1 The Material Heterogeneity of Recorded Sound Rick Alt man When we understood cinema as a text, we borrowed our terminology and our methodology from previously established textual domains. An understanding of cinema as event requires newMemis_andjTipdeJsJbr_a newOy^eofjrjdti-dimensiona^jinalysjs It seems especially appropriate Vs,{ to begin this retooling'process"with the development of a new vocabulary -^v«. for sound analysis, for soundjtsejf is particularly event-oriented itself. ^»J«*^-. Whereas image analysis has given us many terms and techniques fa\\y*Ll*£^ consonant with a textual approach to cinema, s^nďsheterogerieityJias .^Ji sr-much to offer to_an event-oriented aesthetic. In order to reap'"sound's-älMíí Harvest, however, weTňust takeTněw approach to sound, replacing the^^ idealist modeJsjofXered_bv_musical analysis with increased sensitivity to'-|ÍpkÍÉ3 sound's threeKiimensional materiality. -gr "current approaches to film sound systematically borrow a musical^d^jc*. model. The most influential introductory film textbook of the last decade^^L' defines the acoustic properties of sound as loudness, pitch, and timbre .^-^—^z 4 This definition is based on the apparent assumption that all film soundSp,,^^I v,Vtt h§ve lhe nature of musical notes, that is, they are single phenomena,^. produced instantaneously, emitted from a point source, and perceived inj^-je* ah immediate and direct -fashion. With a_ definition ljke „this, one, .wgjcan ^t. «^ explain many aspects of film sounďTsuch as contrasts or confluences in ^4o\,*~-volume, frequency, and tone. i*1-'? Cw In fact, since the terminology is borrowed from the realm of music, we u ^, find that with these terms we can handle almost any of the types of analysis i/^*^ typically practiced on a musical score. We note Hitchcock's suspenseful it-V'-v* diminuendo from a loud slam to muffled scratching, the harmony of Orson Welles' bass and Joseph Cotten's tenor, the melodic gifts of Cary Grant and Katharine Hepburn, the awkward timbre of Zasu Pitts and Jerry Lewis, f. 16 I Material Heterogeneity of Recorded Sound or the varied instrumentation of the "Symphony of Sounds" with which Rouben Mamoulian opens Love Me Tonight. It we could notate all film sounds according to the musical criteria of loudness, pitch, and timbre, then these three criteria would suffice for the analysis of film sound. Yet this is precisely what we cannot do. WJule all film sounds have loudness, pitch, and tmibre, not a single sound in cLnemajľärľbéadequately described with this musical terminology. In fact, not even musical sounds can be fully described with musical terminology. More appropriate for describing musical scores than individual performances, musical terminology pays littlejujaition to the details of any particular performance, concentrating instead on the coiruriön factors joining all performances of the same score. If I attend three concerts of Mozart's "Little Night Music," one in a well-upholstered salon, another in a large concert hall, and a third in a city park, I am in one sense hearing the "same" music three times, that is, music that is represented by a single, identical score. Yet how different are the sounds that reach my ears during the three concerts! Musical notation assumes that each sound is single, discrete, uniform, and unidimensionaL Stressing the formal concerns of music's internal, self-referential aspect, musical notation diverts attention from sound's discursive dimensions, concealing the fact that sound is in reality multiple, complex, heterogeneousJjuidJhree-mmensk)nal. As a concept, middle C exists independently" of space and time, in the abstract notion of abound of approximately 262j;_ycles persecond...As a reality, however, no two versions'öf middle C are identical, becausej^f the different temporal and spatial circumstances in which they originate and are heard. The middle Rick Altman I 17 ?j v'ck I ; u|ucg j^ger in which_spunds are produced. Th£ee elements are required for I tfiě~přoduction ofany sound. First, there must be vibration, such as that of the vocal cords or a violin string. Second, the vibration must take place in a medium whose molecules can be set in motion, such as air, water, or a railroad rail (sound cannot be transmitted through a vacuum). Third, the transmitting medium must absorb_and_transm.it the^original. vibrationsJn the form of changes in pjessure. When a violin string is plucked or bowed, the molecules of the surrounding medium are compressed, with the pressure passed on.f rqmone molecule tojthe next. When the string reaches the eňd^Tltslravel, maximum compression is achieved. As the string 6/^'starts back, the molecules rush back to fill up the void left by the departing vf/ ^'^'string. When the string reaches the end of its travel in the opposite H/,v*Hf. tfirection, maximum rarefactionj>ccurs. In order to create a specific, 4 ; recognizable note this- process must be repeated inj^pid succession hun- ^If'Wv.'dreds or even thousands of times a secpri(JTFor example, the G string H»i| fwtiKon a violin causes the surrounding air to go through 196 compression/ rarefaction cycles per second, commonly expressed as a frequency of 196 cps or 196 Hz. In other words, what we call the musical note G below middle C is in fact a series of rapid changes in pressure. Even taking the three-dimensional nature of sound events into account, however, this description vastly oversimplifies the situation. Whereas an electronic tone generator is capable of producing pure tones, all musical instruments produce notes that combine a fundamental frequency (such as ícíW. ikt^ithe violin's 196 Hz G string) with a senejj^fjwüa[s: harmonics (tones 'H-Ur : whose frequency is a whole number multiple of the fundamental) and C located on the first line below the G clef may be only a concept, but the l* ^n,i^>^L overtones (tones whose frequencyjs related Jo the fundamentaJ.according soundjhat we hear with our ears—whether on the street or in a movie theater—is a heterogeneous event that carries its own temporal and spatial dimensions and constitutes^ full-fledged narrative. When we listen to recorded sound we are therefore always listening to a particular account of a specific event. In order to respect the discursive complexity that is characteristic of all sound events, we can no longer continue to depend on a fundamentally conceptual terminology that remains insensitive to sound's phenomenally. Instead we must have a terminology capable both of respecting sound's heterogeneous nature and of figuring the narrative, component built into the very process of recording and reproducing sound. This article proposes such a terminology, based on a schematic but systematic review of the physical phenomenon that we call sound Sound Events: The Production of Sound What is sound? What happens when a sound is made? While this is hardly a technical treatise, it will nevertheless be useful to recall the *HMi'tö a more complex formula). Depending on the instrument and the way ifls played, the combination of harmonics and overtones can vary tremendously. When played in such a way as to emphasize the upper harmonics, for example, the violin sounds harsh and strident, while a mellow tone results from stressing instead the lower harmonics. If the oboe, trumpet, flute, and cello sound so recognizably different, it is primarily because they produce radixaUx^fferent^onibinations^ofjrjartiajs. While few people are trained to hear harmonics and overtones, most listeners will rapidly recognize their absence, as when music is played through the telephone or over an old record player with limited frequency response. While the loss of these partials reduces our pleasure in listening to music, it may have an even more radical effect on other sounds. S_poken .language becomes far more difficult to understand, voices and familiar sound effects may become harder to differentiate, even our ability to judge the distance and direction of a sound source may be impaired. In other words, the composite nature of sounds is hardly limited to music. In fact, most of the events that we think of as a single sound are not Au " í 18 I Material Heterogeneity of Recorded Sound Rick Altman 119 I > singular at all. The musical model of tone generators and violin strings is "extremely misleading. If a violin note could be produced by a violin string alone, then Stradivarius would never have become a household name. Every violin note is a complex event combining the vibrations of a string, a^wooden case, and the air trapped inside that case. Each of these three contributes to the overall tone of the note playedľ For what we call a sound is typically made up not only of multiple frequencies, but actually has multiple different fundamentals produced oyer a period of time. Think of the following familiar sounds: a refrigerator, snoring, a lawnmower, the wind, a squeaky door. We think of each as a single sound, but none is actually single in the way that an A-440 produced by a tuning fork is unitary. Each of these sounds constitutes an event taking place in time, involving multiple separate sounds organized iríinimíTiaTrŕecognizable fashion. Given the importance of rhythmic and melodic elements for our recognition of each of these sounds, it would be more appropriate to compare; thern ta.musical pJ]oses_than to injjviduai noJes. Yet even individual notes have a temporal djmension. Returning for a moment"to our violin string, consider the difference between plucking and bowing the string. In one case the sound starts suddenly, reaching its full volume extremely rapidly; in the other case the violinist seems to be sneaking up on the note, teasing the molecules into moving rather than suddenly shoving them. Whether violent or peaceful, this initiation of the sound event is termed the attack. It is followedby_the sustain Alow long is the note hold? How long does it stay at full volume? ťinallýTthe" sound fades away. This stage is called the"decay, implying not only a temporal measure but also a qualitative one. Compare/for example," the decay of a plucked string that is simply allowed to spend its own energy and the , decay of a plucked string instantaneously dampened by a finger. * I As parts of the sound envelope, the stages of attack, sustain, and decay / appľý equally to any sound event. Contrast, for example, the smooth í attack of Orson Welles' opening voice-over in The Magnificent Ambersons to the sharp attack of Georgie Minafer's dialogue. How essential to the soundscape of The Wizard ofOz is the gulf separating Margaret Hamilton's staccato attack and nearly instantaneous decay from Judy Garland's ability to ease in and out of speech! Anyone who has ever tried to edit dialogue will understand just how important the elements of the sound envelope are for the establishment of auditory realism. Even when the initial or final words of a sentence are perfectly comprehensible, they create an uneasy feeling whenever part of their attack or decay has been cut off in the editing process. The production of sound is thus a material event, taking place in space and timeT, and involving the disruption of surrounding matter. This doesn't / mean that we have to be molecular physicists or sound engineers to understand sound, but it does suggest a very precise basis for our description of sound events. It is no longer sufficient to analyze a musical score or a written text to understand the effects of a particular performance I event. Recognizing the extent tojyhich sound sets matter in motion— rnetiujid ty,( ajkejt invisibj^dice readily see the importance of developing a vocabulary >W f/Jc <*w and a niéTnodoiogy appropriate to the complex materiality of sound. ck cA| pcu, k instep of describing just a sound's loudness, pitch, and timbre, we stress j^K t^(. the^xtent to which every sound event includes multiple sounds, each with j. k| 1/ ß'^^spärticular fundamental and array of partials, each with its characteristic N^f^^i'e-'ifound envelope, each Pi^s^singjtsjnvn rhythmjj^nj^he^so^d^ej^nťs K ty p-tjifc«^, overall temporal range. The Sound Narrative: The Story of a Sound Event * ft \j'i|w~** v-4. ■"«kilt U oh. ŕ ^ipJÖäJi/''' jM* ■ ' ?7Zo '1* V. Cď In order to understand sound as it is produced, we need to recognize the materia] heterogeneity of sound events. Sound production is only part of the story, however, for sound, like the proverbial tree falling in the forest, must be heard in order to take on itwiajTaUvejnd_ioxial_signj£i-cance. By offering itself up to be heard, every sound everU Joses its autonomy .^surrendering thejower and meaning of its own structurejp the various contexts in whichjtjnjght pe^heard, to the varying narratives that it might construct Beginning as the vibration that induces molecular .movement, sound is not actualized until it reaches the ear of the hearer, which translates molecular movement into the sensation of sound. Justus the sound event necessarily introduces a temporal dimension into the production of every sound, so the process of percgjtjpjialwjys^ajantegs sound's spatial nature. When we speak of language, we implicitly agree to disregard_c£rtain aspects of linguistic discourse as somehow sub-linguistic. Fred Astaire and Ginger Rogers may make something of the difference between ee-ther and eye-ther, but no normal user of the English language shows such a concern. Regional accents and personal idiosyncrasies produce recognizable differences, but these are not taken to be differences in language. Whether it's ee-ther or eye-ther, it's still the English word "either." Our understanding of sound works in a very similar fashion. We lenow that our neighbor's lawn mower sounds very different when it's •mowing on the near or the far side of the house (and vive la difference!), yet that difference does „not change our nomenclature. Whether the sound comes from the near side or the far side of the neighbor's house, jt is still the sound of the neighbor's lawn mower. The sounds are different, but the name of the sounds is not (Metz). Systematically, thejiame of a sound t • ^•^ "W*- 20 / Material Heterogeneity of Recorded Sound rpfersJn th&^rodiiction of sound and not to its consumption, to the object malring thp snupH rather than the person perceivingjt. Yet the hearing process necessarily involves important variables that often outweigh the sound itself in importance. It doesn't take children long to learn that the word "Boo!" does not by itself produce surprise. When a child jumps out from behind a chair at the other end of the room and shouts "Boo!" the reaction is likely to be mild indeed. When my ten-year-old suddenly emerges from beneath my desk, on the other hand, she can be assured of a good return on her "Boo!", however quietly it may be spoken. To be sure, the sharp attack of the letter "b" contributes to the effect. (If perchance you are not convinced of this, try to scare someone with the pastoral attack of a "Moo!") Still more important, though, is the proximity effect obtained by a good surprise. The effect is dissipated if the booer"holds her hand in front of her mouth or looks away from the booee. The reason for this is very simple: the surprise is created largely^ by the sudden arrival of a zone of sound pressure on the ear. Anything that diminishesTfie sharpness of this experience (standing too far away," whispering instead of shouting, facing away from the booee, or uttering the "Boo!" before emerging from the hiding place) spoils the effect. Having learned to distinguish between various versions ofjhe "same" sounds our ears tell us how to react not on the basis of the sound event alone, But also according to our perceived relation to that sound event. How" does a sound"event contrifJü"fe~f(rhearing?"And~v?häTäre we actually perceiving when we hear? In the previous section, I explained the molecular basis for sound's characteristic compression/rarefaction cycle. Vibration creates pressure, which is communicated through a medium. At the other end of sound's path, the human ear collects that pressure and transforms its mechanical enejsyjnto..ete£tric.aL..imr)ulseAibat..the brain understands as sound. Sensitive to frequency (pitch), amplitude (loudness), and many other factors, the human ear is a marvelously sensitive organ capable of" very minute distinctions. The ear hears not only a sound's fundamental frequencies, but its harmonics and overtones as well, thus facilitating the distinction between male and female voices or French horns and saxophones. Through the ear's ability to sense not only pressure but the rate of changes in pressure as well, we are able Jo measure even minute differences in the sound envejope^and thus to distinguisjLhelween individual voice patterns. The ear must do far more than this, however, for until now I have assumed that sound arrives directly to the ear, in a single pencil of pressure. This is precisely not the case. Imagine an actress standing in the center of a stage in a large auditorium, 150 feet wide and 200 feet deep. Since sound travels at about 1130 feet per second at 70 degrees Fahrenheit, the actress's voice takes approximately one-eighth of a second to reach a member of the audi- Rick Altman I 21 But what happens to the sound that goes straight to the spectator on the edge of the auditorium? Certainly it doesn't die there; it must eventually reach the ears of other audience members as well. Radiating out like a ^'Kcone from the actress's mouth, the sound pressure soon films up the_entire 0cfý\v&, auditorium, bouncing off the walls, the floor, and the ceiling, and bending H^v-p^ around audience members, chairs Tana* posts until it is finally completely ^ absorbed.:" The hot ion that sound travels in a straight line from sound ".T^ -event to hearing ear is thus radically incomplete. In addition to direct Ji*r Q*• • sound, there 1 s~~älso a great deal of reflected sound or^reverberation, Jl° H*1^» ^produced by the sound that reaches the hearing ear onJv_after bouncing uelft off one or more surfaces. In a large room, the delay between thejMiival '^foi^k- °f direct sound and the arrival of the last reflections can be quite long. -j>£kuiseS Whenlhe fuireffect of three-dimensional reverberation is considered, ?p«éc*&~' delays of multiple seconds may easily be encountered. Contrary to popular assumptions, even apparently instantaneous sounds r och.kf » thus have a considerable temporal dimension. Our notation systems for sound reinforce a received notion that separately produced sounds are also perceived separately. As they are printed, Hamlet's words "To be or not to be" provide a blueprint for sounds that are clearly separate and sequential. As they are perceived, however, the direct sound of one word is often i ck^k^* _________ ŕ-'-'ltCrrt heard before the reflected sound of the previous word ceases. Musical ř •f.Zř notation 1>yst^elnatically distinguishes betweenjweWv (sequential sounds) ^J^if^sôurharmony (simultaneous sounds). Yet the sounds notated as sequential +■ ■%> ŕ^uare heard as overlapping, thug confounding the received distinction. The /ÍLÍ!£2 relíecte~J sounds of the first beat of the measure continue to be heard as - j£A«*t me direct sound of later beats reaches our ears. 2Jjl1£^/ Such a distinction might easily appear purely academic and theoretical. Wljj)-^ ^ Our ears know, however, that this is not just a question of splitting hairs. Who has not been in a large auditorium, cafeteria, or gymnasium and had trouble making out the speaker's words. The masler of ceremonies may be saying "The winning numbers are seventeen, forty-three, fourteen, and but what we actually hear is more like "The win------ ^m^ [M& Ho.seventy-two>' num- seven- four- four- seven- Because we know what to expect in this context, we easily complete the opening words, but strain as we might, there is no understanding the all-important numbers, for the reflected sound of the first part of each word is bouncing all around the cafeteria—off chairs, tables, floor, walls, and ceiling—long enough to obscure the direct sound of the second half of each number. We are all aware of the difficulty of understanding a telephone message with the competition of a nearby conversation; our ears know that speakers in large halls often provide their own competition. The subsistence of reflected sound does more than block understanding, however. Our ears are marvelously tuned instruments, extraordinarily 22 I Material Heterogeneity of Recorded Sound f ;*£■(? Ijp» puxjU, "K, Rick Altman I 23 Imagine that there are two actors on our stage, one facing the audience, the other facing backstage. The lights are low; the audience cannot always be sure of seeing which actor's lips are moving. Yet we never have any doubt whatsoever about who is speaking. Our ears tell us. The first actor, facing the audience, sends a strong ray of sound directly to each spectator (as well as an infinite number of rays that reach individual spectators as reflected rather than direct sound). The words pronounced by actor two, away from the audience, are prevented by the actbFs own head from reaching spectators directly. In order:to,be_hgard at all, these sounds rhusťrebiSíhUott the set oFBackdrop, thjis takjng up to three times as long to reach spectators. Fortunately, these spectators are also auditors. Their ears rapidly procesjjlus dataauid easily distinguisli between the Ixy^idslhlífluTT^Tňg^sjTokcn directly to them and the words that have Jo 0 k-, U tri eT f f- í*. %' ^^v1-' how long the reflecjejdjoimdjasts, we refine .oiiLcondusiqnsahauL the C , rJ^7(^'Every sound initiates an event. Every hearing concretizes the story of originating space 15> ivvi"^i that event. Or rather, it concretizes a particular story among the many that In fact, we regularly draw still other conclusions from the other aspects^' V > J~cöuTd~blTtold about that event. When the baseball broke the window, I 24 I Material Heterogeneity of Recorded Sound Rick Altman I 25 , , , . v m CTgasingthe raticLofieJlected to direcLsound? This approach will certainly was outside, more than a little worried; I heard the sound ot tne ore« | convince auditors that they are not located in the same sound space as the directly, with little reflected sound, since there are no walls and, cemng * j ^ ^ ^ ^^ ^ ^ m^en(^hJU3iühHmage_siightly outdoors to keep the reverberation going My father was sitting n"™r^-out of focuS; theloTľnd may even app^atto have been collected in a time favorite chair, right next to the broken window; he was -ejected »ot only £fc. ^^ differtnTfrom its'production^ -------------- to the direct sound of the impact but also to a roomful ot icllectedsound ?f ^ (^ __„ ^ ^^^j^ng fur(her COITiplica,es the question of microphone My mother was ironing in the back room; she thought something■ had;„*£ t j,. lwaúo^^^^n honc |ocatinn bc changCJixvcry_limc ihc broken, but the muffled reflected sound that reached her didn t spec y HMp****camera is nio^eTard the^o. changes? Or should sound logic remain whether it was a wimiow. a vase, a car headlight, or somctimg^ independent of image logic? To what extent is consistency of Doing her homework in the second floor back tadroom..my^^^".J^V^Iy L.^t^^J^'sotiiicl c»Ilcotion iiociloay rVlu^rKouncI ^íTTlcot ic.n ~doc:íKi»n» ho KUl>»rcliii»loa u^-.,' to narrative concerns? Under what conditions may the volume and spatial " '■ čhiractenstics of synchronized sound be modified during the editing process? Are there special volume and reverberation requirements for sound „ e^ effects recorded separately from dialogue? All these and many other ,*=*. location. ■L. I Since the very beginnings of sound cinema, filmmakers have been vanou* i^u.uow^v ~ub!i^i___L-- i - . ,■ ■n-r,„„nnp in „n > % convinced that intelligibility is one of the most important requisites of any sound and its perceptiolMmTFicl takes on special sign'™e "^V;^ recording speech. Indeed, nowhere else are the stakes of microphone media that make use of recorded sound. For whatth£rccor£comauisis _»_-f» «n-. s i..... Ljoing nei iioHicwoiiv m in«-.^>--""............-■- - . , —■ knew anything had happened. Al least until she heard my lather bellow.J*^*^ All four of us heard the "same^sound, yet all f°ur_ofus hcarda * ^..r, a^dtffer^ü^ative of the same event. Sound's existenceas.both event aňínSrHtívTm^nlěi^^ UsualTy^di?cü'sYai's'the1rlr5StTr^ sound isf J in fact a Rashomon phenomenon ^existing only in thej>eparate stones o ^ various perceivers of the origjnaievent. Potentially important apropos o media that make use ofrecorded sound. For w^iíg^^^" ^fe Íočätion°so dea7' I magme that we" are recording a sentence spoken by a notthe sound event as such buUrecord^particula , neanng^^cmc _^|^ ^^ ^ ^.^ ^ ^ eak. & ^ ^^ v"eTÍirJíTH"the story ot tKe"söüTiTeventrEvery recording is thus signed, VCIS1U11 Ul uu- jwi; v/. v~ "------------;- ■> - . . . . as it we're with the JréJrTóTtfe particular circumstances.in which it was heard A recording of the shattering window made next to my father s easy chair will be signed in a different way from a recording of the same event made next to my sister's desk. Ľvery recording ca.Ties.the elements of this spatial signature, carried in the audible signs ^Teach hearing s particTTTarities. Even when..those signs are conjradÍ£loj^rJiaw_bejn tampered with, evüTwhenJhsy seem nono niajghjhevisyaiáata provided with ťhesoJnTrecord, they still carry information thaUsjiajiaMe_aQd woman to the man she is facing. While she is speaking a child walks silently past, catching the woman's attention and causing her to turn away from her interlocutor. Now, jnj)rder to maximize the intelligibility ofthe woman's words we might legitimately decide to "pan" the microphone with her, solhat she is always talking directly into the mike, maximizing dířěčTsoundlind thus intelligibility. Note, however, that this decision robs ťhTsound track of its spatial characteristics. Instead of telling us that the woman turned away from her initial position, the sound track implies that she continued to face in the same direction. If, instead, we choose to retain the initial microphone position throughout", the sound track will exhibit a faithful spatial signature, but it will almost certainly reduce our ability to understand the final parts of the woman's sentence. We will realize that the woman has turned her head while talking, but, like the man to whom she speaks, we may miss some spatíäľln nature. , . The siiSaTíoTTis immensely complicated by the tact that soundjecords never convey exactly the same information that a given auditor would experience. Far from arresting and innocently capturing a particular narrative, the recording process simply e*i£G^^ j.lw of her'words'. Recording "choices, as we easily see'from this example 'Just as the upholstery of a particular roun**aP^^ govern our percq5tT51i of particular sound events. Far from simply resound narrative, so the wayjn which sound r^ colleUed^er^e^to^g, ^^, ^g - K^K ^--^ ^.^ s()und evem (hc ^ engjnccr memory becomesjwrt and jjajcelof Uiej)verall «)imd Pnen'>m^ü"; rc4^>actually has the power t0 create, deform, or reformulate that event. In the -EvenTnlhTiínTplest of sound collection systems, decisions regarding ^ y^ ^ illušIritě^TtnTšolindiengineir''mú'sľchookjo.allow cither ; location of the microphone carry enormous importance, especialjy g , , . f . djaloEUC or mistaken perception on the partoj_the the location of the microphon......, ------ . -.. when the sound is to accompany a related image. Should.sound collection take place in the same room as the sound to be recorded? At what distance ? Ohllěrwhalac(ň.stic-condit^ loT-iM thus' reducing volumeľ dampening certain frequences, and .11- deformation of the dialogue or mistaken perception on the partjiQhj, auditor. Nor is microphone location the only variable available to the sound engineer. The microphone ilseljmakes many choices regarding the type. 37 26 I Material Heterogeneity of Recorded Sound amounLjtnd source of sound that will be collected. It is perhaps useful, in an image-oriented world, to think of the microphone as a "sound-camera," a collection device for sound that shares many of the characteristics of familiar image-collection devices. Just as cameras may have wide-angle or telephoto lenses, changing the angle of image collection and thus the apparent distance of the object filmed, so microphones vary from omnidirectional to Jiarrowly focused, thus changing both the angle of souncTcolTection and the apparent distance of the sound source. In addition, the change in the ratio of direct to reflected sound that accompanies a cfíänge in niicrophoneji^jdj^ ajfe^ characteristics. Microphones also vary in their sensjtivity to specific sound frequencies. The familiar carbon microphone in our telephones has an extremely limited frequency response. Sound heard over the telephone thus always sounds dull and lifeless. Close-miking with a telephone mike (or stripping the sound of appropriate frequencies in postproduction) thus gives the impression that all sounds presented are being heard through a telephone. Since no microphone is equally sensitive to all frequencies, the choice of a microphone fairly assures that some sounds will be boosted, while others will be dampened. Many other microphone characteristics may come into play as well. It is often assumed that every microphone produces a faithful sound record. Actually, no microphone produces an entirely faithful sound record. Not only does every microphone have its own particular directional i~haracteristics (omnidirectional^ bidirectional, cardioid,^hoJgjin_and so; on], but ewTy mTCTOptrone~al5ü "'haTlT5"T5wirjfMrtičulár frequency response, sound ronFiauratiôň7"ajTď^weŕT€qíu*irements. In addition, mariyTňľčTôphones produce unwanted sounds of various types (hum, pop, hiss, buzz, crackle and so on) in a wide variety of situations (loud sound signal wind pressure close sound source, vibration and so on). Recorded sound thus always carries some record of the recording process j_superimposed on thelôuňd event itself. Added to theTtory of sound production we always find the traces of sound recording as well, including information on the location, type, orientation, and movement of the sound collection devices, not to mention the many variables intervening between collection and recording of sound (amplification, filtering, equalization, noise reduction, and so forth). Indeed, the recording system itself provides one of the most important determinants of sound characteristics; as such it not only provides a record of sound, it also participates in the overall sound narrative. Think for example of the differing frequency responses of 78 rpm records and digital compact disks. It is so difficult to compare musical performances recorded on these two radically different technologies that the masterworks of Toscanini and Furtwangler seem diminished Rick Altman I 27 without the wonders of digjta[ remastering (which is none other than an attempt to restore the frequencies towhich pre-war disk recording was not sensitive)." * ~~ To record is thus to recall to mind, as the dictionary would have it, but like most mnemonic devices, sound recordings must heighten some aspects of the original phenomenon at the exr3eTisirôror^ís.^ä:aurled ,<í*vcc readings are thujjjiTwáyšrj=e^sej[iUitiqns,_lnt^ŕHä1iions, partial narra- t rV0^5 lives tha^must nevertheless serve as our only access to the sounds of the v_ requirements was rapidly imposed. The words had to be comprehensible;' j?^"?, to that end the amount and duration of reflected sound had to be kept 28 I Material Heterogeneity of Recorded Sound extremely low. Studios thus rushed to create acoustically treated sets open-ended studios, and other devices designed to limit reverberation and maximľzeintelligibUity. The biggest stumbling block of all turned out, Z v r tô hav "nothing to do with film production. It was film.exhibition ľhaTcaľsed all the problems. Built to maximize reverberation, hose drafty & Ä picture palaces made it nearly impossible to understand the Sen word. Whenpjayed overthe loudspeakers of a huge, hard-surfaced fe with three^iddecayjime, even intimate scenes recorded to give the Lpression of small, private spaces sounded as »f they were set m davemous public halls. Carefully recorded speech turned into the same auditory mush that had become the trademark of romantic church organs and mighty theater Wurlitzers alike. Only c^fuU^mi^^^y acoustic treatment were able to solve this problem. -tLu with the practical problem solved, however, the theoretical difficulty remains. Which acoustics amjjistening to? The Hollywood sound stage or the Rialto? Severance Hall or my living room? For that matter which sound am I listening to? The original sound event or its loudspeaker reproduction? In order to understand sound-cinema sound in part.cular-we must recognize both the narrative and the represented nature of sound ľs breaches our ears in the movie theater. The sound system plays the recorďÄe story of an evenT At every point in that chain, new variables enter new elements of uncertainty. Sound heads, amplifiers, lead.loudspeakers and theater acoustics all force new auditory data on the audience juľta! Ihe^ecording process itself had earlier introduced an implicit viewpoint. Hearing Events: Hearing the Record of the Story of a Sound Event Just as sound events remain only hypothetical sound sources until they are actualized by a hearer, so the playing of a sound record takes on mean ľg only i/the presence of an audience. Yet the process of hearing recording differs significantly from listening to a live sound eyent. This hodd come as no surprise to anyone who has contemplated^.difference between a photograph and the scene that it represents. When we look directly at a scene we gain a sense of depth from our binocular vision, by S our head, or by moving to the left or right. If we want to know what's underneath a chair we have but to lean down. In order to get a dearer view of a specific object, we need only adjust the focus of ou eyes Ye all of this is to no avail when we view a photograph. No amount of rotating, moving, leaning, or adjusting will deliver Ration that the photograph lacks. We may have two eyes, but we might as well all be Cyclops when it comes to sensing distance in a photograph for here the ^y F vv ___ ,u,~„„t, c.;-/*» masknp. and detail rather ■Mirr»!"*' n vnen u comes iu »t.naiug, «..««■— — ~ ' . " ' , . •• rnlUar ,f stance is encoded through size, masking, and detail rather Rick Altman I 29 than sensed by the parallax implicit in binocular vision. Without requiring any special education, we have all learned to use our eyes radically differently when we view three-dimensional space and when we view a two-dimensional representation of that space. _ , A similar situation obtains with the sound representations that we call c55^' recordings. When attempting to locate a_cryimx child we normally call^/^'' heavily on our binaural hearing system to provide cues regarding lateral i^gf^" location. When we listen to a recording_o_La crying^chjjrj^jiasuchjbcaliza- —-2-^*' tion is possible. However much we might rotate our heads or change positions, we remain unable to make use of the directional information *"■§??' that was present when ^Re^ sound was produced, but which is no longer ľír----- available in the recording (unless it is in stereo, and even then the location ^v^. of microphones and speakers plays just as important a role as the location ^-^ S(' of the original sound source). For listening to the sound pouring out of a loudspeaker is like hearing a lawn mower through an open window: wherever the lawn mower may actually be, it always appears^to_beJocated on the side of the house where the open window is. When we listen for a crying child, we are marvelously effective at ~a>k"*w cutting out extraneous sounds and concentrating on the cries that we tfFtu/, recognize as those of our own child. Dubbed the cocktail party effect by ^SeQk-Colin Cherry, the process of s^|e£tjye_audkory Mention is far more $£*.,;W difficult when we are listening to recorded material. Whereas live sound p»»»*«* , provides an extraordinary number of variables, each permitting and promoting selective attention, recorded sound folds most of those variables into a single, undifferentiated source. In a live situation, we easily differentiate among the various sound sources surrounding us, but with recorded sound no such clear distinctions are possible. Live sound situations reveal the actual relationship between the sound producer and perceiver, while recordings suggest only an apparent relationship. If I sit in an auditorium and listen with my eyes closed to a series of speeches, I remain constantly aware of the speakers' location. I know what direction they are facing, how loud they are speaking, and what '*w* original bodjjwsitJpjis^olumejjJiLtones. Depending on the type, loca- ^tW tion, and movement of the microphone(s) used in the recording process, ^^(ho^.-. the recorded sound substitutes an apparent sound event for the original olr,r?ft.<Ě. phenomenon. Revealing its mandate to represent sound events rather than '*■**■ *£V-to reproduce them, recorded sound creates an illusion of presence while UM^>-u•'--constituting a new version of the sound events that actually transpired. "Jtief*t^^. What happens in the course of a hearing event is thus not the expected ŕ" i^^Hf' detective activity wherein the hearer searches the recorded sound track for£■£--— clues permitting reconstitution of the original sound event. Instead, weJiurS* 30 I Material Heterogeneity of Recorded Sound r) follow the trail that has been laid for us all the way to an apparent sound event haying_all_Üie aural guarantees of reality but onjy_E>artLal correspondence to theoriginaljoyiMLfflffiDt. Indeed, it is the partial nature of the relationship that makes hearing events so fascinating. If there were no connection between the apparent sound event and the original sound source, recorded sound would not have its extraordinary capacity for ideologicITimpact. It is precisely because recorded sound seemsjorepro-3uce an onglnaTriheriomenon that.recordings, attract.anoíínijwdjences söreadjly j3^eejLt^^ antl the x^SlS)S. represen- tation lies thcfdiscursive power of recorded sound. We hear recordings with the same ears we use for live sound. We reach conclusions about the evidence provided by recordings in the same way that we interrogate and evaluate live sound. We constitute apparent sound events just as we directly perceive live sound events. Yet recordings systematically fail to justify our confidence inthem. Most listeners have learned to concentrate on the aspects of sound events that are most faithfully rendered by recordings and to pay little attention to the aspects introduced or transformed by the recording process. A proper theory of sound will accept no such selective deafness. It will pay special attention to those very points where confusion is possible, recognizing in such moments of imprecision, indecision, or incoherence the very place where sound seizes the opportunity to take an active role in the definition and exploitation of culture. Sound Terminology: Talking about Hearing the Record of the Story of a Sound Event Often called "distortions," on the theory that sound recording is a science of reproduction rather than an art of representation, the variables introduced by sound's material heterogeneity, along with the system constituted to record (that is. represent) it. lie at the very heart of film sound. Though they may constitute distortions for the sound engineer, the marks of the sound narrative and the recording process that appear as part of any sound record constitute the very text of the sound analyst, the fundamental slgňšof the sound semiotician, the basic facts of the sound historian. Central to the interpretation of film sound is the fact that multiple moments and operations must be carried simutaneously by the same final souncTtrack. The characteristics of sound production, sound recording, sound v^xoá\xc^ň[^^^čr^xcmion are all superimposedin a ^iH^Tě~ě^penéňče. When we hear any particular film sound,Jhiow.do we knowto whom to attribute it? Which part of the sound chain has produced, selected, highlighted, or masked it? Does a decrease in the ratio of direct to indirect sound mean that the character has turned away, an obstacle has Rick Altman I 31 been introduced, the microphone has been moved, the sound engineer has fiddled with the dials, or the spectator has shifted her position? To study film sound is to take seriously the multiplicity of possible determinants of anv given audience perception As a complex representation of a complex sound event, cinema sound offers sound designers infinite possibilities for creation and obfuscation. As such, it also offers theoreticians and critics of cinema sound fascinating opportunities to recognize and analyze the techniques, conventions, codes, and ideological investments of the sound chain. This work is only beginning. It will move more quickly if we adopt a vocabulary that reflects the material, heterogeneous nature of sound presented here. 82 09 • ■=a Notes i. The Material Heterogeneity of Recorded Sound 1. Bordwell 1990. Not surprisingly, technical manuals aimed at the production of sound rather than at aesthetic analysis of sound present a far broader terminology The best and most complete of these manuals is by Stanley R, Alten. 2. As a general rule, sound waves will be reflected by (that is. bounce off) obstacles having dimensions greater than the wavelength of the sound. Since audible sounds have wavelengths varying from about one and one-half inches (the C above the piano, 8372 Hz) to 70 feet (the C below the piano, 16 Hz), with the fundamentals of most sounds having wavelengths between one and eight feet most acoustic situations will produce a combination of reflected and refracted (that is, bent) sound. 2. Sound Space 1. "Reproducing Sound from Separate Film," JSMľfi 16 (Feb. 1931), p 152 Just three months later a patent taken out by W. Bouwa, for "Apparatus and Method for Localization of Sound on Screen," was reported on in JSMPE 16 5 (Mav 1931) 643-44. 3 ' 2. On the particular outlook of sound technicians during this period, see Altman 1992. 3. It is difficult, however, to reconcile Maxficld's total mastery over everything related to reverberation with an obvious oversight in his 1938 reprinting of the microphone placement chart from the 1931 article (p. 73). Whereas before 1931 only omnidirectional microphones were available in Hollywood, by 1938 the ribbon bidirectional mike and the cardioid mike were widely used, especially where dialogue had to be recorded with lenses of large focal length, such as those charted in Maxlicld's graph. Now, it is generally recognized that the cardioid (directional) microphone collects far less reflected sound than an omnidirectional mike at the same distance thus permitting placement of the cardioid mike at 1.7 times the distance appropriate for an omnidirectional nuke. Is Maxfield's blind spot perhaps a sign of fidelity to the parent Bell/WL/hRPI complex'.' Whereas Olson's directional mikes were developed lor RCA in the early thirties Wrsii-m ľi,., 78