Cylinders, 78s, LPs and CDs
Some brief comments on the history of audio recording
by Oliver Seely

The World Wide Web has some excellent sites which cover the full history of audio recording. There are pictures and sound clips created on equipment used from the late 19th century and later. References to just a few of those sites are given below.

The brief comments here focus specifically on the process of encoding information on different media, starting with the cylinders of Thomas Edison's first phonograph. The cylinders seen at the left were created acoustically, that is, a stylus mounted on a vibrating diaphragm cut a groove in the cylinder as it rotated. The groove was deeper or shallower depending on the position of the vibrating diaphragm. The diaphragm was set in vibration by sound waves sent down and concentrated in a horn connected to it. A principal difference between the recorders which cut the groove and the phonograph which played them back is that there was a rotating screw on the recorder (or cutting lathe) which governed the position of the cutting stylus. Other than that, the very earliest players looked a lot like the recorders which created the acoustically cut groove. It was common practice for soloists to run up to the recording horn so as to achieve a higher vibration (hence, volume) going into the stylus cutting the groove. An example of an early cylinder phonograph is shown at right.

Standard-size cylinders, which tended to be 4.25" long and 2.1875" in diameter, were 50 cents each and typically played at 120 r.p.m. The early cylinders had two significant problems. The first was the short duration of material recorded, usually around 2 minutes. This necessarily narrowed the field of what could be recorded. The second problem was that no mass method of duplicating cylinders existed. Most often, performers had to repeat their performances when recording in order to produce a quantity of cylinders. This was not only time-consuming, but costly.

A process for mass-producing duplicate wax cylinders was put into effect in 1901. The cylinders were molded, rather than engraved by a stylus, and a harder wax was used. The process was referred to as Gold Moulded, because of a gold vapor given off by gold electrodes used in the process. Sub-masters were created from the gold master, and the cylinders were made from these molds. From a single mold, 120 to 150 cylinders could be produced every day. By mid-1904, the savings in mass duplication was reflected in the price for cylinders which had been lowered to 35 cents each. Beveled ends were made on the cylinders to accommodate titles. Columbia, one of Edison's chief competitors, abandoned the cylinder market in 1912. The original Edison cylinders were cut with vertical "mountains and valleys," rather than lateral "wiggles" as pioneered by Victor. Supporters of the vertical cut argued that it produced a superior sound. The popularity of cylinders peaked around 1905 largely because discs were much easier to mass produce. Edison conceded to this reality in 1913 when he announced the manufacture of the Edison Disc Phonograph. The Edison Company did not desert its faithful cylinder customers, however, and continued to make cylinders until 1929. The Edison Company had been fully devoted to cylinder phonographs, but, concerned with discs' rising popularity, Edison and his associates began to develop their own disc player and discs in secret, but Edison decided to retain the vertical vibrations (mountains and valleys) on his discs. It was many years before the lateral cut was accepted as superior largely because of the effect of lower average force acting in the direction of the vibration. Often the playback stylus (or needle, as it was called) had a higher than necessary mass behind it in the form of a kind of snail- or spiral-shaped expanding hollow metal tube the air hole of which finally fed into the base of the horn. The force exerted on the cylinder by that spiral tube, which could pivot so as to allow the stylus to rise and fall a little with the imperfect radius of the rotating cylinder, was not carefully controlled in early phonographs and ended up producing more wear on a vertically cut groove. The laterally cut groove (with "wiggles") is shown at left and right. The force in the direction of playback (back and forth) was little more than the springiness of the playback stylus in the direction of the wiggle of the groove. Whatever force produced wear at the bottom of the groove did not, at least in principal, produce any vibration (even though the noise of such playbacks testifies to there being all kinds of audio "leakage" from rough groove bottoms). Treat yourself to the sound of some of these early recordings.

Click here to listen to some cylinders at the U.S. National Park Service Web site
click here to listen to some vertically cut discs at the U.S. Library of Congress Web site.

By the middle of the 1920s discs were being recorded at a variety of speeds. Although 78 rpm was said to be the industry standard, there was great variability in recording speeds for many years. English Columbia routinely used 80 rpm into the 1920s. Other companies adopted speeds which varied from 74 to 84 rpm.

Around 1948 the vinyl record was born, that is the first twelve inch "long play" (LP) microgroove records that played at 33 1/3 rpm for up to 30 minutes per side, and seven inch 45 rpm singles and Extended Plays. The format was capable of producing full frequency range. Claims of 20 Hz to 20,000 Hz were common, though the fall- off in volume at either end seemed always either to be conveniently ignored or stated in the unit decibels and in a manner which meant very little to the average audiophile. Still, recordings on a vinyl base with the improved light weight pickup arms and styli offered a remarkable improvement in audio quality over the old 78s . These vinyl records were produced in a similar way to the Shellac 78s, that is, they were pressed from plates produced from a master. By the time LPs appeared, recording on magnetic tape had enjoyed some considerable development and electronic amplifiers produced consistently high sound quality throughout the recording process. The first vinyl disc offered only monaural sound (not stereo) which emanated from a laterally cut groove of constant width as in the images above. Note in the right image above how close one groove comes at several points to the next. A particularly loud passage sometimes produced sufficiently large "wiggles" to break over into an adjacent groove, so in order not to have to limit the loudness, a system of "dynamic separation" was developed to detect the loudness on the master tape a distance ahead equal to one revolution of the disc so that the cutting stylus could be moved further away from the previous groove before the volume increased. Even so, bleed-across was often evident. That is, a loud passage followed by a soft one could be heard one revolution before it arrived because of slight compression of the previous groove by the stylus cutting the next adjacent groove. This was most commonly apparent in at the beginning of any disc the opening passage of which was quite loud and could be heard at a much lower volume one revolution before it arrived.

Monaural discs required information to be stored in a lateral cut moving back and forth parallel to the surface of the record. Stereo was made possible by cutting the groove in a manner which "encoded" the left and right channels in motion at right angles to each other but diagonal to the surface of the disc. Both the cutting coils of the cutting lathe and pickup coils of the stereo turntable or phonograph were arranged as shown in the figure at the left. The signals going into the cutting coils (and then picked up by the playback coils) corresponded to signals representing what would be heard by one's left and right ears. In principle, stylus motion in one direction would produce no motion in the other because the two were at right angles to each other. In practice there was always a little "cross-over" from one channel to the other. Stereo sound was at first achieved with two microphones, often placed close together and pointed at the performing group in a manner meant to approximate one's sensory experience. The signals were picked up, amplified and fed onto two tracks on a tape recorder. Soon that primitive technique evolved into multi-track master tape recordings with subsequent changes made in an editing room to produce more pleasant effects on each channel. In any case, a stereo disc ended up having to use both the lateral cut pioneered by the Victor Company and the vertical cut first introduced by Edison, but the high level of noise which plagued the early shellac discs had been largely eliminated by the smoother vinyl plastic of which the disc was made. Note that the width of the grooves seem to change a bit in the image at the right. There are wide and narrow regions. That is caused by the V-shaped cutting stylus moving up and down as well as back and forth. In particular, look at the place on the lower middle right where the groove seems almost to disappear. That particular point would be a prime candidate for "skipping", a common phenomenon with discs in general in which even the slightest jarring would cause the stylus to jump out of the groove and to fall into another nearby groove.

In 1983 digital recording technology became available for domestic use in the form of the Compact Disc (CD). Instead of recording the digital data on a magnetic medium, the stream of data is stored as tiny pulses on an aluminium disc coated in plastic. The data are then read using a laser

beam, thus eliminating physical wear on the discs. CD technology has settled on a 16-bit audio "byte."

The bits in the byte are in the form of indentations in the disc produced (initially at least) by laser pulses. Note the single commercially produced melted pulse in a disc surface at the left. When linked together to produce a series of audio "bytes" the surface of the disc looks like the image at the right.

Recently, equipment to produce one's own CDs has appeared on the market at prices affordable by the average PC owner. Several formats are available, one to produce audio CDs and another to produce rewritable CDs offering a format which appears no different in one's directory than that shown by a fixed drive or floppy diskette. It is claimed that the pits produced by home equipment are not quite as clean as professional masters. An example of such a claim is shown at the left.

All that having been said, here is a link to a Real Media file (converted from a digital CD) giving the first second or two of the opening bars of Mozart's "Eine Kleine Nachtmusik" (A little night music).

Click here to listen.

The sonogram or image of the vibrations of this music is shown at the right above. Be cognizant that this "image" of the sound in the opening passage is of the original uncompressed digital .WAV file created from a commercial CD of the piece, not the compressed .RM or Real Media file which you hear if you clicked on the link.

The first violin part is shown at the left to illustrate the coincidence between the size of the "blips" on the sonogram above and the note length in the music. The quarter notes are shown as large blips and the eighths as small ones. The total time for a quarter note and an eighth rest is three times that of an eighth note, as one would expect.

If a very tiny portion of this sonogram is copied and pasted to another .WAV file, it looks like the image below. Each of the traces represents one of the stereo channels. The image has been enlarged so that you can see the steps in the digital encoding. This small fragment is only 54 bytes long for each channel. The standard sample rate for commercial CDs is 44,100 samples per second per channel, so this fragment could be heard in a little over one one-thousandth of a second. Not nearly enough time to start it, go into the kitchen for a cup of coffee and expect to hear the rest of it on your return! Each byte is a fixed sampled value. The playback machine must take sequential values and do a digital-to-analog conversion so that an average smooth function results. Compressed audio files, such as MP3 and RM (Real Media) files, are the result of "shortcuts" effected by the compression software. The audio signal of an uncompressed WAV file is analyzed "on the fly" during the compression process and the barest amount of data describing a particular signal shape finds itself replacing the actual shape in the compressed file. An example of such a shortcut would be to approximate the rise one sees from the middle to the right side of the screen using two or three bytes and a code indicating the time required to effect the rise. By so doing 3-4 bytes would be required instead of the 26 or so one can count from the distinct dip to the right side of the screen. Such a shortcut would offer a compression ratio of 26:4 or about 6:1. Much higher rates of compression are possible if one is willing to make greater sacrifices in audio quality.

Thanks to all the people who put out an enormous amount of work in preparing Web pages from which much material was taken. The Web pages I used for this offering may be found at the following sites:

(History of the Edison Cylinder Phonograph)
(History of the Edison Disc Phonograph)
(Reproduction of 78 rpm records)
Ploeg's Recording History page)
(Some photomicrographs, one of which is of an LP record groove)
Real Media recordings of phonograph cylinders
Real Media recordings of vertically cut phonograph discs.