Electronics Guide

Why Synchronized Sound Waited for Electronics

The failure of pre-electronic sound film is easy to misread as a failure of imagination. It was not. Inventors attempted talking pictures continuously from the 1890s, and several systems reached commercial exhibition. They failed on physics, and they failed on the same two counts every time.

The Vacuum Tube Arrives in the Theater

The technical foundation was laid outside the film industry, in the laboratories of the American telephone system. Western Electric, the manufacturing arm of AT&T, and the research organization that became Bell Telephone Laboratories in 1925 had spent a decade building the components a sound film needs, for reasons that had nothing to do with cinema.

The condenser transmitter developed by Edward C. Wente at Western Electric from 1916 gave a microphone with a flat, wide response and a predictable calibration, at a time when the alternative was a carbon button whose sensitivity drifted with packing and temperature. Vacuum-tube amplifiers built for long-distance telephone repeaters gave gain that was stable enough to cascade. Loudspeaker research aimed at public address gave moving-coil drivers coupled to exponential horns, an arrangement that trades bulk for efficiency and was the only way to obtain theater-filling output from the tens of watts a tube amplifier of the period could produce.

Those same components produced the electrical recording system that Western Electric licensed to the phonograph industry in 1925. Applying them to film was an obvious next step for the engineers, and a much less obvious one for the studios, who had watched sound pictures fail repeatedly and saw no reason to gamble a profitable silent business on another attempt.

AT&T eventually organized its film activity into a subsidiary, Electrical Research Products, Incorporated, universally called ERPI, which licensed sound equipment to studios and installed and serviced it in theaters. ERPI's position is central to the economics of the transition. The telephone company did not make films; it made and rented the apparatus, collected license fees on production, and held the patents that made both possible. When the industry converted, a large share of the money flowed out of Hollywood and into the electrical industry.

Sound on Disc: Vitaphone

Warner Bros. was a second-rank studio in the mid-1920s, ambitious and short of theaters, which is precisely the position from which a risky technical bet looks attractive. Working with Western Electric, the company formed the Vitaphone Corporation in April 1926 and put the system before the public that summer.

Sound on Film: The Case Laboratory and Movietone

The alternative approach photographs the sound. A microphone signal, amplified, modulates a light source; the modulated light exposes a narrow stripe running the length of the film alongside the picture; and in the projector a lamp shines through that stripe onto a photoelectric cell, which recovers the waveform. The idea is older than Vitaphone by decades. What it needed, again, was gain: a photoelectric cell delivers a minute current, and reading a soundtrack is impossible without an amplifier immediately behind the cell.

Three lines of work converged. In Germany, Josef Engl, Joseph Massolle, and Hans Vogt developed the Tri-Ergon system from 1919, using a glow lamp to write a variable-density track. Its patents proved commercially consequential in a way its films did not: William Fox bought rights to them in 1926, and litigation over their scope ran until 1935. In the United States, Lee de Forest promoted his Phonofilm process from the early 1920s in a small circuit of theaters. Phonofilm did not fail on principle; it failed on quality, on capital, and on de Forest's inability to interest a major studio in a system he was demonstrating with vaudeville shorts. The company was defunct by 1929.

The decisive contributions came from Theodore Case's laboratory in Auburn, New York. Case had developed the thalofide cell, a thallium-compound photodetector, between 1916 and 1918, initially for a classified United States Navy infrared signaling system tested in 1917. He then developed the AEO light, a glow lamp whose output could follow an audio-frequency signal cleanly enough to expose a soundtrack. Between 1921 and 1924 Case supplied de Forest with the components that made Phonofilm work. When de Forest presented the process publicly in 1923 without crediting the laboratory, the relationship deteriorated, and Case cut off supply in 1925.

Case then found a buyer with a studio behind him. On 23 July 1926 William Fox purchased Case's sound-on-film patents and formed the Fox-Case Corporation, and Case and his associate Earl Sponable moved to Fox to develop what became Movietone. The AEO light went into production and was used in Movietone News cameras from 1928 to 1939 and in the recording of Fox feature films from 1928 to 1931.

Fox's first strategic use of the system was not drama but news. On 20 May 1927 a Movietone crew photographed Charles Lindbergh's takeoff from Roosevelt Field with synchronized sound, and the footage was screened in a New York theater the same night. Regular Movietone News presentations began in December 1927 and the series continued in the United States until 1963. Newsreel was the ideal proving ground: the camera recorded picture and sound on one negative in one pass, so nothing could go out of register, and the material had to reach a screen within hours.

Sound on Film: RCA Photophone and the Variable-Area Track

The competing optical system came out of General Electric and reached the market through RCA. Charles A. Hoxie had built a photographic recorder he called the Pallophotophone after the First World War, originally to record wireless telegraph signals, and adapted it to speech; it captured remarks by President Calvin Coolidge in 1921 for broadcast over station WGY. General Electric began developing the technique commercially for film in 1925, and in April 1928 RCA Photophone, Inc., was organized as an RCA subsidiary under David Sarnoff.

RCA needed theaters and a studio to use the system, and the corporate answer was RKO Radio Pictures, assembled from the Film Booking Offices production company, the Keith-Albee-Orpheum theater circuit, and RCA itself. RKO adopted Photophone as its house system. An FBO release, The Perfect Crime, premiered on 17 June 1928, and Syncopation followed in March 1929 as the first film recorded live with RCA Photophone.

The Photoelectric Cell and the Sound Head

Reading the track is a small analog design problem with awkward constraints, and the solutions adopted in the late 1920s survived essentially unchanged for seventy years.

The reader is an assembly called the sound head, mounted below the projector head on a machine that plays analog optical tracks. An incandescent exciter lamp, run from a well-regulated direct-current supply because any ripple appears directly as hum in the audio, illuminates a mechanical slit through a small objective lens. The slit is imaged onto the moving film as a line a few thousandths of an inch high and as wide as the track. Light passing through falls on a photoelectric cell, historically a vacuum or gas-filled phototube with an alkali-metal photocathode, later a photodiode.

The cell's output is minute, of the order of microamperes into a high impedance, which makes the connection between the cell and the first stage of amplification the most delicate node in the whole theater sound system: a high-impedance, low-level circuit inside a machine containing an arc lamp, a motor, and a good deal of vibration. Preamplifiers were mounted as close to the cell as possible, the wiring was shielded, and microphonics in the first tube were a chronic complaint. This is the point at which the sound film is unambiguously an electronic technology: without gain immediately behind the photocell, there is no signal to work with at all.

The mechanical requirement is equally strict, and it is the reason the sound head is not at the picture gate. Film moves through the gate intermittently, stopping for each frame, because a projected image must be stationary while it is shown. An optical soundtrack must move past the slit at an absolutely constant speed, because any speed variation is frequency modulation of the recovered audio, heard as wow at low rates and flutter at high ones. The two motions cannot share a station, so the sound head sits separately, where the film wraps a polished sound drum carrying a heavy flywheel whose inertia filters the motion, with slack loops above and below isolating it from the gate.

The consequence is that the sound and the picture that belong together cannot occupy the same position on the film. On a 35 mm release print, the sound head is twenty-one frames after the gate, so the optical track is printed twenty-one frames ahead of the picture it accompanies. At twenty-four frames per second that is seven-eighths of a second of advance, and it is a printing convention, not an adjustment: any laboratory making a release print applies it, and every projector in the world is built to the same offset. On 16 mm prints with a mono optical track, the corresponding advance is twenty-six frames.

That offset also explains one of the medium's characteristic failures. When a print breaks and is repaired, the projectionist removes frames from a single point, but the picture and the sound belonging to that moment are twenty-one frames apart, so a splice damages the picture at one instant and the sound at a different one. The penalty is far smaller than the cumulative desynchronization that splicing inflicted on a disc system, because it is local and self-limiting, and that difference is the whole argument between the two formats in miniature.

What the Format Fixed: Frame Rate, Speed, and the Shape of the Picture

Once sound lived on the film, the film's physical parameters stopped being adjustable, and cinema acquired a set of constants it has largely kept.

The Tyranny of the Microphone

The most visible cost of the transition fell on production technique, and for about three years it made sound films look primitive beside the silent pictures they replaced.

The immediate problem is that a camera makes noise. A silent-era camera clattered, and nobody cared, because the only witness was the crew. A microphone in the same room records the clatter. The first remedy was brutal: enclose the camera and its operator in a soundproof booth, a glass-fronted cabinet the industry nicknamed the icebox, and shoot through the window. A camera in a booth cannot pan, cannot track, and cannot be moved without an hour of labor. Since a scene had to be played continuously for the recording, several such booths stood around the set, each rooted in place.

The microphone imposed its own geometry. The omnidirectional condenser microphones first used picked up the set as readily as the actor, so they had to be close, which meant hidden in a flower arrangement or behind a lamp, which meant the actor had to be close to the flower arrangement. Performers stood still and delivered lines toward the furniture. Critics who wrote about the sudden immobility of the cinema were describing an accurate consequence of the recording chain.

The recovery took most of a decade and came from four directions.

First, the camera was quieted rather than imprisoned. Sound-insulating covers built around the camera body, called blimps, and later cameras designed from the outset for quiet running, released the camera from the booth and restored the moving shot.

Second, microphones acquired directionality. RCA manufactured the Photophone Type PB-31 ribbon microphone commercially in 1931 and installed it at Radio City Music Hall in 1932; the Type 44A velocity microphone followed, with pattern and tone controls intended to manage reverberation, and unidirectional ribbon designs came out of the same laboratory during the 1930s. A microphone that rejects sound arriving from the sides and rear can be placed further from the actor for the same ratio of wanted to unwanted sound, and it can be aimed. Mounted on a boom, it could follow a moving performer, which is what finally reconciled the microphone with blocking.

Third, the industry learned to mix. Once several microphones fed a console attended by a sound engineer, coverage no longer depended on a single compromise position.

Fourth, and most consequentially, sound was separated from the moment of photography. Rerecording, which combines dialogue, music, and effects tracks in a later pass, and post-synchronization, in which an actor watches the picture and replaces the dialogue in a studio, were practical by the middle of the 1930s. Post-synchronization restored the property that had made silent cinema flexible: the ability to shoot for the picture and fix the rest afterward. It also created the modern sound department, in which the recording made on set is raw material rather than a finished product.

Wiring the Theater

On the exhibition side, the transition meant installing a small electroacoustic plant in every auditorium in the country.

The chain begins at the sound head and ends at horns behind the screen. Because the loudspeakers must be heard through the picture, the screen was replaced with perforated material, opaque enough to hold an image and open enough to pass sound. Screen perforation is a permanent feature of theatrical exhibition, and it dates from this moment.

Loudspeakers were horn-loaded from the beginning, and not for aesthetic reasons. A direct-radiating cone driver converts a small percentage of its electrical input into acoustic power, while a horn matches the high acoustic impedance at the diaphragm to the low impedance of the room and raises efficiency by an order of magnitude or more. With amplifiers delivering tens of watts, that difference decided whether a large house could be filled at all. Early Western Electric installations paired compression drivers with large exponential horns, and the arrangement set the acoustic character of the American cinema for a generation.

Single-horn systems could not cover the frequency range well, and by 1931 theater systems were being divided into separate low-, middle-, and high-frequency sections with filters distributing the signal among them. The best-known result was the two-way system developed at MGM under Douglas Shearer, head of the studio's sound department, with John Hilliard leading the engineering team and drivers supplied by James B. Lansing's Lansing Manufacturing Company. The Shearer horn used multiple fifteen-inch field-coil drivers in a folded bass horn crossed over near five hundred hertz to a multicellular high-frequency horn, the multicellular construction being a means of controlling coverage across a wide auditorium. The Academy recognized the development with a technical award in 1936, and systems built on the same plan became the industry standard.

Standardizing the loudspeakers made it possible to standardize what was fed to them. From 1938 the industry worked to a specified theater playback response, the Academy curve, which rolled off the high frequencies substantially. It was a practical accommodation to the noise of an optical track and to the horn systems then installed, and once it existed, mixers could balance a film in a studio and expect a comparable result in a distant house. It governed theatrical sound until Dolby's work in the 1970s made a wider response practical.

The building itself needed work. Auditoriums designed for live performance were generally too reverberant for reproduced speech, since reverberation smears consonants and destroys intelligibility long before it becomes objectionable on music. Absorptive treatment on walls and ceilings became a routine part of conversion, and theater acoustics shifted from supporting a live source to serving a reinforced one. Projection booths needed room for amplifier racks and power supplies, many buildings needed their electrical service upgraded, and projectionists needed training in equipment that had nothing to do with the craft they had learned.

What Conversion Cost

The bill for all this was large, and it arrived at the worst possible moment in American economic history.

Wiring a theater for sound cost on the order of fifteen thousand dollars, a sum on the order of a quarter of a million dollars in early twenty-first-century money, and there were more than twenty thousand motion picture theaters in the United States. The money went to ERPI or to RCA Photophone, in equipment purchases, installation, service contracts, and continuing license fees, which is why the transition transferred so much value from the film business to the electrical business.

Conversion was fast where capital was available and slow where it was not. Even in the United States, only about half of the theaters had been wired by 1930, which is why studios released features in both sound and silent versions well into that year. Britain had more than sixty per cent of its theaters equipped by the end of 1930. In France more than half were still projecting in silence in late 1932. In the Soviet Union, as of May 1933, fewer than one film projector in a hundred was equipped for sound.

The uneven pace had structural consequences. A small independent house that could not raise fifteen thousand dollars lost its audience to the wired theater down the street, while chains that could finance conversion across a circuit emerged larger. Studios borrowed heavily to build soundstages and re-equip, which brought investment banks into the governance of film companies to a degree they had not previously enjoyed. The consolidation of the American industry into a small group of vertically integrated majors, which later antitrust litigation would unwind, was accelerated by the capital requirements of sound.

Who Paid: Musicians, Performers, and the Language Barrier

The clearest human cost fell on musicians. Silent film was never silent in exhibition; it was accompanied, by a pianist or an organist in a small house and by a substantial orchestra in a first-run palace. Accompaniment was one of the largest sources of employment for instrumentalists in the United States, and a recorded score eliminated it. The American Federation of Musicians recorded that some twenty thousand musicians lost their theater work within two years of the arrival of talking pictures, and in 1930 the union founded the Music Defense League to campaign against what it called canned music. The campaign lost. It is one of the earliest instances of organized labor confronting the substitution of a recording for a live performance.

Performers were affected unevenly, and the popular story exaggerates. Careers did end because a voice did not suit the microphone, but the larger disruption was that sound imported a different professional standard. Studios recruited from the theater, where actors had trained to speak, and the skills that had made a silent screen performer valuable were not the same skills.

The most far-reaching commercial consequence was linguistic. A silent film crossed borders by having its intertitles retranslated and reprinted, an operation cheap enough to be an afterthought. A talking picture is locked to the language of its dialogue. Hollywood had built an export business on a product that was effectively language-neutral, and sound broke it.

The first response was to shoot the film again. From 1929 studios produced multiple-language versions, remaking a picture on the same sets with the same crew and costumes but a different cast speaking a different language, most often English, Spanish, French, or German. Paramount ran a dedicated facility at Joinville outside Paris for the purpose, and MGM, Universal, Warner Bros., and the German studio UFA all made them. The practice peaked in the early 1930s and then collapsed, partly because shooting a film three or four times was indefensibly expensive and partly because the alternatives improved. Dubbing, which replaces the dialogue track, and subtitling, which requires only a laboratory, were both far cheaper, and by the mid-1930s the market had divided into dubbing territories and subtitling territories along lines that persist today.

What Came After, in Outline

The transition described here settled the format. Everything since has been an improvement within it, and a brief sketch is enough to place the later waves.

Fidelity improved through the 1930s by attacking the optical track's weaknesses directly: noiseless recording that reduced grain noise in quiet passages, push-pull tracks that cancelled even-order distortion by recording two mirror-image modulations read differentially, and better transducers at both ends.

Multichannel sound arrived experimentally with Disney's Fantasound, developed with RCA for Fantasia in 1940, which carried its audio on a separate synchronized film and steered it among loudspeakers in the auditorium. It was too expensive to install widely, and the war ended the experiment.

Magnetic sound and widescreen arrived together in the 1950s, driven by competition with television. Magnetic stripes laminated onto the release print carried discrete channels with a signal-to-noise ratio and a frequency response that optical tracks could not approach: CinemaScope introduced four-track magnetic stereo in 1953, and Todd-AO brought six-track magnetic sound on 70 mm prints in 1955. Magnetic prints were expensive to strike and the oxide wore, so the format stayed at the premium end of exhibition.

Dolby Laboratories changed that in 1976 by making the optical track good enough. Dolby Stereo applied A-type noise reduction to a stereo variable-area track, lifting the medium's dynamic range enough to abandon the Academy curve, and used a four-to-two-to-four matrix to fold left, center, right, and a surround channel into the two optical tracks a print could carry. Center information was distributed equally between the tracks and surround information encoded out of phase between them, so a decoder could recover four channels by summing and differencing. Because a matrixed print plays acceptably as ordinary stereo or mono on unequipped machines, exhibitors could convert at their own pace. Dolby SR replaced A-type on release prints from the late 1980s.

Digital soundtracks arrived in the early 1990s in three incompatible forms, each of which found a different unused part of the print. Dolby Digital, introduced in 1992 and first widely released on Batman Returns, printed its data between the perforations on the soundtrack side of the film, carrying 5.1 channels at three hundred and twenty kilobits per second. Sony Dynamic Digital Sound, introduced the same year, used both outer edges of the print. DTS, founded in 1990, took the opposite approach: it printed only a timecode track on the film and played the audio from separate CD-ROM discs locked to it, an arrangement that debuted on Jurassic Park in 1993 and bought bandwidth at the cost of another object to keep with the print. Prints of the period commonly carried all three digital formats plus a Dolby SR analog track, so that any theater could play any print. That redundancy is a direct descendant of the format war of 1927, and its lesson had evidently been learned.

Digital distribution ended the argument by ending the print. When a feature travels as an encrypted digital cinema package and plays from a server, the soundtrack is a set of files with a timeline, and synchronization is guaranteed by the clock that drives the image. The problem that defeated Edison for twenty years, and that Vitaphone solved so precariously, no longer exists in a form that can fail.

Conclusion

The coming of sound to motion pictures is usually told as a story about The Jazz Singer and about actors with unsuitable voices. The engineering account is more useful. Synchronized sound waited for the vacuum tube because both of its obstacles were, at bottom, the same obstacle. Without gain there was no way to fill a theater from a recording, and without gain there was no way to read a signal weak enough to be carried on the film itself. Once the amplifier existed, both problems became solvable, and they were solved in two different ways within eighteen months.

Sound on disc won the market and lost the medium. It was quicker to deploy because a disc recording chain already existed and could be applied to film with little modification, and it proved to a skeptical industry that audiences would pay. But it kept the sound in a separate object, and every failure mode that followed from that separation, the splice, the reel change, the worn disc, the uneditable master, was structural. Sound on film moved the audio onto the same strip as the image, which made synchronization a property of the material rather than a service performed by a mechanism, and no amount of refinement to the disc system could have matched that.

The winning format then imposed its constraints on everything downstream. It fixed the frame rate at twenty-four per second because the optical track needed a known and adequate linear speed. It put the sound twenty-one frames ahead of its picture because film cannot move intermittently and continuously at the same point. It took a stripe out of the width of the frame, which forced the Academy aperture and gave classical cinema its shape. For a few years it rooted the camera and the actors in place, until directional microphones, blimped cameras, and post-synchronization gave the medium its mobility back.

And it was expensive. It moved a great deal of money from the film industry to the electrical industry, accelerated the consolidation of exhibition, ended a major category of musical employment, and broke the language-neutral export business on which Hollywood's international position had been built, replacing it with the dubbing and subtitling economy the world still uses. Sound was not an addition to the motion picture. It was a rebuilding of it, and the electronics were the reason it could be done at all.

Related Topics