This article is concerned with historical and technical aspects of the means for recording, reproducing and transmitting sound. The material covered in the first section includes, first, a history of recorded sound from mechanical to electrical means; secondly, a history of the various media and technologies used in recording; and lastly, a historical overview of the recording industry. The material covered in the second section includes, first, the conversion of sound into usable electrical energy by microphones; secondly, the employment of microphones and other apparatus in recording and broadcasting studios; thirdly, the recording of sound on disc, tape and film; fourthly, the transmission of sound by radio broadcasting; and lastly, the reproduction of sound through amplifiers and loudspeakers. For the history of music in radio and the influence of radio on musical life, see Radio.
II. Techniques of recording, reproduction and transmission
ARTHUR W.J.G. ORD-HUME (I, 1), JEROME F. WEBER (I, 2–5, 7–8), JEROME F. WEBER (with JOHN BORWICK) (I, 6), D.E.L. SHORTER/JOHN BORWICK (II, 1–5, 12–13), JOHN BORWICK (II, 6–11)
1. Early recording and notating devices.
4. Long-playing records and tape recording.
8. The recording industry after 1948.
Recorded sound, §I: History of recording
The earliest attempts to obtain a permanent record of extempore performance were in the form of ‘notating devices’, normally attached to the action of a keyboard instrument. Apparently the first such machine was proposed in England by J. Creed, who wrote a ‘Demonstration of the Possibility of making a machine that shall write Extempore Voluntaries or other Pieces of Music as fast as any master shall be able to play them upon an Organ, Harpsichord, etc.’, published posthumously by the Royal Society in 1747. A similar invention, called the melograph, was conceived by Leonhard Euler (1707–83) and built by Johann Hohlfeld in about 1752, consisting of two revolving cylinders with a band of paper passing over them. Note positions and duration were marked onto the paper by pencils connected to the piano action. (The concept of the Melograph was revived in the 1950s in an electronic instrument used in musicological research for the continuous graphic representation of melody.) Other similar machines, later employing electromagnetic or pneumatic actions, were developed by J.F. Unger (1716–81), J.J. Merlin (1735–1803; a harpsichord-piano fitted with his clockwork-driven device is in the Deutsches Museum, Munich), J.H. Pape (1789–1875) and Jules Carpentier (1851–1921), though all met with limited success. The most successful method of recording and reproducing a piano performance with all its nuances of expression was developed in the Reproducing piano, introduced by Welte in 1904. (See also Mechanical instrument).
In April 1877 the Frenchman Charles Cros (1842–88) deposited with the Académie des Sciences a paper containing proposals for the reproduction of sound, but he failed to put his theories to practical test. In the same year the American Thomas Edison (1847–1931) came quite independently to the study of sound recording and reproduction as an offshoot of his researches into high-speed telegraphy.
Recorded sound, §I: History of recording
Although Edison's speaking voice first engraved Mary had a little lamb on tinfoil in December 1877, the inventor exploited the cylinder ‘phonograph’ only as a toy from 1878 to 1880 before turning his attention to the electric light. The invention remained dormant for the next eight years. A few tinfoil recordings are preserved at the Edison National Historic Site at Menlo Park, New Jersey.
The history of recorded sound begins in 1888, when Edison entered the market with an improved version of the phonograph. At the same time Columbia began selling its ‘graphophone’, using patents for a floating stylus developed by Alexander Graham Bell and associates. Recordings of celebrities were made in the USA and Europe, and by 1890 both firms were selling cylinder recordings of singers, instrumentalists and comic artists from the vaudeville stage. Columbia recorded the US Marine Band, although its director John Philip Sousa did not participate in the recordings. In 1893 the newly formed Sousa's Band began to record for Columbia, still without its director. The first cylinders were recorded in batches of ten. A few years later the pantograph was used to make duplicates from a master recording; only in 1901 was the process of moulding copies of a cylinder recording developed.
In 1894 Emile Berliner, a German émigré in the USA, began to sell a similar range of entertainment on flat-disc recordings, played on a machine that he called the ‘gramophone’. The disc had an immediate commercial advantage over the cylinder in that many copies could be produced by moulding from a master. Nevertheless, the two media continued to compete for public favour for about 25 years before the disc ousted the cylinder. The first discs were made of hard rubber, which was replaced in 1897 with a shellac compound. The groove was modulated on a zinc disc coated with fatty film, and the grooves were then etched in an acid bath. Eldridge R. Johnson developed a method of cutting a groove in a solid wax disc, and in 1901 he and Berliner joined to form the Victor Talking Machine Co. William Michaelis developed the ‘hill-and-dale’ disc (see §II, 6, below) in England in 1904, calling it the neophone. Recordings issued up to 1901, some of them made privately, can be considered incunabula. Gianni Bettini designed an improved diaphragm for the phonograph which he used to record leading opera singers in New York, but the cylinders were duplicated in small quantities and sold at high prices; virtually all have been lost. Lionel Mapleson, the librarian of the Metropolitan Opera, recorded performances from the prompt-box and from the catwalk over the stage from 1900 to 1903.
The beginning of recorded sound as a serious medium, however, dates from 1902, when Columbia (which had begun to make discs as well as cylinders) and Victor agreed to pool their disc-recording patents. Disc recording began to eclipse the cylinder from this time. In Europe Pathé Frères, which began in 1896 to produce an extensive catalogue of cylinder recordings, adopted the disc in 1906 and abandoned the cylinder in 1910; by 1919 the firm was issuing only lateral-cut discs. Columbia stopped producing cylinders in 1912, but Edison, who began to make discs only in 1913, continued both formats until 1929, when his firm withdrew from the record market to focus on office dictation machines.
In 1897 William Barry Owen went to England to negotiate the sale of Berliner's patent rights. The Gramophone Co. was founded the following year in London, and a manufacturing branch was opened in Hanover. Fred Gaisberg, who had worked briefly for Columbia and then for Berliner, soon joined the firm, which remained closely affiliated with Berliner's by exchange of master recordings and shared use of trademarks. In December 1900 the Gramophone Co. became the Gramophone and Typewriter Co. (G & T), with the recording angel as its trademark. (It was revived in 1953 for the American market.) In 1909, however, the firm reverted to its original name and changed its trademark to the painting of a dog listening to ‘his master's voice’, the same trademark used by Victor. Adopting Victor's method of recording in wax, the Gramophone Co. began in 1900 to record in every part of Europe and Asia. In 1907, however, Japan was transferred from Gramophone to Victor in the division of markets between the two associated companies, and in 1927 the Victor company of Japan was incorporated. During World War I the German government seized the local branch of the Gramophone Co. as enemy property. It was not returned after the war, so His Master's Voice became the trademark of Deutsche Grammophon Gesellschaft in Germany. Their product was exported bearing a different trademark and the label name of Polydor. Only in 1926 did the Gramophone Co. of England re-enter the German market with the Electrola label, using a new trademark. Electrola recovered the classic trademark in 1949.
Columbia, too, established an English branch which began to create its own recordings only after Louis Sterling became its manager in 1909. In 1903 F.M. Prescott, who had directed the International Zonophone Company, launched a new firm, Odeon, to make machines and discs. Odeon produced the first double-faced record, playable on both sides. The Carl Lindström company subsequently acquired Odeon, as well as Beka, Favorit and Parlophon; Columbia (UK) acquired Lindström in 1925 and Pathé Frères in 1928.
Recorded sound, §I: History of recording
The acoustic process used in all cylinder and disc recording involved collecting sound vibrations in a recording horn attached to a stylus that engraved the sound in a permanent medium. The performers were grouped as close to the mouth of the horn as possible (see fig.1) to secure the maximum transfer of sound energy to the diaphragm that closed the throat of the horn; playback involved tracing the sound with a stylus and transmitting it through a similar horn. With the development in the early 1920s of the vacuum tube or valve for radio broadcast, it became possible to use electronic amplification to record sounds captured by the more sensitive microphone, and to reproduce them using either an acoustic or an electric player, which might be part of any radio receiver. Since the sounds could be conveyed by wires over considerable distances, it was no longer necessary to oblige the musicians to perform unnaturally loudly or congregate round a large collecting horn. Sufficient signals could be picked up from conventional orchestral layouts, and concert hall performances by any number of artists could be captured in well-balanced recordings.
The electrical recording process was pursued experimentally for several years and perfected by Western Electric. In 1922 American Columbia sold its English branch to Louis Sterling, who subsequently acquired the parent firm and with it a licence for the recording process from Western Electric. Victor also acquired a licence and shared it with His Master's Voice (HMV). Electrical recording began in the spring of 1925, but the first issues were publicized discreetly in order not to render the back catalogues worthless. In the next four years whole new catalogues of electrical recordings were created. Both Victor and HMV re-recorded some of their records made by Enrico Caruso and others, adding orchestral accompaniment and dubbing in the recorded voice. The results were disappointing, however, and the original acoustic records remained in the catalogues. Deutsche Grammophon allied itself with a smaller American label, Brunswick, which introduced another system of electrical recording developed by General Electric, using a beam of light to modulate the electrical signal. Both firms used this system for about a year, then switched to the Western Electric system in 1926. Odeon and Parlophone adopted electrical recording only in 1927.
The stock market crash of 1929 affected the record business severely, since the radio was already becoming an alternative to records as a source of entertainment in the home. In England Columbia was merged in 1931 with the Gramophone Co. to form Electric & Musical Industries Ltd (EMI). In the USA the Victor Talking Machine Co. arranged to market its electrical playback machines in radios built by the Radio Corporation of America. RCA purchased Victor in 1929, and some years later the record label was changed to RCA Victor.
The weak market for recordings worsened during the Depression. This led Gaisberg and his assistant, Walter Legge, to solicit subscriptions for proposed recordings. After 500 subscriptions were obtained in 1931 for a set of six discs of Wolf's lieder sung by Elena Gerhardt, a long series of ‘subscription societies’ was established. Among the recordings that followed were Beethoven's piano sonatas played by Artur Schnabel, his violin sonatas played by Fritz Kreisler accompanied by Franz Rupp, Haydn's quartets played by the Pro Arte Quartet, five more albums of Wolf's lieder sung by various singers, Bach's Das wohltemperirte Clavier played by the pianist Edwin Fischer, three of Mozart's operas performed at the Glyndebourne Festival under Fritz Busch, and other music not otherwise available on records.
Early discs were designed to play at turning speeds ranging from 70 to 82 r.p.m.; 78 r.p.m. became the standard in the electrical era. In 1931 RCA Victor introduced a record that would play at 331/3 r.p.m., more than doubling the standard playing time. In the depths of the Depression, however, the attempt to market new playback machines was doomed to failure. Moreover, most of the ‘Program Transcriptions’, as they were called, were dubbed from existing masters with inferior results. Only two recordings conducted by Leopold Stokowski were recorded directly in the new medium, Schoenberg's Gurrelieder and Beethoven's Fifth Symphony; both were withdrawn by 1935. For several years RCA Victor exclusively used Stokowski's Philadelphia Orchestra for new orchestral recordings, but in the mid-1930s the Boston Symphony Orchestra and the New York Philharmonic returned to record for the company as well. In England HMV and Columbia retained the London Philharmonic Orchestra from its inception in 1932, with its founder Sir Thomas Beecham conducting on Columbia and others conducting on HMV.
In Hollywood the motion picture industry tried three methods of recording sound for motion pictures. The Vitaphone, used for Warner Bros. first sound pictures in 1926, synchronized a disc to the film projector. This was soon replaced by sound recorded on the edge of the film outside the picture area; RCA's method modulated a variable area, while Western Electric's modulated a variable density. Both soundtracks were read by a beam of light (see §II, 10, below). RCA Victor also made a few recordings from film soundtracks, dubbing the sound to disc for the home market. The first effort to create what would later be called ‘surround sound’ was Fantasia, an animated film made by Walt Disney in 1939. The Philadelphia Orchestra under Stokowski was recorded in six channels, and selected theatres were specially equipped for exhibiting the film. In Germany BASF developed a method of magnetic recording, drawing a tape coated with iron oxide across an electromagnet modulated by an electrical signal. An experimental recording of the LPO was made in 1936, and the method was used extensively in German radio stations after 1939 (see §II, 9, below).
Recorded sound, §I: History of recording
The only recording format available in the home at the end of World War II was the 78 r.p.m. disc. In anticipation of a change, Columbia began about 1940 to record its masters on 16-inch discs for greater fidelity. For a time the quality of their product, dubbed from the original masters, suffered. Then in 1945 Columbia began to develop a new approach to the long-playing record (LP). The speed of 331/3 r.p.m. was again adopted, but a narrower groove 1 mil (a thousandth of an inch) wide replaced the 3-mil groove, and an unbreakable plastic called vinylite replaced the fragile shellac disc.
Columbia, working in secrecy, began to press a large number of vinyl discs in early 1948. The announcement bringing the LP catalogue to market in June 1948 caught the industry by surprise. Meanwhile, RCA Victor had been developing a different technology since 1939. Adopting a speed of 45 r.p.m., which it called a compromise between fidelity and duration, the company chose convenience over longer playing time with a 7-inch vinyl disc. A wide centre hole enabled the playback machine to contain a rapid changer mechanism within the centre post. RCA Victor, rejecting Columbia's LP, introduced its system in January 1949.
For the rest of that year American record companies waged what came to be known as ‘the battle of the speeds’. Several firms adopted the LP for sets of records, but in September Capitol announced that it would issue its records at all three speeds. In January 1950 RCA Victor decided to issue its album sets on LP, and Columbia began to issue single discs at 45 r.p.m. The introduction of three-speed record changers required 45 r.p.m. records to be made with standard centre holes. 78 r.p.m. records gradually disappeared from the market within a few years, while 45 r.p.m. became the standard for popular songs and the 331/3 r.p.m. LP became the only format for album sets in all categories of music.
In Europe L'Oiseau-Lyre issued the first LP records in France in April 1950, followed within a year by Decca, Discophiles Français, Le Chant du Monde and Columbia. In England Decca, which began exporting LPs to the USA in 1949 under its new label, London, entered the British market in June 1950 at the same time as L'Oiseau-Lyre.
The ease of marketing LP records brought a large number of new record labels into existence, especially in the USA. This coincided with the introduction of magnetic tape to the recording studio as well as to remote halls where recordings could be made with ease. Tape machines found in German radio stations at the end of the war were brought to the USA to be copied, and from late 1949 virtually all commercial master recordings were made on tape. The tape recorder as a recording and playback device for the home came to the American market in 1947. The Amplifier Corporation of America, using masters of the Vox label, brought the first recorded tapes on open reels to market in 1949. By 1953 the major labels were issuing their own catalogues of recorded tapes, which were regarded as superior to discs in relative freedom from noise.
Recorded sound, §I: History of recording
The capture of two distinct sound channels from one sound source, just as human ears operate, was perceived to be a method of achieving greater spatial realism in a recording. Bell Telephone Laboratories made experimental stereo recordings of the Philadelphia Orchestra under Stokowski in 1932, and in Germany, Walter Gieseking was recorded playing Beethoven's ‘Emperor’ Concerto in 1944. Early experiments followed two paths. One technique placed two microphones close together to produce a binaural recording that was played back through headphones. The other technique captured two channels of sound, using two or more microphones, which were played back through speakers. The latter system prevailed.
The first stereo playback device for the home consumer was developed in 1952 by Emory Cook, who attached a second cartridge to the LP playback arm, like a sidecar on a motorcycle. This enabled the listener to play discs produced by Cook that had two channels of sound on the same disc, the left channel starting at the edge and the right channel starting in the middle of the disc. RCA Victor began to make most of its new recordings in stereo in early 1954. Later the firm added recorded stereo tapes to its existing tape catalogue. In England EMI began to issue both mono and stereo tapes in 1955. By 1956 most companies were making all commercial recordings in stereo, though only a small number were issued commercially on recorded tape.
Several firms experimented with putting two channels of recorded sound in the same groove of a disc. Western Electric developed a way to combine the two signals in such a way that the sum and difference of the two groove walls yielded stereo sound, channelled to separate amplifiers and speakers. This system was perfected at the end of 1957, when the first Westrex stereo record was offered to a market that lacked any playback equipment. The ploy successfully generated demand in the marketplace, and by June 1958 the stereo disc was a marketing reality.
In the first few years of stereo discs, several labels tried to preserve their monophonic recordings in the marketplace by reissuing them in electronically enhanced stereo. The sound signal was artificially divided into two channels, giving a suggestion of stereo. After a time consumers came to accept the original mono as genuine and no further enhanced discs were issued.
The open-reel tape playback deck was not a simple device to use in the home, so attempts were made to develop a cartridge or cassette that could be played as easily as a disc. In 1959 RCA Victor marketed such a cartridge utilising two inner reels, while in 1963 Columbia produced a small reel using narrow tape that would be taken up inside the player. Several types of four- or eight-track cartridges used a continuous loop of tape flowing between two interior reels.
In 1965 Philips combined a smaller version of the first RCA cartridge with the narrow tape that Columbia had used to produce the small cassette that eventually became the favoured tape format in the consumer marketplace. Originally developed as a low-fidelity device, it was improved to the point that it became as important as the LP in the market. Cassette players were installed in automobiles and Sony developed the portable personal stereo. By 1970 the open-reel tape occupied a peripheral place in the market, although it remains the recommended mode of archival storage of recordings.
In a further refinement of stereo sound, several systems of four-channel, or quadraphonic, sound were developed and marketed. Two more channels were added to the existing channels on a high-frequency carrier, which could then be decoded by the playback amplifier. Two competing formats, SQ and QS, were more or less compatible. Another approach promoted by RCA Victor placed four discrete channels in the walls of a single groove, yielding more complete separation of the channels in playback. All of these discs could also be played using a stereo cartridge, although the quadraphonic sound would not be reproduced. Quadraphonic sound was also marketed in a tape cartridge known as the Q8 format, adapted from the eight-track tape cartridge (see §II, 9, below). These systems were marketed from late 1971, but the failure to establish a standard format made the marketing of home equipment difficult, and quadraphony soon lost its appeal. Surround sound, employing five channels and a non-directional bass channel, was a later development.
Recorded sound, §I: History of recording
The most radical change in sound recording since the introduction of electrical recording was developed by Denon, the Columbia firm in Japan, in the early 1970s. All sound recording since the beginning had produced an analogue pattern in a physical medium. Digital recording, originally called ‘pulse code modulation’ (PCM), reduced the wave form to digital code. This called for more complex electronic circuitry, both to encode the signals and to decode them on replay, but produced a dramatic improvement in the quality of reproduction. The first LP transferred from a digital master was a collection of Telemann fantasies for the flute issued in 1972. Denon built up the first catalogue of digital recordings in cooperation with Eterna (DDR) and Supraphon. By 1980 most firms were using digital recorders and transferring the sound to LP; these discs were identified as digitally mastered. Even greater benefits could be obtained by keeping the music signals in digital form all the way to the listener's home. This was accomplished in 1982 when an agreed standard for a digital record led to the launch of the compact disc, developed by Sony and Philips. Sales of the new CD soon rivalled those of the vinyl record and, within only about ten years, dominated the pre-recorded music market worldwide, although the Philips cassette remained a viable competitor for recordings of wide appeal. The CD is 120 mm in diameter and 1.2 mm thick, yet it easily surpasses LP sound quality in terms of frequency coverage and dynamic range. Maximum playing time is nominally 80 minutes; it has been suggested that the initial playing time of 74 minutes was insisted on by Norio Ohga, the head of Sony, to allow Furtwängler's recording of Beethoven's Ninth Symphony to be accommodated on a single disc.
A new even smaller digital record format was launched by Sony in 1992. The MiniDisc (MD), 64 mm in diameter, offered the same playing time and other features as the CD (see §II, 9, below), although with marginally poorer sound quality. On the other hand, the MD came with a recording facility which was superior to that of the compact cassette tape system. A recordable CD followed, but it was relatively expensive and used mainly by professionals. The MD made slow progress against the dominance of the CD, particularly as the record companies showed little interest in issuing pre-recorded music on the smaller disc. Philips introduced a new digital version of the compact cassette (DCC) at about the same time (see §II, 9, below). Neither the MD nor the DCC did much to convert the buying public from the conventional analogue cassette, which continued to rival the CD. Newer versions of the CD based on a digital versatile disc (DVD) appeared in 1997; the new format was also capable of storing information that made it applicable to the video and home computer markets.
Recorded sound, §I: History of recording
In 1902, after paper labels had replaced engraved text on discs, G & T created a Red Label to identify its most important records. Victor soon copied the style, calling it the Red Seal. Offering serious repertory sung and played by renowned artists, these labels acquired the highest prestige in the marketplace. The operatic arias were accompanied on the piano until 1906, when a semblance of orchestral accompaniment was introduced. Most of the leading operatic singers began to make records, notably Caruso, who became the most celebrated recording artist of his time. In April 1902 he recorded ten arias for G & T, followed by three cylinders issued by Pathé and seven discs issued by Zonofono. In February 1904 he made his first recordings for Victor in New York, an association that continued until his death in 1921. Caruso's records were immensely successful and helped to make recordings artistically significant. A few complete operas were recorded before World War I. In the early electrical era, many popular Italian operas were recorded in Milan by both HMV and Columbia, and some French operas in Paris. To keep the large sets affordable they were made without star singers and conductors and sold at a moderate price. From the beginning of the LP era operas recorded with the foremost singers and conductors became a feature of all the major record labels. The recorded repertory now ranges from newly composed operas to the lesser-known operas of Handel and the earliest works of the 17th century.
Unlike military bands orchestras did not prove suitable for the acoustic recording process, and only brief orchestral selections found their way to disc before 1909. In that year Herman Finck conducted Tchaikovsky's Nutcracker Suite on seven sides for Odeon. A year later Odeon recorded Beethoven's Fifth Symphony complete on eight sides. In 1913 the first celebrity orchestral recording featured the same work played by the Berlin Philharmonic under Arthur Nikisch, with bassoons replacing the double basses. Elgar and Holst also led valuable recordings of their own works. Before 1920, however, symphonies were generally recorded in abridged form; not until the end of the acoustic era was much of the standard symphonic repertory recorded in uncut versions. Frieder Weissmann conducted all of Beethoven's symphonies (except the Seventh) for Parlophone, which also recorded seven of Richard Strauss's symphonic poems conducted by Eduard Mörike. Oskar Fried conducted Bruckner's Seventh Symphony and Mahler's Second for Deutsche Grammophon.
From the early electrical era to the advent of the LP the scope of the standard orchestral repertory on records gradually increased. The new companies that arose in the late 1940s and early 50s recorded neglected works of every era. The repertory expanded to lesser-known composers, contemporary works and music of earlier centuries. From about the centenary of his birth in 1960, Mahler's music went from the periphery to the centre of the repertory, a development that was heard on records even before it reached the concert hall. Beginning with Vivaldi, the music of the Baroque era came to records in abundance, followed by music of the medieval and Renaissance periods. From the beginning of the stereo era, historically informed performing practice became the norm. Recordings fostered the dissemination of interpretations by performers who repeatedly advanced new notions of acceptable performing practice. Even interpretations of Classical and Romantic music were subjected to historical scrutiny.
The improvisatory nature of jazz means that recordings of it, which began to appear in 1917, provide a particularly valuable documentation. The accumulated repertory has acquired a stature that was unimagined in the earliest days of jazz. While the music of the Broadway stage was recorded irregularly before the LP era, after Cole Porter's Kiss Me, Kate in 1949, virtually every Broadway musical was recorded by one of the four major American labels. These original-cast recordings had a parallel in soundtrack recordings, which were made from the original recordings of film scores.
From the beginning of the LP era the most significant interpretations of the past were gradually reissued, and by about 1980 the process had accelerated. RCA Victor, for example, reissued complete recordings of such leading artists as Caruso, Heifetz and Rachmaninoff. In the CD era the expiration of copyrights has permitted an increasing number of recordings to be reissued freely. Digital technology makes it possible to restore the sound of even the earliest recordings with unprecedented fidelity. Original producers have begun to cooperate with freelance firms (EMI with Testament, for example), to provide first-generation sources for use in making the transfers. In the popular field the reissue of boxed sets of the entire output of the more legendary recording artists has become commonplace. While some LP recordings are not yet available on CD, many 78 r.p.m. recordings that were never issued on LP have appeared on CD.
As records became more and more important in the marketplace, retail sales moved from speciality stores to department stores, where only the fastest-selling items were offered. Record shops offering the widest selection are found only in larger cities; music lovers who live far from such centres depend largely on the mail-order firms and record clubs that have sprung up since the early LP era. In the USA both RCA Victor and Columbia (now BMG and Sony) maintain clubs that send records to members on a regular schedule, even licensing recordings of other labels for sale. The Musical Heritage Society has been the most successful club in the USA since 1965. Concert Hall, which began in the USA in 1946, operated clubs throughout Europe and Japan for many years.
The album has been part of record marketing since 1909, when the Nutcracker Suite was issued in a special sleeved album. In the last days of the 78 r.p.m. era Die Meistersinger was recorded at the Bayreuth Festival under Karajan on 34 discs, later issued on five or six LPs. In 1971 Deutsche Grammophon offered 76 LPs of Beethoven's music, packaged in 12 albums, and in 1991 Philips issued 180 CDs of Mozart's music in a pair of carrying cases.
Recorded sound, §I: History of recording
Each technological development in recording has resulted in increased competition in the marketplace and the enlargement of the recorded repertory. Before 1948 both the recording of masters and the manufacture of records required such skill that a few companies dominated the marketplace. In the LP era many more firms were able to make master recordings on tape and obtain pressings from factories owned by the major companies. In the digital era master recordings of the highest quality can be made with much greater ease and pressed at any commercial disc manufacturing facility.
At the end of the 78 r.p.m. era the American market was dominated by RCA Victor, followed by Columbia, Decca (a subsidiary of the English firm until 1948), Capitol and Mercury. The European market was dominated by EMI, which maintained historical alliances with RCA Victor for its HMV label and with Columbia Records for its Columbia label. Decca became a significant label in the English market in 1946, while in Germany Deutsche Grammophon and Telefunken retained a share of the market. During the first decade of the LP, Decca (UK) entered the American market with the London label and established a partnership in Germany with Telefunken. Columbia (US) entered an alliance with the emergent Philips label, while Columbia (UK) entered the American market using G & T's old Angel label. In 1956 EMI acquired Capitol to replace RCA Victor as the American outlet for its HMV catalogue. In turn RCA Victor entered a cooperative agreement with Decca (UK) that provided international distribution for the RCA label. Deutsche Grammophon was distributed by Decca (US) and then by MGM before it established its own American marketing branch.
In the 1960s Philips acquired Mercury as its American branch, while Columbia entered the international market with the CBS label and transferred its Japanese partnership to Sony. RCA Victor separated from Decca (UK) after 11 years to enter the international market with the RCA label. Philips and Deutsche Grammophon became partners by an exchange of stock between their respective parent firms. Philips later became the senior partner and also acquired Decca (UK). Subsequently several record companies were acquired by corporate conglomerates. RCA sold its record business to Bertelsmann in 1989, and CBS sold its record business to Sony in 1988. EMI was acquired by Thorn in 1979, but the union was dissolved in 1996. Warner (later Time Warner), a firm that had its own popular record label, acquired Elektra, Atlantic, Telefunken and Erato; the last was a small French label that had earlier been acquired by RCA. Philips, DG and Decca, which had become known as the Polygram Group, were acquired by Seagram in 1998. This firm had already acquired Universal (formerly MCA), which had earlier acquired Decca (US), and Polygram was merged into Universal. At the end of the 20th century the worldwide record market was dominated by Warner, Universal, Sony, EMI and BMG.
Recorded sound, §I: History of recording
F. Taylor: ‘Extemporising machine’, Grove1
T.L. Southgate: ‘Recording music played extemporaneously’, Grove1, appx
C. O’Connell: The Other Side of the Record (New York, 1949)
R. Wile: Articles on the early history of the gramophone in America, Association for Recorded Sound Collections Journal (1971–96): ‘Adventures in Edisonia’, iii/2–3 (1971), 7–78; ‘The Rise and Fall of the Edison Speaking Phonograph Company, 1878–1880’, vii/3 (1976), 4–31; ‘The Edison Invention of the Phonograph’, xiv/2 (1982), 4–28; ‘Record Piracy’, xvii/1–3 (1985), 18–40; ‘Jack Fell Down and Broke His Crown’, xix/2–3 (1987), 5–36; ‘Etching the Human Voice’, xxi/1 (1990), 2–22; ‘The Development of Sound Recording at the Volta Laboratory’, xxi/2 (1990), 208–25; ‘Edison and Growing Hostilities’, xxii/1 (1991), 8–34; ‘The American Graphophone Company and the Columbia Phonograph Company Enter the Disc Record Business, 1897–1903’, xxii/2 (1991), 207–21; ‘The Launching of the Gramophone in America, 1890–1896’, xxiv/2 (1993), 176–92; ‘The Gramophone Becomes a Success in America, 1896–1898’, xxvii/2 (1996), 139–70
T. Brooks: ‘Columbia Records in the 1890s’, Association for Recorded Sound Collections Journal, x/1 (1978), 3–36
J. Culshaw: Putting the Record Straight (New York, 1981)
E. Schwarzkopf: On and Off the Record (New York, 1982)
P. Martland: Since Records Began: EMI, the First 100 Years (London, 1997)
1. The signal: frequency range and amplification.
2. Microphone characteristics.
6. Gramophone record and compact disc technology.
7. Gramophone record and compact disc manufacture.
8. Gramophone record and compact disc reproduction.
11. Videotape recording and television.
Recorded sound, §II: Techniques of recording, reproduction and transmission
In the studio, the minute rapid fluctuations in air pressure that constitute sound – referred to as the ‘sound pressure’ to distinguish them from the steady atmospheric pressure – act upon the diaphragm of a microphone, which by its movement generates an electromotive force or voltage proportional to the pressure change and causes a corresponding current to flow. The variation of air pressure with time, while complex in form, can be analysed into simple components – a fundamental plus a series of harmonics – with frequencies extending over a very wide range. Ideally, all these audio-frequency components should be represented, in their proper proportions, in the electrical signal generated by the microphone, and this signal should be preserved intact throughout the whole recording or broadcasting chain. It is not possible with present systems of radio transmission on medium and long waves to reproduce the full audio-frequency range. However, the introduction of VHF/FM broadcasting in the 1950s (see §12 below), coupled with improvements in methods of network distribution, overcame many of these difficulties and, with the later change to digital technology, the ideal can now be closely approached. Whereas a frequency range extending from 20 Hz to 20 kHz was for many years regarded as an adequate target, it is now generally agreed that future high-quality sound broadcasting should aim at preserving all components with frequencies between 40 Hz and 15 kHz; subjective tests carried out with a variety of programme material show that few individuals can detect the change in sound quality caused by the omission of components falling outside these limits. Sound recording, without the constraints imposed by the broadcasting chain, can if necessary be made to cover an even wider frequency range.
The signal generated by the microphone is too weak to actuate the rest of the apparatus in the recording or broadcasting chain, and has therefore to be electrically magnified. Originally, this process was carried out by a thermionic valve amplifier. It had been discovered that the heated filament of an electric lamp emitted a cloud of particles, each representing a small quantity of electricity; by catching these particles on a metal plate mounted within the lamp bulb and connected to a battery, the stream of electricity could be made to flow through an external circuit. Later it was found that if a wire grid was interposed between filament and plate, quite small signal voltages applied to this grid could be made to encourage or inhibit the stream of particles; the complete device thus constituted an electrical valve, which could regulate the flow of current as a tap regulates the flow of water, producing in the process an amplified copy of the original signal, with all its moment-to-moment fluctuations faithfully reproduced. An amplifier incorporating a single thermionic valve could be made to increase the signal power a thousandfold, while even greater magnification was obtainable by a multi-stage arrangement in which two or more valves operated in sequence.
For most purposes the thermionic valve has been superseded by the transistor, a tiny semiconductor device that produces an equivalent effect without the need for a heated filament; the necessary electrical conductivity is achieved by the introduction, during manufacture, of slight impurities into what would otherwise be a poorly conducting material. Frequently ‘integrated circuits’ are used in which a number of transistors, together with the associated electrical components – resistors and capacitors – required to give the equivalent of a multi-stage valve amplifier, are all fabricated on a single wafer of material only a few millimetres square. (Amplifiers of one kind or another perform a number of essential functions in a broadcasting or recording system. They are needed particularly in studios not only to magnify the output signals from the various microphones but to make good the loss of signal strength incurred in various processing operations, and to produce the high audio-frequency power required to operate loudspeakers and recording equipment.)
Recorded sound, §II: Techniques of recording, reproduction and transmission
To indicate the standard of performance of a microphone, and to distinguish one type from another, various characteristics have to be specified. The frequency characteristic or frequency response, for example, shows the way in which the sensitivity of the microphone varies with the frequency of the sound; in general, the less variation the better (i.e. the frequency response should be uniform). Another important characteristic shows the way in which a microphone’s response to a sound depends on the direction from which that sound is coming. It is convenient to represent this characteristic in graphical form by starting from a central point, representing the location of the microphone, and measuring off in each direction a length that represents, to some scale, the sensitivity to sound coming from that direction. The resulting outline is known as a ‘polar diagram’ or ‘polar characteristic’. Fig.2 shows some examples; in each case, M represents the position of the microphone, the front face of which is towards the top of the diagram. This two-dimensional representation relates to sound arriving in the horizontal plane; similar diagrams, which may or may not have the same shape, could be drawn for sound arriving at various angles in the vertical plane.
The circular polar diagram of fig.2a refers to a ‘pressure’ microphone, in which the sound acts only on the front surface of the diaphragm, the rear being enclosed: such a microphone is described as ‘omnidirectional’ because it responds equally to sounds coming from any direction. The figure-of-eight characteristic of fig.2c is obtained by exposing both the front and the rear of the diaphragm to sound; this construction produces a ‘pressure-gradient’ (or ‘velocity’) microphone, which is completely insensitive to sounds coming from either side (or from above or below). The figure-of-eight characteristic can be used to exclude unwanted sounds; in drama, moreover, an actor can give the effect of retreating into the distance simply by moving round to the ‘dead’ side of the microphone, so that the direct sound vanishes, leaving only the reverberation (see below). Probably the most generally useful form of directional characteristic is the cardioid of fig.2d, which covers a wide angle in front and has a ‘dead’ region at the rear. This can be synthesized by combining the signals from two microphone elements having individual polar curves of the circular and figure-of-eight forms respectively. Usually, however, the desired effect is achieved by a ‘phase-shift’ microphone, a variant on the simple pressure-gradient type, in which the diaphragm is directly exposed to sound in front and only partly enclosed at the rear; the same arrangement can be designed to produce the hypercardioid curve of fig.2e, which gives a lower response at the sides and a ‘dead’ region at about 120°.
In a studio some of the sound emitted by the source reaches the microphone directly and some indirectly by repeated reflections from walls, floor and ceiling; the indirect component, which persists for a short time after the direct sound has ceased, is also known as ‘reverberation’. If the microphone is brought nearer to the performer, the amount of direct sound reaching it is increased, but the amount of reverberant sound, which is distributed fairly uniformly throughout the studio, remains about the same; the ratio of direct to reverberant sound received is therefore greater. This ratio, other things being equal, determines the auditory perspective, the sense of nearness or remoteness of the reproduced sound.
All the microphones shown in fig.2, when pointed towards the performer, will give the same response to direct sound; but the response to reverberant sound, which arrives equally from all directions, will depend on the polar characteristic. Thus, an omnidirectional microphone, fig.2a, will respond to all the reverberant sound present, while a directional microphone, fig.2b, c, d or e, will respond to only a part of it. It follows that for a given ratio of direct to reverberant sound (a quantity determined by the acoustic perspective required) a directional microphone can be placed further from the source of sound than an omnidirectional microphone; with the more distant position it is easier to cover uniformly a large group of performers, such as a choir or large orchestra.
Ideally, the shape of the polar diagram should be the same at all frequencies, that is the frequency response should be the same for sounds coming from any direction. If this requirement is not met, the tonal quality of the direct sound will depend on the angle at which the sound arrives, while the ratio of direct to reverberant sound, and hence the acoustic perspective, will vary with frequency.
The frequency characteristics of pressure-gradient microphones and, to a lesser extent, of phase-shift microphones, show a progressive increase in bass response with diminishing distance from the source of sound. This phenomenon, which is known as the ‘proximity effect’ and becomes appreciable at distances of 50 cm or less, has long been used by singers of popular music to give the voice a more resonant character. For a given working distance, it can be compensated by electrically attenuating the low-frequency components of the signal, or by designing the microphone to produce the same result; in some broadcasting commentators’ microphones, the effect is used to reduce the amount of background noise picked up. Some artifices used in microphones to avoid the proximity effect are described in §3 below.
Modern studio microphones can accept the loudest musical sounds without introducing distortion. But the volume range that can be transmitted is limited by background noise, usually in the form of a slight hissing sound, which may be heard during a quiet programme. This effect is produced by random fluctuations in the conduction of electricity within the microphone or its associated amplifiers. A full microphone specification includes a statement of the amount of background noise; this quantity is often expressed as the level of acoustic noise in the studio that would produce the same effect. In specifying noise levels, an allowance (known as ‘weighting’) is made for the unequal sensitivity of the ear in different parts of the audio-frequency spectrum.
Recorded sound, §II: Techniques of recording, reproduction and transmission
Most microphones used in the recording and broadcasting of music and drama fall into one of three categories according to the way in which the electrical signal is generated. In the dynamic or moving-coil microphone, constructed like a miniature loudspeaker, the signal is generated by the movement of a coil of aluminium wire, attached to the diaphragm, in the radial field of a permanent magnet. The motion of the diaphragm is to a large extent constrained by the air enclosed within the body of the microphone; the degree of constraint, and hence the response of the system to sound, is determined by a carefully designed arrangement of air channels and cavities. Many dynamic microphones are of the pressure type and are nominally omnidirectional, but become partly directional, as in fig.2b, at high frequencies, at which their dimensions are no longer small compared with the wavelength of the sound. Others are of the phase-shift type, in which a cardioid polar characteristic is obtained by allowing sound to reach the back of the diaphragm through an aperture, at the rear of the microphone, forming part of an acoustical delay or phase-shift network. In these microphones, the proximity effect can be much reduced by arranging for the external path length from the front of the diaphragm to the rear aperture to be increased at low frequencies. This can be achieved by providing two microphone elements mounted in a common housing; one has a long front-to-back path and operates at low frequencies only, while the other, of normal construction, covers the upper end of the range (see fig.3a). A similar effect can also be obtained by a single phase-shift microphone element with a number of spaced alternative rear sound entrances, those giving the greatest front-to-back distances operating at the lowest frequencies.
In the ribbon microphone, the signal is likewise generated by the movement of an electrical conductor in the field of a permanent magnet. In this case the conductor is a flexible strip or ribbon of aluminium foil or leaf, usually 30–50 mm long and some 5 mm wide, which also serves as the diaphragm. The ribbon is mounted between parallel magnet poles so arranged that the magnetic flux traverses it from edge to edge. Most ribbon microphones are of the simple pressure-gradient type, with both faces of the ribbon freely exposed to sound, giving a figure-of-eight polar diagram. Some, however, are of the phase-shift type, the back of the ribbon being partly enclosed as described above; a variety of polar characteristics can be obtained by adjusting the size of the rear aperture.
The condenser (or capacitor or electrostatic) microphone has a diaphragm of paper-thin metal or metallized plastic, mounted in front of a fixed, electrically conducting back plate, the gap between the two being only a few hundredths of a millimetre. Deflection of the diaphragm by sound pressure alters the gap, and hence the electrical capacitance, between the two surfaces. In most of these microphones, a constant electrical charge is applied to the system; changes in capacitance then produce changes in the voltage between the diaphragm and back plate, which, when amplified, yield the required sound signal. The fixed charge is usually derived from an external power supply; alternatively one may use a diaphragm of special material, known as an ‘electret’, which is capable of retaining an impressed charge indefinitely, as a permanent magnet retains its magnetism. In some condenser microphones the changes in capacitance between the diaphragm and back plate are made to vary the frequency or amplitude of a radio-frequency current (i.e. an alternating current having a frequency of the order of megahertz); this variation is then converted, before amplification, to an audio-frequency signal by an action similar to that of a radio receiver. For technical reasons the first stage of amplification or other electronic circuitry has to be incorporated in the microphone; in some cases, however, the diaphragm and back plate are mounted in a detachable capsule that may be separated from the rest of the equipment by an extension tube of up to a metre or so in length.
Condenser microphones may be of the pressure, pressure-gradient or phase-shift type; in some cases, interchangeable capsules giving different polar characteristics are provided (see fig.3b, c, d). A variable-directivity arrangement has also been produced, with two diaphragms symmetrically disposed on either side of a common back plate, which is perforated to allow the passage of sound. This system may be regarded as two phase-shift microphone elements placed back to back; by varying the charging voltages applied to the two diaphragms (an operation that can be controlled from a remote point), the contributions made to the sound signal by the front- and rear-facing elements can be added, subtracted or altered in amount and a variety of polar characteristics thus synthesized (see fig.3e).
For some purposes a microphone much more directional than any of those so far described is required, so that a wanted sound may be picked up at a greater distance while avoiding excessive reverberation or extraneous noises. Such microphones have been devised, but because of the inherent difficulty in maintaining the same form of polar characteristic over a wide range of frequencies, their performance falls short of the best obtainable with other types; and for music, at least, their use is confined to special cases (such as, in TV, the pin-pointing of a single instrument that happens to be in shot at the moment) in which their long-range capability outweighs all other considerations. The best-known device in this category is commonly described as a ‘rifle’ (or ‘gun’) microphone (fig.3f). It consists of a microphone element of one of the types already described, to which has been added a straight tube, usually 0·5–1 metre long and some 20 mm in diameter, through which the sound must pass in order to reach the front of the diaphragm. The tube is provided with a narrow slit or a series of closely spaced holes, extending over its entire length, so that sound waves can enter at any point. The tube is aimed at the source of wanted sound; sounds coming from any other direction arrive at the diaphragm by paths of different lengths, and their total effect is much reduced by mutual interference. The directional effect is less pronounced at low audio frequencies, and at the bass end of the range approaches that which would be obtained without the tube.
Recorded sound, §II: Techniques of recording, reproduction and transmission
The relatively simple case of single-channel or monophonic recording or broadcasting will first be considered; the same basic principles apply also to the multi-channel systems stereophony and surround sound. In the case of works originally intended for direct listening, the object of the process is, in principle, to enable listeners to hear the performance, as nearly as possible, as if they were in the studio or concert hall. In general, the optimum microphone position will be determined by the need to maintain the balance between the different instruments of the orchestra, and to arrive at a ratio of direct to reverberant sound that gives a sense of spaciousness and perspective without loss of clarity. A soloist performing with an orchestra is usually provided with a separate microphone at relatively close range; the output from the soloist’s microphone is then used to reinforce that of the main microphone by an amount sufficient to give greater clarity to the solo part but without introducing any noticeable incongruity in perspective through the combination of near and distant sounds. Allowance has also to be made for the difference in timbre of the sound radiated by a musical instrument in different directions; with the violin, for example, to make the most of the higher harmonics, the microphone should be on a line roughly at right angles to the belly of the instrument, but to reduce bow noise a position slightly to one side may be preferred.
A high proportion of recorded and broadcast music, however, is produced in a form not intended for direct listening, and the microphone technique is radically different from that outlined above. A large number of microphones are used, some of them placed very close to individual instruments or groups of instruments, so that their combined output, when reproduced on a loudspeaker, gives an effect different from anything that could be heard in the studio. In addition, the signal from any of the microphones may be modified in various ways, for example by increasing the response to sounds in the 1–4 kHz region so as to enhance the sense of ‘presence’, or a microphone designed to produce the same effect may be used. Extra reverberation may be added by a distant microphone or one with its dead side towards the performers; alternatively, artificially generated reverberation may be introduced. An electronic device known as a ‘compressor’ may also be used to reduce the volume range of the reproduced sound (artificial reverberation and compression are discussed below). The result, as heard on the studio monitoring loudspeakers, represents a newly created sound rather than a reproduction, and there is no question of fidelity to an original. The artifices described were originally confined to the recording and transmission of light entertainment and pop music, but some of them, notably the multi-microphone technique, are now increasingly applied to the presentation of classical works.
In stereophony the studio output consists of two signals that may be said to represent the left and right aspects of the array of sounds to be reproduced; at the receiving end of the chain, these signals are applied to the left and right loudspeakers respectively, and can then give the illusion of sounds originating at various points between the two. This process enables the listener to tell how far to the left or to the right the various instruments are placed in the studio, and it may be necessary to modify the orchestral layout to avoid an unbalanced effect when the sound is predominantly on one side or the other for too long at a time.
There are three basic methods of microphone placing in stereophony. In the first, the left and right sound signals are derived from a pair of directional microphones mounted close together, sometimes in a common housing, with their axes pointing respectively half-left and half-right; the relative strengths of the signals from the two microphones then depends on the direction from which the sound is coming. This arrangement is suitable for drama and for musical performances in which the players are grouped within the angle between the two microphone axes. The second system uses a pair of microphones, usually of the directional type, spaced 3 metres or more apart, each covering rather more than half the area occupied by the performers, so that there is some overlap in the centre region. This arrangement by itself has the disadvantage that the reproduced sound appears to be concentrated to left and to right – the so-called ‘hole-in-the-middle’ effect; but the deficiency can be made good by the use of a third microphone, centrally placed, with its output signal equally divided between the left and right channels. This last artifice forms the basis of a third system, which can also be used to supplement the other two. As in some monophonic transmissions, many microphones are used, each placed close to one instrument or group of instruments. The signal from each microphone is then divided unequally between the left and right channels, the ratio of the two contributions determining the position from which the reproduced sound appears to come. The resulting sound lacks reverberation, but this can be added, if required, by a pair of distant microphones connected to the left and right channels respectively, or by some artificial reverberation device with independent left and right outputs. (Reverberation from a single source divided between the left and right channels is unsatisfactory because the resulting sound, which, to give a natural effect, should be distributed across the space between the two loudspeakers, appears in this case to emanate from a single point.) Anomalous directional effects sometimes occur in stereophony, the apparent position of a particular instrument varying according to the pitch of the note being played. This phenomenon, which arises from interference between sounds reaching the microphones by paths of different lengths, has to be dealt with by trial and error.
Quadraphony was an extension of the stereophonic principle much in the news in the mid-1970s; in this case, four signals were generated at the studio and applied, at the receiving end of the chain, to four loudspeakers placed at the corners of the listening area. Several quadraphonic encoding systems competed for public attention but the idea did not catch on and was soon abandoned. By the early 1980s, however, a more sophisticated approach to surround sound began in the cinema using the four-channel Dolby Stereo system. This was extended to the domestic environment through video cassettes and discs, as well as through the medium of television (see §9 below).
In the recording and broadcasting of stereophonic and surround sound programmes, allowance has to be made for the fact that not all listeners are equipped to reproduce these in their original form, and compromises may be necessary in microphone placing and mixing to ensure compatibility between systems. Thus, the two signals making up a stereophonic programme must give a satisfactory effect when combined and reproduced monophonically (i.e. on a single loudspeaker).
Recorded sound, §II: Techniques of recording, reproduction and transmission
The reverberant sound transmitted by studio microphones can be supplemented artificially and the effect of a longer reverberation time simulated. This artifice is used when reverberation is lacking (as in the case of singers who hold the microphone close to the mouth), when the original reverberation time is too short and has to be brought up to normal (for example, if music has to be performed in a heavily damped television studio), or when a long reverberation time is required for special effects in drama. Various devices suitable for particular applications are available, but few are entirely satisfactory for all purposes.
In the earliest form of artificial reverberation, part of the signal from the studio microphones is applied to a loudspeaker in a reverberation room – often referred to, incorrectly, as an ‘echo chamber’ – with hard, bare walls, floor and ceiling; the reverberant sound, picked up by a microphone with its ‘dead’ side towards the loudspeaker (for stereophony, two microphones, spaced 2 metres or so apart, can be used), is then added to the studio output. This method has the disadvantage that the length of the reverberation time cannot be varied except by introducing acoustic absorbent material; moreover, for economic reasons, the rooms are small, and as a result the quality of the reverberant sound produced is not acceptable for all purposes.
Magnetic recording devices using a continuous tape loop or disc provided with a number of playback heads can be made to provide a series of artificial echoes progressively diminishing in strength; if these are sufficiently numerous and irregularly spaced in time, the effect of reverberation can be simulated. In such devices, the equivalent reverberation time can easily be adjusted by electrical means; they may however give audible ‘flutter-echo’ effects, particularly noticeable with impulsive sounds such as gun shots, while with sustained sounds, components at certain frequencies may be overemphasized.
It is also possible to produce artificial reverberation by mechanical systems with a sufficiently large number of resonance modes to simulate the acoustical resonances of a room. Part of the signal from the studio microphones is applied to an electromagnetic drive unit, which produces corresponding vibrations at some point in the system; this unit is equivalent to the loudspeaker in a reverberation room. At another point in the system, a pickup device detects the vibration and generates a corresponding electrical signal. In one system operating on this principle, a thin metal sheet is made to vibrate in the flexural mode; in another, a long spiral spring is driven in the torsional mode. The reverberation time of the metal sheet can be controlled by acoustic absorbent material brought close to its surface, and that of the spiral spring by damping, applied electrically via the electromagnetic drive.
Artificial reverberation as described above is added to the signal coming from the microphones, and is audible only when the programme is reproduced on a loudspeaker. A variant on the magnetic recording system has however been devised in which the delayed signals from a series of playback heads are applied to a number of loudspeakers distributed about the studio, so that the artificial reverberation is audible to the performers. With this arrangement, sometimes known as ‘ambiophony’, great care has to be taken to minimize the amount of sound from the loudspeakers picked up by the studio microphones, otherwise unwanted emphasis at certain frequencies can occur or, in the extreme case, the system may produce a sustained oscillation (colloquially described as ‘feedback’ or ‘howl-round’). Developments in digital techniques provide improved methods of delaying and storing audio-frequency signals, and are increasingly used as the basis for all-electronic artificial reverberation systems.
A compressor is a device designed to amplify audio-frequency signals by amounts that vary automatically with their strengths; the weakest signals are amplified most and the strongest least, so that the final volume range is ‘compressed’ to less than that of the original, and, for a given maximum volume, a higher average is achieved. The amplification rises to its maximum value during pauses in the programme; on the arrival of a signal, it is automatically reduced to the appropriate value in a few thousandths of a second. On cessation of the signal, maximum gain is restored in half a second or less.
If compression is applied to a signal derived from more than one source of sound, the variations in amplification brought about by the predominant component are inevitably imposed on the others; with a sporting commentary, for example, the volume of crowd noise falls and rises according to the volume of the commentator’s voice. For this reason, compression in music programmes is usually restricted to the sound of a single voice or instrument, picked up by a single microphone; if necessary, however, several compressors may be used, each operating on the signal from a different microphone. Care should be taken to minimize the amount of reverberant sound from a compressed soloist picked up on other microphones not subject to compression, or the acoustic perspective will vary according to the soloist’s sound output.
Decisions on such matters as the deployment of microphones and the proportion in which their output signals should be mixed are necessarily based on the whole effect as heard on the monitoring loudspeakers provided in the studio control room. Any peculiarity in the characteristics of these loudspeakers will therefore influence, indirectly, the balance of sound reproduced in the listener’s home. Monitoring loudspeakers are usually chosen on the basis of fidelity to the original sound as heard in the studio; for the purpose of this evaluation, a straightforward single-microphone arrangement is preferable, as this reduces the number of arbitrary factors involved.
In aural monitoring at the studio, it has to be borne in mind that many listeners will be using loudspeakers of inferior quality, or will be receiving the programme under adverse conditions that do not allow the full frequency range to be reproduced. It is not practicable to compensate in the recording or transmission for these shortcomings, but the producer should ensure that the essential features of the programme do not depend on effects audible only to a minority. Similar remarks apply to the disparity between the realistic sound volume at which monitoring is usually carried out at the studio and the much lower volume of a great deal of domestic listening; because of this difference it is sometimes necessary to check the balance of the programme with the studio loudspeaker operating temporarily at reduced volume.
From the microphone onwards, every piece of apparatus in the recording or broadcasting chain is designed to carry signals up to a certain maximum strength. If this limit is exceeded, even momentarily, the reproduced sound may become distorted, equipment may be damaged, or (in the case of broadcasting) interference may be caused to other programmes. It is therefore necessary to regulate the signal at the studio so that the prescribed maximum value is never exceeded, even with the loudest passages of music. It is also necessary to ensure that the quiet passages in the programme are not drowned, at the receiving end of the chain, by background noises arising from radio interference, traffic or other local disturbances; the volume of the weaker signals, therefore, has to be artificially increased and the total volume range of the transmitted programme thus reduced.
To satisfy both these requirements without detriment to the artistic effect of the programme, the usual practice is to employ a skilled (and, where appropriate, musically trained) operator to observe the strength of the signal leaving the studio, with a special indicating instrument, and to regulate the volume in an unobtrusive manner by adjusting the amount by which the microphone output is amplified. The operator is provided with a script or score of the work being performed, and is thus forewarned of any large changes in volume that may require his intervention. For example, a sudden fortissimo in an orchestral programme may be prepared for by surreptitiously reducing the amplification in advance, either slowly and continuously, or in a series of steps, each timed to coincide with some change in the character of the music.
In addition to the volume range, the relative loudness of different broadcast items has to be regulated to avoid, as far as possible, the need for a listener to adjust the volume control of his receiver at each change of programme. To achieve the best compromise in this respect, account must be taken of the probable conditions under which a particular programme will be reproduced in the home. Thus, in broadcasting music of a kind likely to be used as a quiet background to other activities, any announcements should be at least as loud as the rest of the programme. If, on the other hand, the music is likely to be regarded as a concert performance in the home, and reproduced at as near natural volume as circumstances permit, announcements should preferably be kept at a lower level.
Attempts have been made to avoid the need for manual control at the studio by using automatic devices to restrict the volume range and to prevent the signal from exceeding the prescribed maximum level. Quick-acting devices such as the compressor referred to above are not satisfactory for every kind of programme material; because they operate from moment to moment, their action is more obtrusive than that of a human operator, and, as already pointed out, the sound from one prominent instrument can affect the sound level from the remainder. More refined forms of automatic volume regulator have been developed and used with some success in simple cases, such as talks and discussions, as well as in less demanding music in which the volume range is small and the quieter passages, if any, are brief. In general, however, the task of the skilled operator is beyond the capabilities of any practicable robot system. Manual or automatic volume regulation at the studio is sometimes supplemented by an automatic regulator at the broadcasting transmitter; transmitters are in any case protected by a ‘limiter’, a kind of compressor that operates only on the strongest signals, automatically reducing the amplification by the amount required to prevent the volume from exceeding the prescribed value. Any audible effects produced by either of these pieces of apparatus are not, of course, heard on the studio loudspeaker.
Recorded sound, §II: Techniques of recording, reproduction and transmission
Oddly, perhaps, sound recording was achieved before anyone had the idea of being able to reproduce it afterwards. In about 1857 the French typographer and physicist Léon Scott built his ‘phonautograph’, which consisted of an inverted megaphone or horn with a thin membrane stretched across the narrow end. A bristle was attached to this membrane and could be brought to bear on the smoke-blackened surface of a glass cylinder (later covered with a roll of paper). As the membrane responded to the vibrations of sound waves, its movements caused the bristle to etch a wavy line on the cylinder. Louder sounds caused the line to move further from side to side (greater amplitude); higher-pitched sounds produced a higher rate of vibration (higher frequency). With such simple means Scott had developed a unique tool for his purpose, which was to study and analyse different speech sounds, the harmonic contents of musical sounds, etc. The pure tone of a tuning-fork, for example, was found to produce a cyclically repeating waveform: the number of such cycles inscribed in one second could be measured on the paper record and accurately related to musical pitch. Raising the pitch by one octave doubled the number of cycles per second, and accordingly halved the length occupied on the paper by each cycle (the wavelength). If the speed of turning the paper was doubled, the recorded wavelength for any given frequency was similarly doubled. This strict relationship between frequency, wavelength and the speed of the recording medium remains basic to all systems of recording. It sets a limit to the highest frequencies that can be reproduced, and determines the optimum tip dimensions of pickup styluses, tape-head gaps and so on.
Thomas Edison’s ‘phonograph’ consisted essentially of a grooved cylinder covered with tin foil and rotated by a crank (see fig.4). A sort of speaking-tube was connected with the cylinder by a sharp metal point, which indented the tin foil in response to the sound vibrations in the air of the tube; on lowering the needle at the starting-point, and again turning the handle, the chattering of the needle in the indentations produced sounds that somewhat resembled the original speech. A poor imitation it must have been, yet the first appearance of Edison’s commercial phonograms was greeted with tremendous excitement. In 1888 the Illustrated London News said of a recording of Israel in Egypt made at the Crystal Palace that it ‘reported with perfect accuracy the sublime strains, vocal and instrumental’ (see fig.5). But a few people had reservations: Sir Arthur Sullivan made a recording that same year including the words ‘I am astonished … and terrified at the thought that so much hideous and bad music may be put on record for ever’.
After the experiments with foil Edison began using soapy wax for his cylinders; the surface could be shaved with a sharp blade to ‘erase’ the recording and make way for another. The stylus moved in and out to follow the vibrations, so that the modulations in the groove became referred to as ‘hill and dale’, to distinguish them from ‘lateral’ recordings in which the stylus moved from side to side in the groove, producing waveforms like those of Scott’s phonautograph. Lateral recording was adopted by the next important pioneer of recording, Emile Berliner, whose ‘gramophone’ used flat discs instead of cylinders (see fig.6). As with Edison’s commercial machines, the recording (and reproducing) head was driven across the recording surface at a predetermined pitch (grooves per inch) by means of a lead screw that shared its drive with the motor causing the disc, or cylinder, to revolve. Berliner first used a disc of 12·7 cm diameter revolving at 70 r.p.m., with the spiral groove cut from the centre outwards.
The disc system had two technical drawbacks. First, with the constant turning speed of the disc (78 r.p.m. became an early standard, after a period when anything from 74 to 82 r.p.m. was used) the linear recording speed falls continuously as the groove spirals towards the centre; so, as Scott had shown, the recorded waveform is progressively cramped towards the centre until, at the end of each side, the shortest wavelengths (highest frequencies) can be traced only inefficiently by the reproducing stylus. There are other inherent causes of distortion because of the changing parameters associated with this slowing down of recording speed. Attempts to combat ‘end of side distortion’ were ingenious but short-lived. They included motors that speeded up as the pickup tracked across the record, to keep the recording speed roughly constant, and the recording of alternate sides of record sets from the inner groove outwards, so that the difference in quality at the changeover point would be less obvious. The cylinder avoids this problem as it does the second technical difficulty of the disc. This is ‘tracking error’, which comes about because the reproducing stylus does not strictly follow the path across the record taken by the cutting stylus. The latter, as noted, is driven by a lead screw and moves along a radius directly towards the record centre. At all times, therefore, the lateral vibrations of the cutting stylus are at right angles to the groove. On playback, with a pickup arm pivoted at its remote end, the head will track in an arc across the record and so the stylus plane of movement will mostly be at some angle other than 90° to the groove. A few, more advanced tangential tracking arms have appeared, but for most record players this form of distortion remains.
With the introduction of electrical recording in the mid-1920s (see §I, 3 above), the fidelity of recordings was enormously improved: a much wider range of sound frequencies and dynamics could be impressed on the record; electrical cutter heads, which acted like microphones in reverse by converting the energy of the electrical currents into mechanical vibrations of the plough-shaped cutting stylus, were both more efficient and less liable to uneven frequency response than the old acoustically driven cutters; and, above all, the bass frequencies were at last reproduced and gave a fullness to the sound matched by cleaner reproduction of the treble. At first electrical recordings were played on existing acoustic gramophones, but people soon changed to electrical reproducers. In electrical reproduction the gramophone pickup head is again a generator in which the movements of the stylus produce an equivalent electrical current. This tiny signal is amplified and applied to a loudspeaker that reverses the process, being constrained to perform vibrations and radiate sound waves into the air. Soon the record players added such refinements as volume and tone controls, extension loudspeakers and automatic record-changing mechanisms for which ‘automatic coupling’ sets of records were produced.
World War II seriously interrupted progress in sound recording. In the USA, for example, there was a ban on the use of shellac for non-military purposes as well as a crippling dispute with the American Federation of Musicians that prevented new records being made for over two years. Immediately after the war one technical advance followed another at a rapid pace. First came Decca’s ‘ffrr’ (full frequency range reproduction) records, which added a couple of octaves to the span of earlier recordings and which the Gramophone described as ‘a new and very exciting page of gramophone history’. The EMI group similarly extended the range of their recordings; gramophones and radiograms too were soon matching the records in sound quality. In the USA 275 million records were sold in 1946 and 400 million in 1947. This was all the more surprising considering that the shellac discs, with abrasive filler powders such as emery, were in a form hardly changed since the early 1900s. They were brittle, scratchy and had to be turned over or changed every four minutes or so.
In June 1948 Columbia Records gave the first demonstration of their new non-breakable microgroove disc. Each 30-cm LP contained about 25 minutes of music per side, the extended playing time being achieved both by slowing down the turntable and packing more grooves to the inch. The new running speed at 331/3 r.p.m. quickly became a world standard. The new pitch, or groove spacing, was about 100 grooves to the centimetre instead of under 40. The quality of the reproduction was universally acclaimed and was in no small measure due to the entirely new pressing material; this was the unbreakable plastic vinylite, or polyethylene, whose smaller grain structure made the microgrooves possible, and potentially increased the frequency and dynamic ranges of recording, also giving a substantial reduction in surface noise compared with the abrasive-filled shellac records. Again, as lightweight heads were obligatory for the new, softer materials, no filler was required to give strength to the record material. Thus lighter, flexible records became possible which, though relatively easy to scratch, did not break easily and had important storage and handling advantages quite apart from their much longer playing times per side.
A further step towards high fidelity was the introduction of stereophonic records in 1954. These restore to the listener the sense, totally lost in single-channel monophonic recording, of space and of separate sound sources. The aural mechanism of this process is not fully understood, but stereophonic reproduction depends on the recording of two (or more) signals from different positions (see §II, 4 above). Suitably impressed in the record groove and thence reproduced through a spaced pair of loudspeakers or special headphones, these signals combine to re-create a spatial image at the listener’s ears more or less closely resembling the sound spread that would be experienced at an actual performance.
Having perfected the techniques for stereophonic recording and reproduction, the designers moved on to four-channel or ‘quadraphonic’ sound. This uses at least four microphones for recording and requires at least four loudspeakers, with the listener in a roughly central position, and thus surrounded with sound. The recording and reproduction of four discrete channels is simple on magnetic tape. For gramophone records or direct broadcasting, however, a matrix system was generally used, which encoded the four channels on to two for disc cutting; a decoder had to be added to the user’s record player, as well as the extra two amplifiers and two loudspeakers, before the quadraphonic effect could be enjoyed.
The late 1970s saw the beginnings of a new development with greater potential for technical advance: ‘digital’ recording, in which the signals are not direct imitations or ‘analogues’ of the sound waveform but are first converted into a series of coded pulses. Many record companies and broadcasting stations changed to digital recording for their master tapes; this eliminated sound degradations associated with the tape process, such as speed irregularities, overload distortion and tape noise. In particular, digital recordings are proof against interference during storage and transmission since the reproducer has only to detect the presence or absence of pulses, not their amplitude. Records and broadcasts made from digital masters are thus potentially of higher quality. The advantages of compact discs are equally impressive. The digitally encoded signals are pressed in to the upper surface of a transparent polycarbonate substrate in the form of a spiralling track of tiny pits (see fig.9). The pitted surface is then given a thin reflective coating (usually aluminium) and a final protective coat of lacquer on which the label can be printed. The track begins near the disc centre and spirals outwards. Unlike the LP, the linear tracking velocity is kept constant and so the disc rotational speed has to begin high, about 500 r.p.m. and fall steadily to about 200 r.p.m. at the outer edge. The digital track contains a synchronizing code (clock) which automatically maintains correct speed, and all unwanted speed fluctuations are eliminated.
Instead of a mechanical stylus bearing down on to the recorded surface, the playback head is located beneath the disc and makes no physical contact with it, thus avoiding wear (see fig.10). The head comprises a laser light-beam source, sharply focussed through the clear PVC on to the light underside of the recorded track, and a light-sensitive photo detector which registers the presence or absence of light reflections from the pits. This generates an output electrical signal which recreates the original stream of digital data. Dedicated codes embedded in the data provide various functions which would be impossible to an analogue medium. These include almost instantaneous cueing to any track on the disc; visual displays; repeat, random and programmed track sequence playback modes; and error correction which enables the player to ignore the effects of dust or scratches.
Recorded sound, §II: Techniques of recording, reproduction and transmission
Originally it was the practice to record for the gramophone on wax discs of 33 cm diameter and 2·5–5 cm thickness. These were difficult and time-consuming to set up and operate, a nuisance during the inevitable false starts, re-takes and technical hitches of a recording session. Magnetic tape machines have greatly improved the versatility of master recording; they are easy to set up, can be run in sequence to cover extra long takes or just as easily be restarted to take in any number of short repeats. Once every note has been recorded, together with any repeats or corrections, the main advantage of tape becomes apparent – the ease with which it may be cut and spliced. First the recording manager arranges a listening session with the musicians at which the best of any alternative takes are selected. Details of these are marked on a cue sheet and orchestral score and passed through to the tape editor who, while consulting the score, splices the selected passages together to produce the final complete master tape. Editing music tapes is a highly skilled job and, because many of the joins are at the end of a phrase where a cut in studio reverberation or sustaining pedal stands out, any slight mistake is liable to show.
When the master tape is assembled and passed, it is then transferred, or ‘dubbed’, to disc. Operation of the cutting lathe and associated apparatus tends to be fully automatic so that the lead-in, lead-out, finishing and banding grooves are programmed beforehand. The pitch or spacing between grooves is greater at these points, which means that the speed of the motor driving the lead screw carrying the cutter carriage across the disc has to be accelerated at the right moments, and arrested altogether for the finishing groove that completes a circle. Fine control of the lead screw velocity, in an invention by Arthur Haddy of Decca, has produced the technique known as variable-groove recording; this makes more economical use of the recording area by moving the grooves closer together during quiet passages (less side-to-side amplitude in the recorded waveform) and spacing them further apart to accommodate loud passages. Sensing heads on the tape replay deck scan the tape at a distance equivalent to one revolution in front of the reproducing head proper. They collect advance information as to the nature of the recorded amplitude, frequency and vertical–lateral distribution (in stereo) of the signal. This information is automatically used to vary the groove spacing and even the depth of cut. Marked improvements in dynamic range and playing time per side have resulted.
For the next stage in record processing, the direct disc recording is submerged in an electro-plating bath and a copper ‘master’ is ‘grown’ on to it. When peeled off, the master is a copy of the disc in reverse, with ridges instead of grooves, and could be used for pressing out records of some suitable plastic material (see fig.11). However, since it would soon wear out, the plating process is twice repeated so as to produce first a ‘mother’ (with grooves like the original) and then a ‘matrix’ or ‘stamper’, which is given a hard chromium plating and used for the actual job of pressing out records; when a stamper is in any danger of wearing out, further stampers may readily be produced from the mother. After finding where to drill the centre hole in the stamper, done by reference to a concentric line cut on the outside of the original disc for this purpose, the stamper is then placed in one jaw of a hydraulic press and the stamper that will produce the other side of the record is placed on the other jaw. Then the two paper labels are put into position on the vinyl plastic record material, which is in the form of Geon chips or a tarry-looking ‘biscuit’. The heated jaws close to plasticize and press out the material, then are rapidly cooled and opened. There is a fuzz of surplus material round the disc that is roughly cut away and the edge is buffed smooth on a rotating machine.
The manufacturing cycle for compact discs is very similar (see fig.12). Again it begins with the recording of a studio master, which is almost invariably in digital form whether recorded on a tape machine or a computer-controlled digital disc recorder. The digital medium produces a technically superior master with enhanced editing facilities. The signals are then transferred to the photo-sensitive coating on the surface of a glass disc master, using a laser beam which is interrupted in accordance with the digital bit-stream to expose a series of tiny areas along a spiralling track.
Then, in a process that resembles photographic developing, the coating is dissolved away to leave pits in the surface. As in gramophone record processing, this master disc is put through a series of electro-plating stages to provide a nickel father from which a mother and successive stampers can be produced. A high-speed automated press then stamps out discs of clear polycarbonate, having the required spiral of pits on to which the reflective metal layer is deposited, followed by the protective layer of lacquer and label printing.
Recorded sound, §II: Techniques of recording, reproduction and transmission
As already described, the waveform corresponding to the complex sequence of changing sound pressures picked up by the microphone(s) is ultimately moulded into the fine grooves of a gramophone record. The reproducing head or ‘pickup cartridge’ has the task of reconverting this waveform into a corresponding electrical signal that can be amplified as necessary and sent to the loudspeakers or headphones. All pickup cartridges therefore consist of a fine stylus, suitably pivoted so that motion of the tip in following the groove undulations will cause a cantilever arm to rock correspondingly, and a miniature generator capable of transforming this rocking mechanical energy into an electric current. Most cartridges use the principle of electromagnetism to generate the current. The ‘moving magnet’ cartridge (fig.13a), for example, has a tiny permanent magnet fixed to the upper end of the stylus cantilever; the magnet rocks close to a coil of fine wire and the moving field of the magnet induces an alternating current in the coil. In a reversal of this idea, the ‘moving coil’ cartridge (fig.13b) has a coil fitted to the cantilever and the magnet is stationary. It is this latter form of construction that is the basis of the cutterhead used to etch the waveform on the original master disc. The alternating current signal to be recorded is passed through a coil pivoted in the field of a fixed magnet; this causes the coil to rock about its pivotal point and the plough-shaped cutting stylus inscribes the resulting waveform in the soft lacquer surface of the disc. At the same time, the whole head is tracked towards the centre by means of a lead screw so that the familiar spiral groove results. ‘Crystal’ and ‘ceramic’ cartridges depend on the piezo-electric principle in which crystals of certain materials, notably quartz, exhibit an electrical voltage between opposite faces under the influence of an applied bending force (fig.13c). Another type of generator is met in the ‘electret’ cartridge, a tiny capacitor carrying a permanent electrical charge. Vibration of the stylus causes the capacitance to oscillate about its average value and thus generate the required alternating current.
For monophonic gramophone records the plane of movement of the cutting stylus, and therefore of the reproducing stylus, is effectively parallel to the surface of the record. To introduce stereophonic recordings, which require two channels of information with a minimum of interference (crosstalk) between them, it became necessary to apply the driving force to the cutting stylus simultaneously in two planes mutually at right angles. At first, a combination of lateral and hill-and-dale recording was tried with the left-channel signal, say, driving the stylus in the horizontal plane and the right in the vertical plane (fig.14a and b). However, this arrangement is obviously unbalanced and a better solution was quickly established. In this 45°/45° system of stereo recording, the left and right signals are still applied at 90° to each other, but the axis of reference is turned through 45° so that the left plane is at +45° to the record surface and mainly affects the left wall of the groove, while the right plane is at –45° (fig.14c and d). A stereo pickup cartridge therefore has twin generators so as to produce separate left and right output currents boosted in a twin (stereo) amplifier before being sent to the required pair of spaced loudspeakers. An important advantage of the 45°/45° system, apart from the obvious one of improved symmetry, is that such a stereo groove can be traced satisfactorily by a standard monophonic pickup. The resultant lateral motion of the stylus effectively generates a current that is the sum of the left and right stereo signals and so an acceptable mono signal is reproduced. Reverse compatibility is also achieved in that a stereo pickup will play mono records perfectly well; in following the lateral recorded waveform in the mono groove, the stereo stylus produces identical signals in its left and right output circuits and so the stereo loudspeakers similarly radiate equal sounds and the listener hears the music reproduced monophonically as if from a point midway between the loudspeakers.
In the early days of 78 r.p.m. shellac records, the disc material was very hard and the relatively soft steel or thorn needles were quickly worn to an exact fit in the groove. Modern plastic records are of soft, and less noisy, material and modern styluses are made as hard as possible. This is to ensure that the precise tip dimensions created in manufacture will be maintained over a playing life of many hours. Diamond-tipped styluses are preferred; these give 1000 or more hours of playing before the ‘flat’ worn due to friction against the walls of the groove is sufficient to cause audible distortion or endanger the recorded waveform. Cheaper record players are sometimes fitted with sapphire-tipped styluses but these last only about 40 hours.
Best results from LPs require precise dimensions of the stylus tip so that the stylus will sit squarely on the shoulders of the groove without sinking too low. If ‘bottoming’ occurs, there is a serious increase in background noise. The most common tip shape is hemispherical and the optimum tip radius is 0·065 mm for 78 r.p.m. records, 0·025 mm for mono LP records and 0·0125 mm for stereo records. Although they are more difficult to manufacture, and therefore relatively expensive, elliptical-tipped diamond styluses are usually fitted in more complex record players. The preferred major and minor axis radii are 0·018 mm and 0·0075 mm respectively. The stylus must be precisely mounted in the cartridge so that the major axis sits across the groove (fig.16a); the minor axis is better able to follow the smaller waveforms associated with high frequencies in the music, particularly as the waveform becomes more cramped towards the centre of a record.
Some distortion is inevitable in the reproduction of gramophone records since the groove waveform was originally inscribed by a sharp, plough-shaped cutting stylus. A hemispherical playback stylus tends to be pinched by high-frequency waveforms so that it rides up and down and so produces unwanted harmonics of the original tones (fig.16b). An elliptical stylus is a better approximation to the cutter shape and so the ‘pinch’ effect is less marked.
A compact disc player substitutes a microprocessor-controlled disc platform for the ordinary gramophone motor and turntable platter. The gramophone pickup and stylus are replaced by a travelling head assembly comprising a laser light beam source and light-sensitive detector that converts the stream of reflected on/off pulses from the pitted track into a digital electrical signal. This is taken to a digital-to-analogue converter which supplies a suitable output for passing to an amplifier and (for stereo) pair of loudspeakers.
The physical dimensions of CD tracks are microscopically small. A single disc contains about 5 billion pits and there are about 6000 tracks per centimetre, compared with 100 on an LP. Great precision is clearly needed in the tracking system and the focussing of the laser beam, yet CD players have generally proved reliable and relatively inexpensive. Most of the forms of distortion inherent in LP reproduction are completely eliminated, though a certain hardness of tone characterized early players and discs, until designers fully understood the importance of screening and anti-vibration mounting of the disc platform and the optical carriageway, and developed more accurate digital-to-analogue converters (see fig.17).
Recorded sound, §II: Techniques of recording, reproduction and transmission
The origins of magnetic recording can be traced back almost as far as those of cylinder and disc recording. Valdemar Poulsen (1869–1942) patented his ‘telegraphone’ in 1898 after about four years of research into methods of recording telephone messages for later transmission at a higher speed to economize in telephone line time. His original device recorded sounds, as a pattern of magnetized domains, on a steel piano wire about 1·5 metres long, wound on a stationary drum (fig.18). An electromagnet tracked along the wire to record the signals, and retraced it to reproduce the sounds. In about 1900 Kurt Stille introduced a modified machine in which the steel wire was drawn past a fixed head and wound from one reel to another. By about 1929, notably in the ‘Blattnerphone’, the wire was replaced by a flat steel band, 6 mm wide and 0·2 mm thick, because the round wire was inclined to twist. Such recordings could be edited, and high-frequency erasing and bias were developed about 1927.
It was not until flexible coated tapes were introduced that magnetic recording gained real momentum. BASF showed their first tapes and ‘magnetophon’ recorders at the 1935 Berlin Radio Fair. Paper was used at first, then plastic-film-based tapes. By 1947 tape recording had been adopted by broadcasting and gramophone studios all over the world, and it took less than two years to oust disc recording for the storage of radio programmes and the preparation of studio masters (see §I, 4–5 above). Its more important advantages are ease of setting up, compared with direct-cut discs, increased length of playing time, ease of editing and the possibility of recording a number of tracks simultaneously. These tracks of course remain synchronous, and it is a simple matter to record the outputs of various microphones or groups of microphones on separate tracks. The final mixing operations can be postponed to a later, and perhaps less expensive, time. Recordings are often made without some of the performers and their tracks are added later.
In magnetic recording the record head is a shaped electromagnet with a fine coil of wire and a core of magnetic material formed so that the magnetic field, produced as the signal current flows through the coil, is concentrated across a narrow frontal gap (fig.19). The magnetic coating on the tape consists of many tiny elongated particles that can be regarded as miniature bar magnets. As the tape is drawn past the record head, these particles will be pulled into alignment with the polarity of magnetism existing at the instant of passing the head gap. This leaves a record of the alternating-signal current in the form of a sequence or pattern of magnetic polarities and densities of magnetization along the length of the tape. The magnetic recording is reasonably permanent, though there is a risk of the signal’s being lost or weakened if a stray magnetic field (e.g. from a loudspeaker magnet) is brought too close to the tape. To replay the recording, the tape is wound back to the beginning and then re-run past a replay head. This resembles the record head (indeed, they are one and the same on the simpler domestic recorders) and the changing magnetic flux on the tape sets up a field at the frontal gap that induces the required electric signal current in the head coil. Ideally, during the recording process the tape should be in a completely neutral or demagnetized state when it reaches the record head; this is achieved by placing another sort of head just in advance of the record head. This ‘erase’ head is fed with a strong current at some supersonic frequency (generally at least three times the highest sound frequency to be recorded). Then, as the tape passes the erase head, it is first saturated with the high-frequency signal but, as each portion of the tapes moves on, it is subjected to rapidly reversing cycles of magnetization that continually diminish in strength so as to leave the magnetic particles in a completely random (i.e. de-magnetized) state. In practice, a portion of the supersonic tone is also mixed with the sound signal being fed to the record head. This is known as high-frequency ‘bias’ and has the effect of reducing distortion.
The first commercial tape records issued in the early 1950s met with little success. For the quality of tape reproduction to rival that of discs, a relatively high running speed (19 cm per second) and track width (no more than two tracks on a 6 mm tape) seemed necessary. This put tape at a distinct economic disadvantage, since the basic raw tape was much more expensive than the small amount of vinyl needed for a gramophone record. Also, tape duplicating requires the laborious, though much speeded-up, recording of every new copy. The introduction of ‘stereosonic’ recordings in the mid-1950s aroused some interest in tape records. However, very few people seemed ready to buy the elaborate stereosonic tape player units, and conventional mono machines could not reproduce both the stereo ‘half-tracks’ through the necessary twin loudspeakers. Inevitably, therefore, when stereo discs appeared in 1958, tape records were again relegated to the background. Successive economy measures were introduced in an attempt to make tapes more competitive with discs: the slower speed of 9·5 cm per second became popular; quarter-track tapes (and machines) were launched, to give a further halving of the raw tape needed. Unfortunately, these changes were to some extent counter-productive because owners of older machines could not play all the new tape records and potential buyers of new machines found the conflicting formats confusing.
A drawback of all these ‘open reel’ machines was the awkward need to thread the tape on to the empty spool before playing. Accordingly, ready-threaded and enclosed magazines, cartridges and cassettes were developed, including the RCA cartridge (1958), the Fidelipac cartridge (1962) and the Earl Muntz cartridge (1963), but with no more success. The Philips Compact Cassette system has had a wider impact than any other. It was planned as a long-term project, beginning with a small portable recorder in 1963: this used cassettes of 3·8 mm tape running at only 4·75 cm per second, and many manufacturers took up the system so that the existence of large numbers of machines soon made it profitable for Philips and other record companies to start selling pre-recorded cassettes. It was helpful that Philips avoided the tangle of stereo–mono incompatibility into which open-reel tape recorders had fallen. From the outset, cassettes were recorded in quarter-track stereo, but with the left and right signals for each tape side recorded on adjoining tracks (1 : 2 and 4 : 3) instead of divided (1 : 3 and 4 : 2) as on the conventional quarter-track system. Thus all mono (half-track) cassette machines could scan both tracks of a cassette simultaneously and produce an acceptable mono (sum) signal.
In serious competition with the Philips cassette, though mainly designed for use in cars, is the eight-track continuous-play stereo cartridge. This was developed by the Lear Jet Corporation in 1965 and taken up by RCA, Motorola and the Ford Motor Company. The eight-track cartridge tapes are of the standard 6·3 mm width and run at 9·5 cm per second, which, it is claimed, gives better fidelity than the 4·75 cm per second of the cassette; it does not usually incorporate the facility for making one’s own recordings, but the tape is wound in a continuous loop so that it gives uninterrupted music (fig.21). The eight tracks comprise four stereo programmes on pairs of tracks (1 : 5, 2 : 6, 3 : 7 and 4 : 8) with an automatic changeover mechanism that moves the playback head at the end of each track.
Successive technical improvements to the Philips compact cassette medium increased its popularity as a home recording format and persuaded the record companies to issue prerecorded cassettes (musicassettes) simultaneously with almost every LP, and later, CD version. The improvements included a change from the iron oxide tape coating to chromium dioxide, better duplication technology and Dolby B noise reduction. While sound quality remained inferior to that from LP records, and more especially CDs, it was quite acceptable for music on the move in cars or portable players.
A leap forward in sound quality from cassettes became possible in 1992 when Philips introduced their DCC (digital compact cassette). This came close to CD sound quality, though data compression was used, and DCC machines offered a digital recording facility along with backward compatibility which meant that the machines would also replay, but not record on, the conventional analogue cassettes. However the greater cost of DCC and the simultaneous appearance of Sony’s MiniDisc (see §I, 6 above; fig.22) hindered its public acceptance and it was eventually abandoned.
Recorded sound, §II: Techniques of recording, reproduction and transmission
The development of the microphone and the thermionic valve amplifier, primarily for radio, advanced gramophone recording and reproduction from the acoustical to the electrical era. Aided by another invention, the photoelectric cell demonstrated by Lee de Forest in 1923, they also transformed the silent cinema into the ‘talkies’. Initial troubles delayed the mass production of sound-on-film until 1928, when The Jazz Singer had its sound track recorded on a 40 cm disc at 331/3 r.p.m.
To record the sound track on film, the first steps were, as usual, to set up the microphone and amplifier to produce an alternating electric current whose fluctuations corresponded to the frequencies and intensities in the original sound waves. The current was then passed to an optical system consisting of an electric lamp whose light could be projected through a slit to illuminate a narrow strip of the moving film. The signal current was arranged to modulate the intensity of the light falling on the light-sensitive film. The effect was to record a track of constant width, say 3 mm of a 35 mm film, alongside the picture frames. The film was a negative and, when developed, had a succession of brightness patterns corresponding to the sound signals. When the film was run through a projector, a light of constant brightness was directed through a slit on to the sound track. Light passing through the film actuated a photoelectric cell that converted the light signal into a proportionate electrical signal. This could then be treated in the same way as the signal from a gramophone pickup, namely amplified and fed to one or more loudspeakers placed near the cinema screen. The system just described is called ‘variable density’ optical recording. There are advantages, in terms of the sound intensity range that can be achieved, in the ‘variable area’ method, which allows a constant light intensity to pass through the slit during recording but modulates the width of the track; this is done by an electromagnetic device into which the signal is fed, and the envelope of the recorded track is in fact a replica of the sound waveform.
Optical recording has the advantage of keeping the sound and picture in perfect synchronism and allowing combined prints to be made in any desired quantity by normal photographic processing. It has the disadvantage that the producer must wait for the film to be developed before things like the quality and accuracy of editing can be checked; also, in the early days optical recording fell noticeably short of other recording media in terms of the quality of sound reproduced. For these reasons magnetic recording increasingly recommended itself to film producers and distributors, especially where the musical content of films was of first importance. Using the ‘striped film’ method, the signals are recorded on to a special film that has a track or stripe of magnetic iron oxide coating applied alongside one side of the picture area, with a thin ‘balancing stripe’ on the other so that the film will not lean over when spooled. This method clearly gives optimum synchronizing of sound and picture.
In the late 1980s a breakthrough in high-quality optical film soundtracks was introduced with a four-track system designed to provide a surround sound effect in suitably equipped cinemas. The system, called Dolby Stereo, fed sound signals to left, centre and right front speakers plus one or more ‘effects’ speakers arranged behind the audience (see fig.23). In 1992 the idea was taken further when film soundtracks and cinema installations adopted digital encoding techniques. The so-called Dolby Stereo Digital 5·1 format, and a number of rival encoding methods, provided an enhanced surround sound effect using five tracks for the front speakers and left/right stereo effects channels plus a sub-bass channel to reproduce the popular blockbuster movie sound effects. Domestic television and video systems were introduced with similar playback facilities, using suitable decoders and extra speakers, to form what became known as ‘home cinema’ (see fig.24).
Recorded sound, §II: Techniques of recording, reproduction and transmission
The advantages of magnetic tape for the recording of sound signals were carried over to vision by the introduction in the 1960s of videotape recorders by the Ampex Corporation. Whereas a bandwidth of 20 kHz is sufficient to accommodate sounds of all kinds, the more complex electrical signals involved in the transmission of television pictures necessitate a bandwidth of several megahertz. This requires a very high running speed of the recording medium past the recording and playback heads. The Ampex solution was to run the special 5 cm tape at a normal speed of 38 cm per second but to mount the video heads in a spinning block to give a much greater relative speed. The sound signals were carried on a conventional track near one edge of the tape. Later developments of videotape recorders led to much smaller mechanisms and the introduction of digital encoding to high-quality colour pictures. The gramophone and tape record companies in turn took up the idea of issuing recordings of combined audio and visual productions.
Music programmes are seldom completely satisfactory on TV. In televising an orchestral concert, for example, the smallness of the screen makes the performers much too tiny for a complete long shot to be satisfactory; the producer therefore tends to use several cameras to give alternative close-ups of the conductor or sections of the orchestra, and to cut from one to the other in time with changes in the music. Appropriate paintings, scenery or abstract patterns may be shown on the screen, but the optical memory is as sharp as the optic nerve is sensitive, and it is doubtful whether such pictorial patterns, even if satisfactory for a single broadcast, would stand up to frequent replaying in the way that records do. Operas and certain types of popular music lend themselves to visual presentation, but for most music considerable research remains to be done to discover satisfactory answers to the question.
In the late 1970s there was fierce competition between two videotape cassette formats, Betamax and VHS, which required similar but incompatible VCRs (videocassette recorders) that could be plugged into an ordinary television set for playback and off-air recording. The VHS system prevailed and was soon taken up by all VCR manufacturers.
VHS cassettes use 12·7 mm wide tape loaded into a magazine offering playing times of three hours or more. To accommodate the very high frequencies needed for the video signals, a spinning head-drum is used as in professional video recorders to produce a helical scan, the tape/head speed being 5·8 metres per second whilst the linear tape running speed is only 2·4 cm per second. A conventional mono soundtrack is recorded along one edge of the tape and, at this slow running speed, is of relatively poor quality, with high frequencies sadly lacking.
Recorded sound, §II: Techniques of recording, reproduction and transmission
The commonest method of impressing the sound signal on the carrier is to make the strength of the radio-frequency current depend on the strength of the audio-frequency current. This process, known as ‘amplitude modulation’ (AM), is illustrated in fig.28, in which curve a represents the sound signal arriving at the transmitter and curve b the corresponding aerial current. The time interval from A to B illustrates the conditions during a silent period, when the sound signal is zero and the aerial current is steady. Between B and C, a sound is produced in the studio, generating the complex signal waveform shown in curve a; in the corresponding section of curve b it will be seen that the tips of the radio-frequency current wave, which represent the amplitude (i.e. the maximum value reached during each cycle), trace out the form of the sound signal. In the receiver, there are corresponding variations in the strength of the radio-frequency current; these variations are detected by an electronic device, producing a replica of the original signal.
Although it is not obvious from fig.28, the process of amplitude modulation generates a number of additional radio-frequency components; these cover two regions, known as ‘sidebands’, extending respectively above and below the carrier frequency, the components furthest removed from the carrier being associated with the highest audio frequencies in the sound signal. Ideally, to avoid interference, no two transmitters capable of being received at a given location should have their carrier frequencies so closely spaced that their sidebands overlap. Unfortunately the number of different transmissions that have to be accommodated in the available space in the radio-frequency spectrum is so great that some overlap has to be allowed; the situation is aggravated because for transmissions in the widely used medium frequency (or medium wave) range, reflections from the upper atmosphere after dark can lead to interference between broadcasting stations hundreds of kilometres apart. To minimize interference from unwanted transmissions, medium-frequency receivers are commonly designed, as a compromise, to reject part of the sidebands of the wanted transmission; the components thus sacrificed are those corresponding to the higher audio frequencies, the upper limit of the reproduced sound being then restricted to some 5 kHz or less. In addition, medium-frequency transmitters are often provided with quick-acting compressors, similar to those sometimes used in studios, to increase the sound volume in quiet passages that might otherwise be drowned by interference at the receiver; this expedient likewise represents a compromise at the expense of sound quality, since such compression can produce objectionable effects on some kinds of music (see §5 above).
Recorded sound, §II: Techniques of recording, reproduction and transmission
Recorded sound, §II: Techniques of recording, reproduction and transmission
H.F. Olson: Musical Engineering (New York, 1952, rev. 2/1967 as Music, Physics and Engineering)
J. Borwick, ed.: Sound Recording Practice (London, 1976, 4/1996)
G. Ballou, ed.: Handbook for Sound Engineers: the New Audio Cyclopedia (Indianapolis, IN, 1987, 2/1991)
H.F. Olson: Acoustical Engineering (New York, 1957)
D.E.L. Shorter: ‘A Survey of Performance Criteria and Design Considerations for High-Quality Monitoring Loudspeakers’, Proceedings of the Institution of Electrical Engineers, cv/B24 (1958), 607–23; repr. in Journal of the Audio Engineering Society, vii/l (1959), 13–27
M.L. Gayford: Electroacoustics: Microphones, Earphones and Loudspeakers (London, 1970)
M. Colloms: High Performance Loudspeakers (London, 1978, 4/1991)
J. Borwick, ed.: Loudspeaker and Headphone Handbook (London, 1988, 2/1994)
J. Borwick: Microphones: Technology and Technique (London, 1990)
D.E.L. Shorter: Operational Research on Microphone and Studio Techniques in Stereophony, BBC Engineering Division Monograph xxxviii (London, 1961)
A. Nisbett: The Technique of the Sound Studio (New York, 1962, 4/1979)
D.E.L. Shorter and W.I.Manson: The Automatic Control of Sound-Signal Level in Broadcasting Studios, BBC Engineering Division Monograph lxxvii (London, 1969)
D.M. Huber and R.E.Runstein: Modern Recording Techniques (Indianapolis, IN, 1974, 4/1995)
D. Mellor: Recording Techniques for Small Studios (Edenbridge, Kent, 1992/R)
A.W. Miller: Choral Recordings as History: a Study of the Recording Techniques of Five Choral Organizations (diss., Florida State U., 1992)
R. Gelatt: The Fabulous Phonograph (Philadelphia, 1955, enlarged 2/1977)
P. Wilson: The Gramophone Handbook (London, 1957)
H.G.M. Spratt: Magnetic Tape Recording (London, 1958, 2/1964)
O. Read and W.Welch: From Tin-Foil to Stereo (New York, 1959, enlarged 3/1994)
J. Culshaw: Ring Resounding: the Recording in Stereo of ‘Der Ring des Nibelungen’ (London, 1967)
‘The Phonograph and Sound Recording after One Hundred Years’, Journal of the Audio Engineering Society, xxv/10–11 (1977) [whole vol.]
K.C. Pohlmann: Principles of Digital Audio (Indianapolis, IN, 1985, 3/1995)
J. Eargle: Handbook of Recording Engineering (New York, 1986, 3/1996)
J. Eargle: ‘An Overview of Stereo Recording Techniques for Popular Music’,Journal of the Audio Engineering Society, xxxiv (1986), 490–503
J.M. Woram: Sound Recording Handbook (Indianapolis, IN, 1989)
J.G. Spencer and G.J.Phillips: ‘Stereophonic Broadcasting and Reception’, Radio and Electronic Engineer, xxvii (1964), 399–416
E.R. Rout and A.H.Jones: ‘The Use of Pulse Code Modulation for Point-to-Point Music Transmission’, Radio and Electronic Engineer, xxxvii (1969), 199–207
E. Pawley: BBC Engineering 1922–1972 (London, 1972)
D.E.L. Shorter and J.R.Chew: ‘Application of Pulse-Code Modulation to Sound-Signal Distribution in a Broadcasting Network’, Proceedings of the Institution of Electrical Engineers, cxix (1972), 1442–8
A. Chayes and others: Satellite Broadcasting (London, 1973)
J.R. Bittner: Broadcasting: an Introduction (Englewood Cliffs, NJ, 1980, enlarged 2/1985 as Broadcasting and Telecommunication: an Introduction, 3/1991)
J.R. Dominick, B.L.Sherman and G.A. Copeland: Broadcasting/Cable and Beyond: an Introduction to Modern Electronic Media (New York, 3/1995)
