needle faithfully etched the soundwaves in its surface. When the wax had dried, the recording was played back. It sounded faint, thin, shaky and noisy: sound qualities one tends to associate with a recording that was made a hundred or so years ago. This sound, however, had been recorded by the very same machine that played it back to us; in the very same room, and only a few minutes earlier.

Quite unexpectedly, and despite my knowledge of the workings of this machine, I found the experience profoundly moving. It made me realise how, in today’s world, the experience of listening to technologically reproduced sound has become so overwhelmingly mundane that one hardly ever thinks about it. As it reproduces sound without electricity, cables, microphones or loudspeakers, however, the phonograph, this simple yet ingenuous mechanical device, reconnected my sonically overstimulated ears with the exceptionality of that experience. Listening to sound waves captured and (re)produced by the first type of machine to ever do so, made me ask once again and with renewed curiosity what this experience of listening to technologically produced sound actually means.

At first thought, one might focus on the experience of sonic historicity. In the case of the phonograph recording, its thin, shaky sound, combined with the clicks, hums and noises of the device itself, seem to sonically confirm that that we are, in fact, listening to something that happened in the past. Indeed, the very process of sound reproduction seems to contain a latent promise of capturing time: taking a singular acoustic event in full and holding on to it forever. Especially when compared to the almost seamless and thus mostly undetectable digital operations carried out by the twenty-first century black boxes in our pockets, the audible traces of the phonograph’s operations are loaded with this, as of yet unfulfilled, promise of complete sonic replication. After all, this argument goes, if sound reproduction originated with these relatively simple machines that transmit, store and replay physical sound waves with limited spectral resolution and limited dynamic range, and developed into highly complex technologies that record, transmit, analyse, manipulate, produce and reproduce the physical characteristics of any sound with a definition that equals the input in all its sonic complexity, is it not logical to think that we are slowly but steadily approaching what might be called perfect replication?

On the contrary: precisely because of this increasing technological complexity and accuracy, and especially the narrative of continuous progress associated with it, the discourse on sound reproduction remains under the spell of an ideal that goes back to its earliest days. Sterne (2003, 218) describes this ideal as the “desire to capture the world and reproduce it ‘as it really is,’” based on the doctrine of the “vanishing mediator.” If only the technical components are improved, this doctrine tells us, if only the recording process is streamlined, if only all external noise is reduced, the precision is increased, the bandwidth is widened; if one only prevents, eliminates or at least maximally reduces all traces of the medium itself, the complete capture of any sonic event in real time is within reach. By presupposing an intrinsic, unbroken relationship between input and output, according to which the latter can and thus eventually will fully coincide with the former, this myth of perfect fidelity, as I call it, upholds the idea that complete similitude between originals and copies is within the realm of the possible.

This myth is very much part of the modernist assumption that, as Latour (2010, 482, emphasis in original) puts it, “although every state of affairs deploys associations of mediators, everything is supposed to happen as if only chains of purely passive intermediaries were to unfold.” Instead of such techno-optimism, a more nuanced understanding of technologically sound reproduction should account for the basic information theoretical observation that, as Serres (1982, 150) writes, “every relation between two instances demands a route. What is already there on this route either facilitates or impedes the relation.” What this means is that, in contrast to what the myth of perfect fidelity tells us, the input and output of a recording chain are idealised notions that reduce complexity and conceal the influence of the route itself.

Following the principles of information theory, each channel in between “two instances” (source and destination, transmitter and receiver) affects the signal in specific ways. On the one hand, this influence must be repressed to enable the signal to get through. On the other hand, the physical presence of a channel is indispensable for the transmission to take place at all.² This is the fundamental dilemma at the heart of all signal transmission: what is required for successful transmission also impedes its accuracy. To understand technologically reproduced sound, one should therefore not look at the beginning or the end of the route and “deny the gap between causes and consequences” (Latour 2010, 483, emphasis in original), but focus on everything that happens during the journey itself.

More concretely, in their book on aural architecture, Blesser and Salter (2007, 150) describe music “as a sonic energy package” (a number of compound sound waves) “that progressively passes through a series of passive acoustic objects” (like walls, air, musical instruments, furniture, etc.), “each of which then radiates and couples energy to other acoustic objects, and eventually to listeners.”³ By the time any technologically mediated sound reaches our ears, this “sound energy package” has typically travelled through and refracted from a series of such passive acoustic objects, but also travelled through many passive and active technological components. Along the way, each link in this chain of microphones and walls, amplifiers and furniture, cables and air, compressors, effect modules, loudspeakers and living bodies “couples energy” to the signal, changing its frequency composition, altering the waveform and thus shaping its sonic characteristics.

Each link in this transmission chain constitutes a gate or passage through which the signal passes; and each gate or passage affects or filters the signal in a specific way. Because of this, what is heard by the receiver is not the original signal combined with some random artefacts that are added by the medium itself, but which can be erased, removed or reduced at will. Instead, the signal is fundamentally shaped by all the channels – all the filtering passages – in the transmission chain. As material traces of all transmission channels in between, the artefacts of signal transmission mark the route from the moment of recording to its potentially many moments of physical reproduction. As such, these traces are key to understanding technologically reproduced sound.

In other words, the experience of listening to recorded sound and music cannot be explained by our marvel over technological ingenuity and the myth of perfect fidelity. Instead, as I will show in the following, technological sound reproduction is defined by the somewhat paradoxical combination of, on the one hand, the repetition of sound events that took place in the past, and, on the other hand, the unmistakable presence of all technologically reproduced sound in the present. Throughout the ages and across cultures, this combination of the inherent pastness of sound (the fact that all sound is always already gone) and its continuous unfolding in the here and now (the fact that all physical sound is heard in the present), has been central to musical experiences. By emphasising precisely this double temporality even more urgently, technologically sound reproduction evokes a radical sense of sonic presence whilst simultaneously reminding us of the impossibility to fully capture and reproduce the flow of events.

The plane of the ideal filter

According to the myth of perfect fidelity, the ultimate goal of technological sound reproduction is to produce perfect copies.⁴ Any distorting or disrupting effects should be prevented, reduced or eliminated, because only when all random, transient and unexpected events are expelled, complete control over and exact knowledge of the process becomes possible. The level of such control and precision, however, is fundamentally restricted, because at the most elementary physical level, a compromise or trade-off between incompatible extremes is required. This compromise is the result of two closely related uncertainty principles that – like Heisenberg’s quantum mechanical uncertainty principle of 1927 – limit the possible accuracy of reproduction or representation. As soon as accuracy in one domain increases, it decreases in another; and vice versa.

The first of these principles is related to the irreducible presence of random background noise.⁵ It constitutes a trade-off between the frequency bandwidth of a transmission channel and its sensitivity to very faint (low amplitude) signals. Such signals do not carry enough energy to exceed the minimum amount of background noise (or noise floor) and are drowned out. These signals could be amplified, but amplification always raises the volume of the noise floor as well: ultimately, noise will inevitably overtake the signal. Alternatively, one could instal some filter that narrows the channel’s bandwidth and filters out noisy frequencies. In the case of broad band signals – like most music – however, such a filter almost always also affects the signal itself. In other words, it would not only filter out “unwanted” noise but remove part of the signal as well.

In the hypothetical extreme of this second scenario, an infinitely accurate filter might filter out all frequencies and produce just one, pure, noise-free frequency. This single frequency, however, would convey just as much (or just as little) information as the noise that overtakes the signal in the amplification scenario: it confirms that some signal is there, but because all its unique characteristics disappeared, there is no way of telling what kind of signal it is. Following the first uncertainty principle, then, the wider the bandwidth of a channel, the less sensitive it is to low amplitude signals. Conversely, the more sensitive a channel is to low amplitude signals, the narrower is its frequency range.

The second principle follows directly from the first. It constitutes a trade-off between the system’s sensitivity to very faint (low amplitude) signals and its sensitivity to their duration. Significantly, it thereby also involves the factor of time. Because any system requires a minimum amount of time to process an incoming signal, every output is produced with a slight delay. Like the uncertainty relation between the bandwidth of the channel and its sensitivity to low amplitude signals, this delay is “proportional to the narrowness” of the channel (Moles 1966, 86). In other words, the time between input and output increases and decreases proportionally in relation to the channel’s bandwidth: the narrower the bandwidth, the more frequencies it filters out and the more time it requires to produce the output (Moles 1966, 86).

The frequency spectrum of most natural signals (music and speech included), however, does not remain stable, but changes rapidly and continuously. If such a change happens faster than the minimal processing time of the system, it goes unregister. This means that the duration of the signal is not processed correctly. Following the second uncertainty principle, then, a gain in sensitivity to low amplitude signals goes at the cost of the sensitivity towards their duration. Most significantly, at the limit case of this second trade-off, in which the aforementioned infinitely accurate filter would filter out all noise to produce just one, noise-free frequency, the processing time would become infinite. A pure, noise-free signal, in other words, will never appear.

This single, noise-free, idealised frequency produced by an absolutely accurate but infinite filtering operation is called a sine wave. Following the uncertainty principles, the closer any system comes to producing a perfect sine wave, the narrower its bandwidth should be, and the more time it requires to complete the operation. Only as the temporal factor $t$ (in mathematical terms) or the response time of the filter (in engineering terms) tends to infinity, a sine wave would appear. The ideal sine wave is therefore positioned on one extreme of the uncertainty relation between time and frequency. Squeezed through an impossibly narrow frequency filter and cleansed of all possible noise, pure sine waves stretch endlessly into the past and the future. Despite (or perhaps exactly because of) this theoretical purity, a sine wave has no physical existence.

If we widen the bandwidth of our ideal filter, we allow a larger spectrum to seep through. This will shorten its response time and reduce its delay: the temporal factor $t$ turns back from the idealised $t=\mathrm{\infty }$ into a finite, real-world timeframe, with a beginning and an end. Continuing this process, gradually widening the bandwidth and shortening the response time, one ultimately arrives at the other extreme of the uncertainty relation between time and frequency: the Dirac impulse or delta function.⁶ In this case, the delay of the filtering channel is reduced to zero, which means that the timeframe of a Dirac impulse is infinitesimally short. The signal is produced instantaneously. Because delay and response time affect precision, however, such a hypothetical instantaneous filter would not filter out anything. All frequencies would pass through unaffected. A Dirac impulse therefore represents the smallest grain of time, a pure now, at which everything happens at once. It is the exact inverse of a sine wave: the latter is a single frequency of infinite duration, the former an infinite frequency spectrum at an infinitesimally short time.

An ideal reproduction system with perfect fidelity would reproduce signals with absolute spectral and temporal precision. It operates on the plane of ideal filters, combining the spectral clarity of sine waves with the temporal precision of delta functions. Achieving perfect accuracy and unlimited resolution, its operations would administer what can be called a “clean cut” that seamlessly removes the singular sonic event from the flow of time and turns it into infinite series that stretch out in both dimensions: temporally infinite, like ideal sine waves, and spatially or spectrally infinite, like Dirac impulses.⁷ The limit imposed by the uncertainty principle, however, marks the difference between such an ideal system and the capabilities of technical media that process signals in physical reality, or the domain of technical filters.

The domain of technical filters

In the domain of technical filters, an infinitesimally small event can never be analysed in full, because it contains an infinite number of frequencies; and an entirely determinate spectrum can never be fully processed, because it would require an infinite amount of time. Whatever we gain in one dimension, we lose in the other. Sound signals in the domain of technical filters extend somewhere in between the two extremes of infinite sine waves and infinitesimal Dirac impulses.

In an essay first published in 1967, physicist and cybernetician Wiener (1976) recounts how, at a talk in Göttingen in 1925, he used a musical example to explain the consequences of this physical balancing act. Every waveform, he writes, occurs within a finite timeframe. It cannot be infinitesimally short, because it must at least be long enough to allow the waveform to complete one full cycle. In the case of complex harmonic waveforms – like most musical sounds – a listener can generally identify its frequency spectrum when the fundamental frequency completes at least one cycle. As Wiener explains: “if you take a note oscillating at a rate of sixteen times a second and continue it only for one twentieth of a second, what you will get is essentially a single push of air without any marked or even noticeable periodic character.” When the fundamental frequency of the waveform is cut short, it “will not sound to the ear like a note,” but comes across as a short, transient noise, blow or impulse (545).⁸

This is the uncertainty principle at work: beyond a certain limit, tending towards (but never reaching) the infinitesimally short timeframe of a Dirac impulse, it becomes physically impossible to shorten a sound without losing its sonic identity. Beyond this threshold, the clarity of sine waves gradually gives way to the instantaneity of Dirac impulses. Clearly definable frequencies turn into fuzzy, undefined spectra, until all that remains is a transient blow, pip or noise. Still, such fuzzy, non-periodic transients are unavoidable, because every physical signal has a beginning, a longer or shorter duration, and an end. Even an almost periodic signal does not continue forever; at some point it will stop. Precisely these starts and stops (being disruptions of absolute periodicity) Wiener (544–545) writes, cause “an alteration of [the] frequency composition which may be small, but which is very real.” The beginning and end (or, in musical terms, the “attack” and “decay”) of every sound adds elements of non-periodic transience to even the most periodic signals.

The absolute spectral purity of sine waves is always shot through with transient events that negate absolute periodicity; and precisely these “small, but very real” alterations of the frequency composition give each sound its uniquely identifiable timbral quality. Whereas periodic frequencies are largely responsible for determining pitch and overall harmonic composition, the specific tone colour (but also, in the case of speech, specific vowel colour and consonant shape) is just as much determined by those non-periodic components that composer Cowell (2004, 23) calls the “noise element in the very tone itself.” As crucial determinants of differential meaning for auditory communication, these elements – this “noise of sound” – sonically mark the fundamental difference between the plane of the ideal filter and the domain of technical filters.

“Just as the gods confined us to finite lives in the temporal domain,” Kittler (2006, 72) writes ominously, “our bodies restrict us to a limited spectrum in the immeasurable range of frequencies.” Inspired by our persistent desire to overcome such bodily limitations, human beings designed machines that are able to go beyond the limited capacities of their senses, to process and analyse frequencies their ears cannot hear and their eyes cannot see. On the one hand, these machines capture, produce and reproduce physically real signals that bear a statistically non-arbitrary resemblance to those they are supposed to reproduce; and with every new, even more sophisticated machine, the copies seem to resemble the originals even more. The physical limitations posed by the uncertainty principle, on the other hand, also assure that even these machines cannot instantaneously process infinitesimally detailed, real-time events. The output is always marked by the processing channels in between.

In contrast to the clean cut of an ideal filter, real-time operations in the domain of technical filters require a compromise between sine waves and Dirac impulses. Precisely this compromise becomes audibly apparent in the form of the traces left by every link in the recording and reproduction chain: the non-periodic oscillations of transient events that add random and unpredictable sonic artefacts to the signal. As they cling to any sound on its route from input to output, these transient elements introduce a certain, if only the slightest, amount of fuzziness that negates the possibility of a clean cut and dispels the myth of perfect fidelity.

Because of the physical effects of the uncertainty principle, any technologically processed sound contains traces of everything it encountered, whether natural or technical, whether air, copper or glass fibre; whether acoustically, electro-acoustically, electronically or digitally – and of the specific circumstances at which it was encountered: humidity, air pressure, altitude, etc. etc. Each passage or gate adds transient noises to the signal that fundamentally preclude any possibility of absolute periodicity and create an irreducible acoustic difference between consecutive sound. As they make each moment sonically different from the next, the transient traces of the filtering channels therefore not only generate spectral differences. Most importantly, they emphasise the irreversible flow of time.

The pastness of sound

Eight months before his death in October 2011, during the after-talk of a lecture on music and mathematics in Cologne, Kittler (2012, 48) characterised the plane of the ideal filter as follows:

In the time domain, we are mortal, and in the frequency domain, in the Fourier domain, we are immortal. […] It is the essence of the sine and cosine that they do not have a beginning or an end and are therefore immortal.⁹

By redefining the infinite timeframe of sine waves in terms of this almost theological discourse on mortality and immortality, Kittler offers a different perspective on the temporality of technologically reproduced sound. The word “immortal” points in two directions that initially seem contradictory or even mutually exclusive, but eventually prove to be complementary. On the one hand, the rhetorical paring of the all-too-human dream of immortality and the idealisation that is necessary for the mathematical analysis of physical signals connects the symbolical purity of sine waves to spiritual motifs of heavenly purity and eternal life. On the other hand, the invocation of “immortality” just as much emphasises the physical impossibility of such mathematical idealisations, and their fundamental absence in the domain of technical filters.

“When we measure frequencies,” or when we apply mathematical Fourier analysis to symbolically shift from the time domain to the frequency domain, Kittler (2006, 69) writes, “we are on the other side of death, in an immortality that has replaced the old gods.” By interpreting the infinite timeframe of the sine wave as a form of “immortality,” the mathematical “infinite” is reinterpreted as the more theological “eternal.” As such, the absolute a-temporal periodicity of the frequency domain is turned into an “eternity” – the temporality of gods. As a reviewer of the newly invented phonograph (cited in Kittler 2015, 105) famously noted in 1877, with the invention of sound reproduction, “speech has become, as it were, immortal.”¹⁰ When one emphasises this superhuman potential of technology, sound media indeed seem to strive towards the infinite clarity and perfect repetition of the Fourier domain.

However, Kittler (2012, 48–49) continues his argument, although “it is the essence of the sine and cosine that they do not have a beginning or an end, this property is quite annoying, as we do not only want to know frequencies, but events as well; for instance, when something has taken place.” Despite all its analytical prowess, frequency analysis only provides information in one domain at the expense of information in the other. Indeed, in a famous paper from 1947, physicist Gábor (1947, 591) already noted that the fact that “sound has a time pattern as well as a frequency pattern finds no expression either in the description of sound as a signal $s\left(t\right)$in function of time, or in its representation by Fourier components $S\left(f\right)$.” In other words, spectral analysis only provides half the story. Gábor suggests a way to include the other half as well: by chopping a signal into very small bits or “windows” and plotting the frequency information of each of these windows on a temporal axis, one can account for both time and frequency. Due to the uncertainty principle, however, even the accuracy of this time-frequency analysis remains limited. The windows cannot be as short as an ideal Dirac impulse and also require a negotiation between representing complete spectra on the one and exact durations on the other.

Following Gábor’s observation, Kittler’s gesture to link the infinity of the frequency domain to concepts like eternity and immortality, whilst subsequently stating that this same infinity is “quite annoying,” therefore stretches a crucial observation: any promise of infinity or immortality in the frequency domain only makes sense in relation to the fundamental limitations of the domain in which things have a limited duration and in which life is, indeed, mortal. The conceptual limit cases in the frequency domain (sine waves) and time domain (Dirac impulses) thereby show how such mathematical idealisations always already presuppose their own negation. Their possibility on the symbolic plane of the ideal filter only exists by virtue of their impossibility in the physical domain of technical filters. Although analysis might suggest that a physical signal endlessly tends towards the limit case of the analytical representation, precisely the signal’s inability to coincide with this asymptotic limit underscores its physical existence in space and time. Understood like this, Kittler’s conceptualisation of the immortality of perfect sine waves can only be properly understood by taking into account the fact that this immortality always also evokes its opposite. In short, the immortality of sine waves only becomes conceptually meaningful in contrast to the finitude that marks everything in the time domain.

Only when time is infinite, all transience is reduced and “paradise,” as Serres (1982, 68) writes, “then is there.” By contrast, as long as transients remain present, the timeframe cannot be infinite, sine waves are not immortal, and the eternity of paradise remains forever out of reach. Hence, the traces of the impossibility of a clean cut, which are so important for the specific character of each sound, sonically negate the possibility of divine eternity, heavenly immortality and temporal infinity. They accentuate that a signal at some point began and will eventually end, thus confirming the irreversible flow of time, the fundamental inaccessibility of eternity, and the finitude of all physical phenomena. All signals decay and die out. Time flows irreversibly in one direction. So, if the symbolic sine wave signifies infinity and immortality, the transient noise of sound, as sonic marker for the physical cuts that produce every physical signal, signifies finitude and mortality.

When one assumes the possibility of a clean cut, the magic of sound technology holds the promise of immortality. “As long as a turntable is spinning or a CD is running,” Kittler (2006, 68) imagines, “an old magic emerges despite the fading of years, hair and strength. Time stops, what more do hearts want?” Time, however, does not stop; and that which, under bright theatre lights and accompanied by dramatic music, looks and sounds a lot like magic always turns out to be nothing but simple trickery. Besides techno-religious dreams of immortality, Kittler’s conceptualisation of the immortal sine wave therefore also points in another direction, as the word immortality implicitly connects the a-temporality of the frequency domain to its conceptual opposite: mortality. The figure of the ideal sine wave also carries with it a sense of the temporal finitude of the physical world.

Whether recorded years, decades or more than a century ago, sound waves only physically exist in the present. Until they are transduced back into sound waves at the moment of playback, signals that are acoustically, electro-magnetically or digitally stored on some kind of storage medium are just grooves, magnetised particles or pits in a surface. Despite this undisputable physical presence of any sound in the here and now, however, the subtle, non-periodic traces of its journey through a great many technical channels also mark its pastness (Ernst 2012, 22). In contrast to the infinite stasis, clarity and periodicity of the frequency domain, and the immortality promised by the infinite of sine waves, the ever-present noise of sound, tending towards the absolute transience of a Dirac impulse, continuously pushes time forward. In the final analysis, this randomness and transience thereby resonate with the irreversibility of time and the fact that, as Prigogine (1997, 62) puts it, “we are all transient.”

The presence of sound

Besides this decisively melancholic side of technological sound reproduction, the noise of sound therefore produces a different temporal experience as well. The technological possibility of playing a recording backwards notwithstanding, as reminders of the complexity of a world in continuous flux, the sonic traces of each link in the technological chain continuously emphasise the unidirectional flow of time and, by extension, the fundamental presence of the present.¹¹ In his late essay Athens, Still Remains (2010), Derrida discusses this continuous, yet unrepresentable presence of the present, to make a case against what he calls the Western philosophical “tradition of being-for-death” from Socrates to Heidegger (59). He does so in the form of an extensive analysis of the click of the photographic shutter and the multi-layered temporality of photographic images.

Implicitly following Barthes’ (1981, 27) concept of time as “punctum,” Derrida (2010, 3) initially approaches photographic capture – freezing an moment in time and space – as an act to which the photographer always arrives too late. At the very instance the shutter opens and the picture is taken, the moment itself already passed. As a testament of the attempt to capture what is always already gone, Derrida writes in Heideggerian fashion, the photograph therefore confronts its viewer with the fact that life is nothing but a “temporary reprieve” from the time when one is no-longer alive (29).¹² Because it produces images that inherently depict things that already happened, and are no longer happening, photography represents the transience of life and the realisation that each passing moment only “suspends the coming due” (27). As this frozen capture of what is always already gone, the photograph reminds us of the catastrophe of death. It thereby emphasises how we, as Derrida puts it, always exists in relationship to a fundamental delay between the moment something takes place and the moment we can process it (17). As soon as we grasp the presence of the present, it already passed. Unable to represent the here and now at the very moment it takes place, we can only process or represent the past.

The “click” of the photo camera, the short time between pressing the release and the closing of the shutter, constitutes the cut that defines this delay. It is not a clean cut: it tends towards the impossible instantaneity of a Dirac impulse – the ideal now – but never fully converges with it. Following the logic of the uncertainty principle, Derrida’s ontological delay is thus analogous to the response time and delay of a technical filtering operation. The uncertainty principle postulates that the response time of any filter cannot be zero, because nothing happens instantaneously. Only in the idealised instance of a Dirac impulse, when the timeframe is infinitesimally small and the frequency range infinite, this delay would be absent. In real time, the now can only be stored as the no-longer-now, because we are always already too late to grasp it. As a consequence, we are left with an impression in our memory or some representation inscribed on media hardware.

In the final section of Athens, Still Remains, however, Derrida introduces an alternative perspective. He shifts the focus away from the inherent lateness of the photographic image and the impossibility of the clean cut to conceptualise the “at-present of the now” by exploring the possibility to rethink “instantaneity on the basis of the delay” (2010, 17, emphasis in original). This rethinking of the unrepresentable experience of the present and its relation to its inevitable pastness implies a different approach to the very short instance between the opening and closing of the shutter. It requires a focus on the point-like moment of the cut itself to develop a perspective that is diametrically opposed to the static infinity of sine waves. By temporarily suspending the inherent delay that is contained in the photographic image and zooming in as close as possible on the moment of capture itself, we can ideally, Derrida writes, attempt to “refuse [the] debt” that points to the inevitable coming due (63). This exercise allows us – if only for an infinitesimally brief instance – to pay attention to and stay with (or within) the moment of the click itself.

With this conceptual move, our analysis can shift from Kittler’s preoccupation with immortal sine waves to the inherent temporality of all sound. When one refuses to acknowledge its asymptotic impossibility (the fact that it tends towards, but never attains the infinitesimally short timeframe of a Dirac impulse), the click or cut represents the unimaginable nowness of every moment. At each successive, infinitesimally short instance, the end is kept at bay and we can try to be, as Derrida (2010, 63) puts it, “an innocent living being who forever knows nothing of death.” Radically unconnected to past and future, this is the perfect transient experience, the ideal event. It comes and goes as instantaneously as a flash of lightning. Like a Dirac impulse, it contains an infinite amount of information; too much to process or filter in any limited amount of time. As such, such pure transience cannot be captured, nor reproduced. It only exists, can only exist, in the radical present. The ideal Dirac impulse thus mathematically represents an unfathomably briefly, non-reflexive moment which is nothing but present. At this infinitesimally short moment, Derrida concludes poetically, the inherent transience of life can be ignored, and we can imagine to be “infinite […]. We are infinite, and so let’s be infinite, eternally” (63).

By pulling us away from dreams of complete replication towards a focus on the infinitely complex and indivisible present, Derrida hereby offers a compelling counterpoint to Kittler’s take on the stasis of the Fourier domain. Although “we are infinite” seems to echo “we are immortal,” they are not the same. They are opposites. Derrida’s analysis of the click of the shutter and call to be infinite, understood in terms of the temporally uncorrelated transience of a Dirac impulse, capture an aspect of the technological sound reproduction that Kittler’s emphasis on immortality and sine waves does not. Following the uncertainty principle, if time contracts to the most infinitesimally short instance, the corresponding frequency spectrum becomes infinite. Understood in terms of the perfect Dirac impulse, the frequency spectrum of this pure, infinitesimally short moment must therefore be infinite. Whereas Kittler’s “immortality” is spectral, Derrida’s “infinity” is temporal.

The infinity of the Dirac impulse and the immortality of sine waves are idealisations at opposite extremes of the uncertainty principle. Besides the melancholic pull of pastness caused by the cuts of physical filtering operations, introducing transient traces that underscore the inability to achieve the purity of ideal sine waves, the randomness of these non-periodic transients also produces a continuously evolving sonic present that contains the promise of an infinite now. In the domain of technical filters, the physical reality of sound technology, the combination of, on the one hand, the transient temporality of the moment of the cut itself and, on the other hand, the impossible dream of infinitely pure sine waves confirms the fundamental correlation between the radical presence of each sound in the here and now and the simultaneous experience of its inherent historicity.

Listening to passed and passing time

The “destruction of the delay,” Derrida (2010, 51) writes, is “the very desire of philosophy.” It also motivates the rationalist desire of positivist science: the creation of models that perfectly fit the things they represent, without gaps, disruptions, breaks or discontinuity. It thereby encapsulates the wish to immediately and completely process all events at the very moment they occur; the dream to give everything, down to the most infinitesimally small detail, its proper time and place; and ultimately, the desire to ward off the imminence and unpredictability of death. Like Kittler’s dream of immortality, this dream of a non-self-reflexive subject outside of the delay is therefore purely symbolic. It is sought after like the fountain of youth but can only be realised in theory – in mathematical formulas or philosophical stories. As such, it is the very essence of the myth of perfect fidelity.

If we were able to wallow in the eternal presence of an infinitesimally short instance, we would never have time to pinpoint what was what; and to analyse, process, or define the event. Would we be able to grasp every frequency in perfect clarity, time would never progress and everything would forever stay the same. The delay in the response time of technical filters is therefore not just a physical or technical limitation of media processes, which may or may not be overcome at some time in the future. It postulates, as Peters (2008, 11) puts it, “an ontological point about the nature of things and an ethical point about the uniqueness of every act.” Because of the absolute physical limit imposed by the uncertainty principle, “all empirical representation,” Peters writes, “both depends on and crashes into the wall of finitude” (19).

Not unlike the age-old Pythagorean story of celestial harmony, immortal sine waves suggest the purity of perfectly periodic frequency spectra. The transient noise of sound, in contrast, tends towards the infinite spectral complexity of a Dirac Impulse. Physical sounds exist somewhere on the continuum between the two; between sine wave and Dirac impulse, infinite time frames and infinite frequency spectra. When a technologically reproduced sound arrives at the ears of human listeners, it includes sonic traces of everything it encountered in between its first attack and its ultimate decay. Given the specific physical properties of each channels and the unique conditions of every transmission, each playback is irreversibly singular. In the specific frequency composition of every signal, we hear its journey over space and time.

On the one hand, all these transient elements that stick to the signal highlight the impossibility to capture what Derrida calls the “at-present” of the now and represent or reproduce an event exactly as it occurred. They draw attention to the fact that, although a sound is unmistakably present in the here and now, it has travelled over time and through many different places. It was captured, cut off from its origin and changed. On the other hand, however, the same transient singularities – traces of the physical cuts of technical filters – resonate with listeners in the present. They confirm the radically contemporaneous passing of all sound through the now. In other words, noise, distortion and randomness mark passed time and passing time. Passed time in the sense that, as indexical traces resonating with the temporal irreversibility and transience of all physical phenomena, they affirm the inherent finitude of any signal. They mark passing time, on the other hand, in the sense that these traces of the materiality of communication also confirm the signal’s constant flow through the now, producing the experience of a continuously renewed sonic presence.

“It is precisely under mediatic conditions,” Kittler (1992, 68) explained in 1992, “that what cannot be processed, what is impossible, is brought into ever sharper focus.” This sharper focus on “what cannot be processed” defines the operations of technological sound media as well. The symbolic representations produced by ideal filters assume the complete reduction of all the noise of signal transmission. The channels of technical sound media, by contrast, apply physical cuts that shape and change the signal in ways that fully belong to the recording and reproduction process itself. These traces of everything that happens “in between” only appear at the very moment the signal is produced, which is why they cannot be represented by anything but themselves. They constitute precisely what Kittler calls “that what cannot be processed.” In other words, what is “brought into ever sharper focus” under the mediatic conditions of sound technology, is the fundamental fuzziness caused by the uncertainty principle: the structural entanglement and infinite divisibility of the relation between time and space.

Striving for clearly delineated spectra and infinitely oscillating sine waves, the myth of perfect fidelity adheres to a rationalist ideal of a world in which every part of every sound has its proper and unchanging place. Assuming a clean cut that separates signal from noise, it presupposes the possibility of a perfect filtering operation, much like the one through which Western classical music theory upheld its symbolic separation between “music” and “noise” for centuries.¹³ As part of what Latour (2010, 473) calls “the modernist grand narrative of Progress,” this symbolical suppression of all artefacts of signal processing serves to symbolically connect a supposedly “original” input with an ideally identical output and deny the existence of the in between all together.

Truly understanding the experience of listening to technological reproduced sound and music, however, requires a break with this dominant, modernist myth of perfect fidelity. It requires us to fully come to terms with the fuzziness of everything that happens on the journey itself. Instead of assuming an unbroken connection between input and output we should turn our attention to everything that emerges on the crossroads between the two extremes of the uncertainty principle. Doing so means to leave the idealist figures of sine wave and Dirac impulse for what they are and embrace the complexity of things that exist in the middle.¹⁴ Contrary to the infinite purity of rationalist models, such a focus allows us to consider the agency of the material medium itself and how it defines the ways in which technologically (re)produced sound generates a continuous push and pull between pastness and presence.

Whereas the sheer presence of a sound signal (or the fact that we can hear it unfold in the present) physically confirms that a filtering operation has taken place, the very moment of filtering, the moment of its reproduction or transmission itself, always escapes our grasp. Still, the sonic traces that inevitably shape the signal reveal that this filtering operation took place. Because this fundamental filtering capacity of the channel is a prerequisite of all signal transmission, media scholar Donner (2006, 25) writes, “it actually makes sense to talk about a media-filtered perception.”¹⁵ To acknowledge the irrepressible influence of mediatic filtering operations on the sounds we hear and the way we hear them, the idealist logic supporting the myth of perfect fidelity should therefore be replaced by a fundamental logic of filtering. This foregrounds the operations of the channel itself as the primary point of reference for understanding how media shape the sound of technologically reproduced music.

Whether it is heard in the control room of the music studio, in the comfort of one’s own living room, while driving in a car or dancing in the club, the signal at one end of the chain is both radically different and fundamentally the same as the signal that went in. Radically different in the sense that its spectral contours and temporal flow are singularly unique in comparison to those that went in; and fundamentally the same in the sense that, regardless of this difference or similitude, it is just as physically real and present as the input signal. The transient traces left behind by the cuts of technical filters thereby confirm the primacy of the unrepresentable moment of filtering. They put an end to the idea that sound recordings are incomplete or flawed reproductions of some “original” sonic event. Instead, the logic of filtering emphasises that a technological produced sound is never an ideal replication of some supposed “original,” but always a singularly complex and essentially new sound altogether.

Acknowledgments

For Doris. This article was made possible by the support of research project Sound and Materialism in the 19th century, hosted by the Faculty of Music, University of Cambridge, as well as Corpus Christi College, Cambridge and the Amsterdam School of Cultural Analysis (University of Amsterdam). Many thanks to David Trippett, Melissa van Drie and the two anonymous peer reviewers for their valuable commentary, remarks and suggestions. I also want to thank my other colleagues in Amsterdam, Cambridge and elsewhere. Lastly, this essay would not exist without the invaluable support of Sander van Maas.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

This work was supported by the H2020 European Research Council [638241].

Notes on contributors

Melle Jan Kromhout

Melle Jan Kromhout works on the intersection of musicology, sound studies and media studies. His work focusses on the conceptual relations between music, sound and media from the nineteenth century to the present. He completed a PhD at the Amsterdam School for Cultural Analysis (University of Amsterdam) and recently worked as postdoctoral research fellow at the Faculty of Music and Corpus Christi College, University of Cambridge. His first monograph, The Logic of Filtering. How Noise Shapes the Sound of Recorded Music, is forthcoming with Oxford University Press.

Notes

1. At the Cambridge Science Festival 2017, I joined a project initiated and put together by Melissa van Drie, in which a group of researchers and artists performed several pages from Cornelius Cardew’s Treatise, a 193-page graphic score that can be interpreted in any way one sees fit. As part of the performance, British artist and researcher Aleks Kolkowski brought one of the phonographs from his collection of early recording devices to make a recording.

2. In Claude Shannon’s information theory of 1948, noise is not conceived as external disturbance but considered to be internal to the communication system. Although this allows one to calculate the amount of noise that accumulates during a transmission, it also shows that complete noise reduction is fundamentally impossible. A clear and concise introduction to these principles in the context of audio technology can be found in Sterne (2012, 81).

3. In Earth Sounds, Kahn (2013, 62, emphasis in original) uses the term transperception “to denote the perception of those characteristics” acquired “through the course of their propagation, acoustically and electromagnetically.”

4. On the history of the concept of fidelity, see for instance (Sterne 2003, 221/276; Thompson 1995, 131–171; Siefert 1995, 417–449).

5. Moles (1966, 84) writes that Einstein proved how “in the last analysis, background noise is due to the agitation of electrons in conductors.” This means that random noise is present down to the level of elementary particles, and proportional “to the absolute temperature.”

6. The Dirac impulse or delta function is named after British physicist and mathematician Paul Dirac or the sign that represents the function, the $\delta$.

7. Siegert (2003, 251) describes the impossible temporal exactitude of the infinitesimally short Dirac impulse as a “cut that freezes the movement.” In his discussion of Kant’s aesthetics in The Truth in Painting (1987, 89), Derrida writes about the “sans of the pure cut” (“le ‘sans’ de la coupure pure”) or the paradox between the finitude that is inherent to the application of a cut and the ideal infinity of a pure cut that leaves no traces of its cutting. In a different context, Barad (2007, 114) discusses the problem of separating observer and observed in quantum mechanics: “So the question of what constitutes the object of measurement is not fixed: as Bohr says, there is no inherently determinate Cartesian cut. […] The apparatus enacts a cut delineating the object from the agencies of observation.”

8. Kittler (2006, 71) also notes that “before a deep organ tone can turn into an event, many high trebles have already been recognized.” He does not credit a source, but it seems highly likely the example is from Wiener’s article.

9. English translations of untranslated German sources are my own.

10. Kittler and many others attribute these words to Edison himself, but Sterne (2003, 298) notes that they actually appeared in an 1877 editorial comment on Edison’s invention in The Scientific American.

11. It is possible to technically reverse a sound, but this procedure always leaves an audible mark, because it turns the attack into a decay and vice versa. Even though one hears the reversal of a signal’s temporal flow, time itself is still experienced as irreversibly flowing in one direction.

12. Following Heidegger’s notion of Dasein’s being-towards-death, the possibility of the event of death, which remains beyond representation, separates the domain of technical filters that administer a physical cut from the plane of the ideal filter, administering a clean cut. Just like the awareness of the, as Heidegger (1962, 310) puts it, “indefinite certainty” of death highlights the “not-yet” of not having died yet, the transient presence of sound signals resonates both with the inherent finitude of life and with the current being-alive of Being.

13. Up to and including Hermann von Helmholtz’s mid-nineteenth century work on sound and hearing, the separation between musical “sound” and unmusical “noise” was a basic fundament of Western music theory. Although Helmholtz (1875, 101) acknowledges that non-periodic noises accompany most instrumental sounds and “facilitate our power of distinguishing them in a composite mass of sounds,” he still consistently differentiates between (periodic) “musical tones” and (non-periodic) “noises” (11–13). For Helmholtz as well, the periodicity of musical tones supports the ideal of well-ordered music, which is diametrically opposed to the non-periodicity of noise.

14. Serres (1982, 65) famously calls “what is between, what exists between,” or “the middle term,” a parasite.

15. The models produced on the basis of “so-called natural laws,” argues Flusser (2011, 46), are not objective descriptions of physical processes, but ways to process and decode the “gigantic quantity of indications, signs, clues” that we are confronted with. They symbolically create order and reduce complexity. Similarly, every transmission, reproduction or representation requires a reduction of physical complexity – a choice, a focus, a selection – that allows the signal to be transmitted.