Sound and Its Reproduction.
The mechanical, physical and psychological factors entering into the perception of sound and its reproduction are complex and varied, and the simultaneous satisfaction of these, constitute, it is thought, one of the most remarkable achievements of science and engineering.
This paper discusses, in general principles, the nature and perception of sound, physical and psychological problems encountered in its reproduction, and a visual and aural demonstration was given of the effects of various distortions, phase differences and other phenomena associated with the subject.
The nervous system of the body is such that nerve termini are exposed so as to be excited by physical disturbances surrounding the body. This commotion is transmitted to the central organism producing commotion there, and we experience what we call a “sensation.” The physical disturbance producing a sensation is known as a “stimulus”.
The disturbance to which any particular set of sensory nerves is especially sensitive is called the “proper stimulus.”
The nerves of hearing terminate in end organs which float in a fluid contained in a bone-walled cavity called the “inner ear”. The inner ear is provided with two windows, the “oval window” and the “round window” which are closed by thin membranes. The “middle ear” is an air-filled cavity communicating with the mouth cavity by the eustachian tube, and having three small bones, the hammer, the anvil, and the stirrup, forming a chain bridge between the oval window and the “eardrum”. The eardrum vibrates by excitation from the vibratory motion of the external air transmitted down the “ear canal”. The motion of the eardrum is transmitted by the chain of bones to the oval window causing the inner ear fluid to surge back and forth through the complicated channels of the inner ear between the oval and the round windows. This surging fluid excites the end organs of the nerves of hearing.
The sensation from this excitation is called “sound”. The physical disturbance which constitutes the proper stimulus of the nerves of hearing is also called “sound”. The study of sound, the sensation, belongs in the science of psychology. The study of sound, the proper stimulus, is in the science of physics. The study of the reception of the proper stimulus and its transmission to sensation is in the specialised medical field of otology.
The recording and reproduction of sound is in effect the interposition of a further dimension, that of space or time, in the proper stimulus. The study of this field is the science of sound engineering, with its subsidiary subjects of acoustics, electronics and electrodynamics.
The field which we have to cover in “Sound and Its Reproduction” is therefore very wide, and in the time at our disposal it will be possible to treat only briefly of each phase. This must serve also as an excuse to those who may consider some portions of the treatment sketchy, as they are, and in many cases lacking in rigour.
Sound in General.
Ordinarily sound consists of wave disturbances in the air caused by a vibrating body. These waves spread out in all directions from the sounding body very much as water waves spread out from the point where a stone is thrown into a pond of water. The wave velocity is a function of the temperature and humidity of the air and recent determinations give the velocity as 331 meters per sec. in dry air at 0° C., that is 1080 feet per second, or 736 miles per hour.
Sound disturbances may be of two types, wave pulses and wave trains. A wave pulse consists of an isolated wave group, commonly called a transient. A wave train consists of a recurrent disturbance repeated in equal intervals of time, being termed in general, periodic. If the disturbance in addition to this latter quality also follows a definite law, the sine law, the disturbance is harmonic.
Sound for our purposes is termed noise, notes and tones, which correspond to the types of disturbance given above. Noise is, strictly, sound which is neither periodic nor harmonic. In common usage, however, it is more nearly defined as unwanted sound.
The distinction between the technical and the effective or personal definitions must be kept in mind. One can see here very clearly the confusion between sound the stimulus and sound the sensation, that is between sound the physical and sound the psychological. Technical noise, that is the bang-crash sounds, may be said to be objectionable to everybody, excluding the small boy, therefore when we meet with any sound (even in the note-tone category) which we consider an objectionable sensation, we call it by a psychological transfer, noise.
Notes are sounds which are periodic but not harmonic. They constitute by far the greater proportion of sounds in general experience, in fact one is hardly doing violence to truth to say that all the sounds anyone, except the specialist worker, hears, are notes.
Tones are pure sounds; being harmonic their form is definite and invariable.
They consist of waves of density variations about the normal unstressed density of the medium of propagation. They have, strictly speaking, only two dimensions, amplitude and frequency. The subsidiary dimension of wavelength is a variable equal to the velocity of propagation divided by the frequency. The velocity of propagation varies widely in different media, from 1120 ft./sec. in air at 15° C. to 4700 ft./sec. in water and 16,500 ft./sec. in steel.
In the recording and reproduction of sound these dimensions are subject to transfer from one form of energy to another, and it is the control of this transfer which is the function of sound engineering.
Let us trace this transfer in the general case. Sound exists originally as an air disturbance. In recording, this disturbance is picked up by a microphone whose function is the conversion of acoustic energy to the mechanical energy of a diaphragm, and from this to electrical energy. This electrical energy is transferred from one electrical circuit to another, being amplified from stage to stage until it is of the required power. At this point enters the presence or absence of the time factor. If the recorded sound is immediately reproduced by the conversion of the electrical energy into magnetic energy, from magnetic energy into mechanical energy, and mechanical energy into acoustic energy, that is, back into its original form, there is no time delay. The original sound has been merely increased in power or transmitted to a distance, and the two performances, original and reproduced, occur, for our purpose, simultaneously.
If, instead of immediate reproduction, the electrical energy is converted to, say, either of the two common recording forms, sound on film or sound on disc, we have a conversion from electrical energy to mechanical energy, mechanical to light, light to chemical, chemical conversion, and the product of a more or less permanent record on film. In the case of sound on disc, the mechanical energy is used to act upon a mass so as to change its shape.
The most important aspect here is that the sound is recorded in a form in which it is practically independent of time. As with a photograph or a motion picture we can reproduce at will the performance of a past event. The factors affecting the fidelity of this reproduced event are, in particular, our province.
I have purposely dwelt at some little length on the energy conversions involved because I consider that in the end result they constitute one of the greatest achievements of modern times. An achievement which is, in fact, obscured only by that facile conceit with which the average spectator views common experience.
When it is realised that the original sound must go through all these energy transformations both of kind and magnitude, in the course of which it may be subject to successive amplifications of a total of a million to a billion times, the wonder is not that there is anything wrong with the reproduction but that there is anything right with it.
We have already seen that the principal characteristics of sound are amplitude and frequency. The simple sounds or tones have only one frequency, the complex sounds, notes and transients, consist of combinations of frequencies. The lowest frequency in a note is called the fundamental and additional frequencies are “overtones” or harmonics. In a musical note there are always multiples of the fundamental, such that we have the fundamental or first harmonic, double frequency or second harmonic, third frequency or third harmonic, and so on. Since in a note we may have any combination of harmonics in any amplitude it is obvious that the range of possible notes is practically limitless.
Transients, in which is included noise and some notes, usually contain a very wide range of frequencies. In the case of noise the frequencies are not harmonically related with the result that we get a jumble of tones and notes which usually proves very irritating. The principal cause for this is what is known as cross modulation. If we sound two tones together, say 100 cycles and 150 cycles, these tones will add to form a tone of 250 cycles, and will subtract to form a tone of 50 cycles. We will thus have a note composed of a first harmonic of 50 cycles, a second harmonic of 100 cycles, a third harmonic of 150 cycles, and a fifth harmonic of 250 cycles. Depending upon the relative intensities of these tones, the ear will hear a note which will be precisely the same in character as if it had been originally composed of these tones. Thus the ear can hear a note which does not exist in itself, and it is for this reason that much reproduction which is actually poor, appears to sound better than it is. These tones are called combination tones and exist in all cases where two or more fundamental tones exist in the same medium. When two notes, or more, exist together with their associated harmonics the combination of tones becomes most complex. In music the scales and instruments are so designed that the combination tones are always in harmonic relationship to the fundamental tones. When this harmony is departed from we get the effect known as dissonance.
In reproduction, when any portion of the reproducer is non-linear in character it produces harmonic distortion. This means that frequencies which were not present in the original are being produced in the reproduction. It is often the case that these added frequencies combine to produce tones which are not harmonically related to the tones and notes in the original, with the result that dissonance is produced and we get the familiar effect of the cheap or improperly handled radio or reproducer.
Precerption of Sound.
The ear recognises three qualities in sound—
Loudness or Intensity.
Timbre or Quality.
The ear is generally insensitive to pitch unless it has a comparison tone, when it becomes extremely critical. Thus a variation in the speed of a re-
producer is quickly detected, since the tones themselves form the comparison tone.
The range of pitch heard by the ear is given as 16 to 22,000 cycles by the Bell Telephone Laboratories, but varies widely between individuals. The loss of higher frequencies becomes quite serious with advancing age. Musical tones go to around 5000 cycles from a low bass of 60 cycles. The grand organ has a low C of 16 cycles. A high soprano is about 1300 cycles and over and above these tones, overtones or harmonics occur up to 15,000 cycles. For perfect reproduction the total range required is from a minimum of 30 cycles to an upper limit of 15,000 cycles.
Transients require in general the greatest range. The percussion instruments produce transients and are therefore the most difficult to reproduce. The piano is rather peculiar in this respect. The lowest note on the piano is 27 cycles, but it produces no fundamental lower than about 100 cycles. The lower notes are produced by the combination of rich harmonic tones, which give difference tones simulating very low frequencies. This is probably the reason why the piano usually sounds well even under adverse conditions of reproduction. Since the tones most difficult of reproduction are not produced fundamentally they do not have to be transmitted and require the transmission only of the harmonics producing them.
Loudness or intensity is a measure of the energy contained in the sound. At the most sensitive portion of the spectrum the ear can accommodate a range of intensity greater than a million to one. At the lowest intensity or threshold of hearing we have the flutter of leaves corresponding to a sound pressure of 0.000204 dynes per square centimeter, or 16−16 watts per square cm. at the eardrum, while very intense sounds may reach pressures of 100 or 200 dynes per square centimeter or 10−4 watts per square cm. at the eardrum. This wide range of powers is handled analytically by means of the unit called the decibel. One decibel is about the smallest change in sound pressure which the ear can detect, and it should be noted the ear does not hear linearly but logarithmically, obeying the Weber-Fechner Law. As a result equal increases in loudness are not equal increases in intensity but equal ratios of intensity. Thus 6 decibels means a doubling in intensity, and it does not matter whether this is an increase from one-thousandth of a dyne per square cm. to two thousandths of a dyne per square cm. or from 100 dynes per square cm. to 200 dynes per square cm., the ear will interpret the two changes as equal.
The dyne per square centimeter is called the “bar” in the U.S.A., and the “microbar” in English meteorology.
The range of hearing is about 132 decibels, or an intensity ratio of 4 × 106, and above the maximum intensity the ear passes into the threshold of feeling where sounds are felt rather than heard, and if the intensity is further increased the sensation passes into the region of pain. The sound pressure at this point is about 600 dynes per square cm. or 1 milliwatt at the eardrum. The magnitude of these powers can hardly be comprehended. Conversational speech results in a power at the cardrum of about 50 millionths of a watt.
The range of perceived intensity narrows as the limits of frequency perception are approached, and of course becomes zero at the limits. This results in a complex effect in reproduction called volume distortion. If we assume a certain range of frequencies being reproduced at a certain volume the ear has a certain frequency range. If now the volume is reduced the frequency range of the ear becomes reduced and frequencies which are present in the original are no longer heard, giving in effect a restricted frequency reproduction. The reverse effect occurs if the volume is raised, when new frequencies which were not audible in the original become heard. The effect becomes noticeable on a radio when the volume is turned down. The reproduction loses its body and becomes thin because the low frequencies fall below the threshold of hearing.
Timbre or quality in a sound is the characteristic which distinguishes one note from another. It is produced, of course, by the combination of fundamental and harmonic tones both in number and in magnitude, which accompany any particular note. For reproduction it is necessary to reproduce these without change and without subtraction or addition, that is, the identity of the note must be preserved. This requirement places the greatest strain on the reproducer, and is the requirement which is very seldom met adequately. The timbre of the reproduction is affected by any type of distortion, often in a complex manner due to a combination of effects.
The distortions which affect reproduction are, generally speaking, four: (i) Amplitude or volume distortion, (ii) Frequency distortion, (iii) Harmonic or non-linear distortion, (iv) Phase distortion. Other effects may be present which mar the reproduction and are in effect distortions, but there are usually types or combinations of the above classes.
Amplitude or volume distortion is defined as a lack of constancy of the ratio between the R.M.S. values of the stimulus and the response at different amplitudes of the stimulus.
In general terms the volume level of the original and that of the reproduction do not preserve a linear relationship. This type of distortion has already been mentioned, and it occurs when the reproduction is made at a volume level which is incorrect for the performance. The distortion may occur during recording, due to limitation in the recording characteristic or in failure to secure the proper tonal balance, but, since the recording process is practically always under the controlled of skilled personnel and very great pains are taken to secure the correct recording, it is not often that the recording is at fault.
Volume distortion more usually occurs during the reproduction when the acoustic power is either greater or, usually, less than that which would apply in the presentation of the actual performance. It is not our place here to enter into a discussion upon this aspect. The subject is allied to questions of acoustics, reverberation and relative sound levels, but the simple fact remains that there is a correct level for any reproduction, and amongst competent observers you will find general agreement on this. The mechanism of this type of distortion has already been explained, and it will be seen that it is really frequency distortion in the acoustic field, due to the varying response of the ear.
Another type of volume distortion is inherent in all known recording systems. Owing to the very wide range of sound intensities, which we have already seen cover a range of 4 × 106 to 1, it is not practicable to record the full range of volume which the ear can appreciate. The restriction is principally mechanical as it is almost impossible to get mechanical systems which will respond to or accept operating ranges of this magnitude. In the case of disc records the range is reduced by the necessity for avoiding overcutting of the grooves. The conflict here is one between playing time, frequency response and volume range. In order to record high frequencies the recording groove must pass the stylus at a rate which, in general terms, makes the size of the stylus a small fraction of the wavelength being recorded. As the frequency is increased and the wavelength decreased the ability to record frequency consequently falls off and a loss of frequency, or frequency distortion, results.
Given an upper desirable limit of response the length of groove is then fixed, and to get this on a reasonably sized disc a certain number of grooves per inch must be used. This limits the amplitude of cut which can be used without cutting into the next groove. The result is that the volume range is restricted and we get volume distortion. Frequency distortion also results because the low frequencies carry practically all the energy and have a correspondingly great amplitude. This full amplitude cannot be recorded, and the intensity of these notes is therefore reduced.
The restriction in recorded volume range is usually done manually, the maximum level being maintained just below the overload point of the system. Volume expanders, which operate to increase the volume from loud passages and reduce the volume from weaker passages automatically, have been used for a number of years, but they suffer from serious disadvantages in practice, which has greatly restricted their use.
The principal of these disadvantages is that the initial volume compression is arbitrary in both magnitude and speed, and may vary considerably from record to record. Consequently it is practically impossible to reintroduce the exact complimentary expansion. Excellent results can be achieved in the case of some records, but adjustments have to be made for each record, and they are not simple to make, generally speaking, with the result that the effect is more trouble than it is worth, and it has never become popular.
The volume range of recorded music has been very greatly extended in recent years, particularly in the case of sound on film, and a range of 50–60db is achieved.
Suggestions for automatic control of levels by the use of supersonic pilot frequencies, etc., have been made, and are quite practicable, but suffer from the disadvantages of complication and lack of standard equipment.
On theatre equipment some automatic control is commonly used, and works very well.
Frequency distortion is defined as a lack of constancy in the ration between R.M.S. values of stimulus and response, at different frequencies of the stimulus.
For our purpose here we will consider only the distortion occurring in the reproducing process.
It is usual to take the mid band as the norm, commonly 1,000 cycles, and to express the frequency characteristic in so many decibels plus or minus 1,000 cycles.
The characteristics may fall at any point or over any band. Commonly it falls at the high and/or the low frequencies, becoming rapidly worse as the extreme of the range is approached. In high quality reproduction it is common to require a characteristic of ± 1 decibel over a range of 50 to 10,000 cycles, and for “perfect” quality this range requires to be from 30 to 15,000 cycles. Plus or minus one decibel means that the amplitude must be accurate to within 10%.
The appreciation of frequency range in quality is generally fairly widely deficient in people due to lack of training in musical appreciation, defective hearing (i.e., frequency distortion in the ear itself), and psychological adjustments. People in general hear what they want to hear. They hear a band—not the music a band is playing. This characteristic is becoming more and more pronounced with the common use of “background music” and “radio music” until it may be fairly said that most radios and gramophones are musical instruments and not reproducers. There is a point I would like to make here. A reproducer never has “a nice tone,” whatever that term may mean. A musical instrument, a piano, a violin, has tone, a reproducer has quality. If, of course, you wish to run your gramophone or your radio as a musical instrument and hack the original recorded quality into something that you like, you are, in a free country, perfectly at liberty to do so, but you are not entitled to say that it is good quality or good reproduction, only that you like it that way.
Frequency distortion is closely allied with an indefinite quality in reproduction which is associated with the overall distribution of the frequency response. It is a balance between high and low frequencies such that a reproducer with a limited range in both high and low register will show better quality than if it lacks only the high or the low register. The proper taper in response, where this is limited, is most important.
In gramophone reproduction the range of frequency response is limited by a number of causes. Three types of recording are used, constant amplitude, constant velocity, and constant acceleration. In constant amplitude recording the amplitude of cut is restricted, as we have seen, to a certain maximum value determined by the number of grooves per inch to avoid cutting through to the next groove. Because the energy in the frequencies is proportional to the square of the amplitude, this results in a droop in the low-end response. Over the range 300–3000 cycles the recording is straight line or constant velocity, which produces a flat response curve. Above 3000 cycles, on account of the difficulties of recording high frequencies, the recording becomes constant acceleration and a droop in the high-end response results.
The requirements of disc recording are such that no improvement in present recording can be expected unless some radically new method is developed. We have been dealing, of course, solely with lateral recording, in which the recording is done across the groove.
Other methods of recording, such as hill-and-dale, in which the recording is done on the depth of the groove, and film recording, are not subject to many of the limitations of lateral recording, but they are not, on account of existing equipment, likely to come into general use. Magnetic recording, in which the record is obtained by the magnetisation of a steel tape or wire has received considerable research in recent years, and the improvements in it are such as to indicate that it may well become a serious competitor to disc recording. However, we are not concerned here with recording, but reproduction, and the only excuse for introducing the subject at all is because of the limitations the record places upon the reproduction.
Harmonic or non-linear distortion is defined as a lack of constancy of the ratio between the instantaneous values of the response of the system and the corresponding values of the stimulus.
This means in general that harmonic distortion occurs when any frequency not present in the original is introduced in the reproduction.
This type of distortion, as has already been noted, is responsible for considerable trouble, due to its facility in producing combination tones. It is practically always due to non-linear elements in the reproducing chain. An amplification can be considered as a straight-line graph having a slope equal to the amplification. If the graph line remains straight but the slope changes it is obvious that a linear relationship will still be maintained between input and output, but the absolute ratio of the two will change with the slope. If this slope varies with frequency, we have frequency distortion. Consider now the case where the line is not straight but is curved so that the slope varies along the length of the line. In this case the output will not be proportional to the input at all points, but will vary from point to point, with the result that the output is not a replica of the input, and this means that additional frequencies are being introduced into the output.
This effect may be due to mechanical resonances, frequently to magnetic materials, which in practice have to be worked at extremely low fluxes, and curvature in the characteristics of vacuum tubes. In fact, in terms which most of you will readily appreciate, the requirement corresponds to Hookes Law in mechanics. When the Hookes Law of recording or reproduction is departed from we get harmonic distortion.
The effects produced by harmonic distortion are fuzziness, rattles, rawness of tone and dissonances. The effects may be so complicated that no simple results can be predicted. Frequently it is possible to have considerable harmonic distortion without being able to detect it.
The real culprit in harmonic distortion is cross modulation where frequencies are produced by the interaction of other frequencies, and the resulting frequencies are dissonant. The even harmonics are usually very difficult to detect, and in general are quite pleasing in tone. Odd harmonics, on the other hand, nearly always give trouble. The higher harmonics are the most important, due to the very great sensitivity of the ear to these frequencies. For instance, at a frequency of middle C (256 cycles), at a pressure of 2 dynes/CM2, the tenth to twentieth harmonics have to have only one ten-thousandth of the pressure of the fundamental to be quite easily audible. The minimum audible increase or decrease in fundamental is about 1000 times as great as this. Rattles and such noises are frequently composed of trains of high-order harmonics of small amplitude and under conditions of harmonic distortion are frequently simulated, so that a loudspeaker may appear to have a bad rattle and yet, on single frequency test, show no signs of the effect. Lower-order harmonics, on the other hand, may require up to one-tenth of the fundamental pressure to be noticeable. A striking demonstration of this effect of beat frequencies will be given later.
Phase distortion is defined as a lack of constancy in the time relationship between the instantaneous values of the stimulus and those of the system response.
Phase distortion is of small moment in general. There are two types of phase distortion. A note as we have seen is composed of a fundamental and harmonics. These harmonics have a specific time relationship to the fundamental in the original sound. It does not matter, however, whether this relationship is maintained or not in reproduction, as the ear apparently takes account only of the number and magnitude of the harmonics. In other words, provided we have the right ingredients in the right quantities, how they are mixed does not matter. This is very fortunate, because maintaining phase relationship is a very difficult business. This effect, therefore, is easily disposed of.
The second kind of phase distortion is due to a difference in the speed of propagation of different frequencies over a transmission circuit. The result is that, where the time of transmission is relatively long, a delay may occur so that some frequencies are delayed appreciably behind others. This effect is not met with in the reproduction with which we are concerned here, but its effect is interesting in showing how sensitive the ear is to this type of distortion. A demonstration later will show that the ear can easily detect phase distortions of thirty-thousandths of a second.
We have now covered in a general manner the distortions which may occur internally in the system. The final link in the chain of reproduction is the loudspeaker. A chain is no stronger than its weakest link, and the loudspeaker is the weakest link in the chain at present. It is not difficult to record and reproduce perfect quality up to the loudspeaker, by making use of
the best methods. I am not now speaking of commercial methods, but of the best available techniques applied to the specific problem of perfect reproduction. When the loudspeaker is reached, however, we come up against the problem of a mechanical translator and associated problems of acoustics. It is possible to treat of these only very briefly. The production of low frequencies in sufficient energy is one problem. This requires large diaphragm areas, and with large areas we get interference effects at higher frequencies which cause high losses. To avoid interference between the front and back radiated waves from the speaker, it is necessary to separate these by means of baffles.
These baffles require care in their design if undesirable effects are to be avoided.
Acoustic labyrinths of various kinds are also used, but require careful design to avoid excessive back loading on the speaker cone and resonance from enclosed volumes. They can, however, be constructed to smaller dimensions than plain baffles, which require to have a minimum distance from the front of the cone to the back of the cone of at least half of the wavelength of the lowest frequency to be reproduced. This means that the minimum distance from front to back of a speaker must be at least 6ft. to reproduce 100 cycles, and 12ft. to reproduce 50 cycles. In other words, the speaker requires an enclosure of the order of say 6ft high by 12ft. long to reproduce 50 cycles, or half these figures for 100 cycles. The response of the usual radio set can be gauged from these figures.
The high-frequency response suffers from the area of the diaphragm, and interference effects are produced, due to each point of the cone area acting as a separate source. The resulting wave fronts combine to produce areas of high and low response, and at very high frequencies the response tends to narrow into a beam-like zone, and the response outside this beam is practically zero.
The acoustics of the room are also a matter of particular importance, and in the case of large enclosures, such as halls and theatres, become a problem. This aspect of the subject does not have a very great influence, however, in the usual home application, where rooms are comparatively small and well furnished. The difficulties increase with volume, which of course goes up very fast with an increase in dimensions. The placing of a speaker in an ordinary room may, however, have a decided influence on the reproduction by reflection from the walls.
It is not my intention here to demonstrate to you high-fidelity reproduction. To do that would require surroundings much more acoustically acceptable than these, and also the use of equipment and recordings, the installation and adjustment of which would occupy much more time and effort than the use would warrant.
In conclusion, I would like to express my appreciation of the generous assistance which has been given me by Mr. P. D. Mason and Mr. P. E. McIver, of the Western Electric Co. (N.Z.), Ltd., and by Mr Martin Kimble, who has generously donated time and suggestions in the arrangement and presentation of the demonstrations.
Harvey Fletcher. Speech and Hearing.
Olson and Massa. Applied Acoustics.
Webb and Wilson. Modern Gramophones and Electrical Reproducers.
Symposium on Wire Transmission of Symphonic Music and its Reproduction in Auditory Perspective: Bell System Technical Journal, 1934, vol. 13, p. 239.
The Appraisement of Loudspeakers: G.E.C. Journal, Nov., 1936, and May, 1937.
Grandall. Theory of Vibrating Systems and Sound.
Wheeler and Whitman. Acoustic Testing of High Fidelity Receivers: Proc. I.R.E., 1935, vol. 23, p. 610.
Motion Picture Sound Engineering. Academy of Motion Picture Arts and Sciences.
Reference must also be made to the large number of papers in various issues of the Journal of the Society of Motion Picture Engineers and the Bell System Technical Journal.
Mr. Gemmel: Under what circumstances is phase distortion important?
Mr. Brown: In broadcast reproduction when transmitted over long lines. There is also a phase shift in the amplifiers. In U.S.A., where very long lines are used, the delay of lower frequencies is quite noticeable. This is also very important in the field of television, where phase distortion may produce overlapped
images. It can also occur in chain broadcasting. Small phase distortions, although difficult to measure, are detectable by the ear. In communications over wires, one tries to compensate for phase distortion.
Mr. McLennan: Is phase shift similar to echo?
Mr. Brown: Phase delay and time delay are not synonymous, although the terms are often used indiscriminately. Phase delay is delay expressed in terms of the angle, while time delay is expressed in terms of time. In general, the two are equivalent only for one particular frequency. An echo is an extremely long time or phase delay.
Mr. Mudgway: Why are lower frequencies not more attenuated if the ear is more insensitive to them, as in the case of hearing band music at a distance?
Mr. Brown: There is more dissipation of energy in the higher frequencies due to reflections from obstacles. One may use the hydraulic analogy of water waves: waves of long wave-length pass easily around an obstacle, while wave ripples of short wave-length are held back by obstacles and are reflected.
Mr. Wood: Are all three methods mentioned in the paper used for cutting records?
Mr. Brown: All three methods are used, or rather the effect is as if they were used. Constant amplitude, e.g., is watched by the supervising technician. Constant velocity is the normal recording characteristic. Constant acceleration characteristic is produced primarily by the finite speed of the recording disc.
Mr. Hale: How does the signal voltage of magnetic tape pick-ups compare with that of ordinary electro-magnetic pick-ups? Have amplifiers to be more sensitive?
Mr. Brown: I have not measured the output voltages of magnetic pick-ups. The voltage would depend on the type of magnetic tape used. One the whole, the output would be higher than that of the electro-magnetic type.
Mr. Bogle: Can we take it that the quality of reproduction in private homes will never be true because the sound level cannot be high enough?
Mr. Brown: This is a debatable point and depends on the field strength and type of enclosure. Whereas the peak power of a full orchestra is 70 watts, in a home the maximum required is only 8 watts. It is a matter of producing the same or similar pressures.
Mr. Gray: How is it that the tone from a particular record is not distorted as it is with others, when the volume is turned on to a very high level?
Mr. Brown: There does not appear to be an explanation for this. Harmonic distortion depends on amplifier response, ear response and quality of record. The quality of modern records may be generally classed as follows: 80 per cent. are very good; 15 per cent. are quite good; 5 per cent. may be faulty. Therefore, on an average, the reproduction should be good.
Mr. Burton: In the reproduction of the talking record, it was noted that the bass was much more prominent than is your speaking voice. How do you account for this?
Mr. Brown: This is due to volume distortion and will depend on the level of reproduction, which was higher than ordinary direct speech level. In recording talking records, the speaker is usually close up to the microphone and there is a tendency for the lower frequencies to be more pronounced than in reality.