International photographer (Feb-Dec 1929)

Twenty The INTERNATIONAL PHOTOGRAPHER October, 1929 The Electrical Recording and Reproduction of Sound-Part 1 -BY ROBERT L. KAHN Within the last few years, the commercial methods of recording and reproducing sound have undergone some very profound changes. But like everything else, these commercial changes should not be considered as exactly mirroring simultaneously occurring changes in the theory and fundamentals of the art. It is just like the sea battering against a wall, apparently making but little change when suddenly at some stage the whole structure will go toppling. Almost as early as Edison had invented the phonograph, inventors bent their energies toward developing an electrical system for doing everything. But their minds outran their facilities. For one thing, electrical apparatus in the early part of the present century was crude, as measured by our modern requirements for accuracy, reliability, and sensitivity. An amplifier of electrical energy, like the present vacuum tube was unknown. Acoustical devices like transmitters, receivers, sound recorders and reproducers for wax records were not highly developed. The paramount object in those days was to get a good output of energy from the devices. Fidelity of translation had to take a secondary part. Now these binding and circumscribing restrictions have been displaced. The main object is fidelity of translation. Modern radio has educated the public to an appreciation of faithful translation of sound waves into electrical currents and back again to sound waves. The thing could be done well and had to be. The old methods of recording and reproducing sound were not equal to the demands placed on them. Like the noiseless and silent servants we read about in novels, electricity came in and demonstrated the faithfulness, flexibility, and ease with which these desired results could be obtained. The difficulties of mechanical recording on wax are well known. First of all, if an orchestra is to be taken, the grouping has to be very carefully arranged with respect to the horn. Irrespective of the actual volume, certain instruments would not record very well and would be submerged by others. There was a limit to the number of pieces. As a rule, a large symphony orchestra would be pared down to a fraction of their number, often as low as a third. Then they had to distort their rendition of the music. There was a limit to the loudness, for too much volume would record as blasting and might make the grooves so wide that adjacent ones would run in together, since the pitch of the spiral is a constant. Too soft playing would not record at all. Thus that form of distortion which tended toward monotony was a large defect. The methods of electrical recording have not entirely removed this but the range of volume has been greatly extended. The greatest source of evil, however, was the distortion of the various frequencies present in the sound to be recorded and reproduced. An authoritative survey of the totality of the art on Electrical Sound Recording and Reproductive Inventions as set forth by Mr. Robert L. Kahn, until recently examiner in these departments of the U. S. Patent Office, and as published in the Journal of the Patent Office Society, Washington, D. C. Reproduced by special permission of the Journal of the Patent Office Society, Sept., 1929.— Editor's Note. Distortion Due to Mechamcai Resonance To properly appreciate this type of distortion and understand how it has been overcome, it will be necessary to go into the mechanics of a vibrating body in a very cursory manner. Everybody undoubtedly knows that every mechanical element in any sort of a system has a tendency to vibrate at a certain frequency. The mass and elasticity of the body are the determining factors. This body may be able to vibrate at a different frequency if it is forced to do so, but the resonant frequency is the one it will choose if it can. This phenomenon may be considered from a different point of view. The resonant frequency may be considered as that one at which a device will be vibrated most efficiently. Let us assume a string has a resonant frequency of 1000 vibrations per second. Assume that the energy required to set the string into vibration at that frequency is unity. Then apply energy at a different frequency, say 1300 vibrations per second, and try to force the string to vibrate. It will be found that to get the same ampltitude of vibration may take as much as five or ten times the energy as at resonance. If a curve is plotted with frequency as one coordinate and energy as the other, it will be found that at the resonance period, the curve assumes a shape almost like a mound coming up from the ground. Thus it will be seen that the efficiency with which applied energy is transformed into vibratory energy varies tremendously near the resonant frequency. Now the distortion inherent in mechanical recording and reproduction under ordinary conditions may be understood. As the sound enters the horn, it encounters a device that has its own resonant frequency and this device picks out that frequency and transmits it very efficiently to the rest of the system, the remaining frequencies being transmitted at a much less efficiency. This means that most of the sound entering is reduced in volume because of the losses and that one frequency, the resonant one, is very little reduced in volume. The distortion due to that can be readily imagined. The sound then continues to the diaphragm. This has a resonant frequency and as a rule, it is much more marked than the horn. The curve for a diaphragm at the resonant frequency is much sharper than for an ordinary horn. The same thing happens here that happened when the sound was being trans* mitted through the horn. What complicates matters, is that the resonant frequencies of the horn and diaphragm are not often the same or near each other. That means that the sound is much more distorted and losses are greater. If the two elements of the system had identical resonant frequencies, it would not be so bad. It would be like two windows in line with a source of light. The second window would not cut off any more light than the first one had done. But if the second element has a different resonant frequency, then it is like moving the second window sideways. The first window has cut off a certain amount of light and the second one cuts off more. Then the transmission of the energy from the diaphragm to the recording stylus is like a third window. This element has its own resonant frequency. Now considering the system of windows as an example, there is going to be one band that goes through all the windows quite well, the rest of the light being cut off and diffusing through. By band here is meant a rectangle of light and refers to the area of illumination. In the same way, the entire recording system picks out one frequency which all the elements will transmit with considerable efficiency and records that very efficiently — that is with little diminution of volume. The remaining frequencies are transmitted with very great losses and are recorded with comparative weakness. This is only one-half of the process. The very same thing happens in reproduction. The net result is that one frequency is over-emphasized and the rest are recorded, if there is considerable energy in that band, otherwise they are lost. The ordinary record of some years ago was reproduced most efficiently at about 800 cycles per second. That takes the entire recording and reproducing into account. The remaining frequencies fared very badly. The high frequencies of several thousand cycles per second, the ones that lend naturalness to a tone and which distinguish a piano tone from a violin tone of the same pitch, were for the most part lost. That accounts for the difficulty which every one has experienced in attempting to determine what instrument was being reproduced. These grave imperfections in mechanical processes were well understood but little could be done. One well known method of avoiding distortion due to resonance is to make the resonant period outside of the range of frequencies in which it is desired to obtain a uniform efficiency of translation. Thus in voice frequencies, the range is usually from zero or two hundred up to two, five or ten thousand. That is the audible range and the greater the limits are taken as above stated, the more faithful will the translation be. The greatest amount of actual energy lies below the two thousand cycle mark, so everything above that is in the nature of a refinement. By making (Continued on Page I 3 i