US2458641A - Method and apparatus for controlling frequency characteristics in sound recording - Google Patents

Description

Jam; 1949. M. RETHNGER 51' L 2,458,641

A METHOD AND APPARATUS FOR' CONTROLLING FREQUENCY CHARACTERISTICS IN SOUND RECORDING Filed May 8, 1945 2 Sheets-Sheet 1 FeeaufiA/cr 200 A700 2000 LOW p953 RESPONSE IN DB HMPL lF/EB )glvelyToR. pg 1 v BY {mi inqa. I QQ/flW /ATTORNEY.

Jan. 11,1949. Y'M. RETTINGERY ETAL METHOD AND APPARATUS FOR CONTROLLING FREQUENCY D RECORDING Filed May 8, 1945 CHARACTERISTICS IN SOUN 2 Sheets-Sheet 2 I000 FEEQUENC I 500 FREQUENCY INVENTOR. P

{was W46? ATTORNEY.

Patented Jan. 11, 1949 METHOD AND APPARATUS FOR CONTROL- LING FREQUENCY CHARACTERISTICS IN SOUND RECORDING Michael Rettinger, Encino, and Kurt Singer,

North Hollywood, Calif., assignors to Radio Corporation of America, a corporation of Delaware Application May 8, 1945, Serial No. 592,618

9 Claims.

This invention relates to the detection, transmission, and. recording of signals, such as dialogue, music, and sound effects, used in the production of sound motion pictures, and particularly to a method of and means for obtaining a predetermined frequency response characteristic during transmission to produce a sound record which will provide the most natural and pleasing sound in the usual places of reproduction, such as theaters.

It is well known in the analysis of high fidelity sound recording that a large number of factors must be considered as important in determining the quality of the reproduced sound. Not only must the electrical characteristics of the various elements of a recording system be taken into consideration, but the physiology and psychology of speech and hearing must also be considered, as well as a number of acoustic and electroacoustic factors.

The present invention is directed to a sound recording channel in which provision has been made for obtaining certain changes in the frequency response characteristic at certain points in the channel to provide a record or sound track which will have the desired characteristics when reproduced. For instance, to mention specifically a few of the elements which introduce an undesirable variation or distortion into the frequency response characteristic, we can consider the high frequency loss due to the recorder slit used in the commercial type of studio recorders. Then there are also losses introduced into the high frequency range due to film processing and reproducer slit attenuation. From the physiological and psychological standpoints, we have the unnaturalness occurring due to the use of a reproducing level, which, on the average, is approximately 5 db. above thenormal speech level, such a high level being necessary because of audience and other noises in the theater. In view of this fact, there is a resulting low frequency accentuation caused by the hearing characteristic of the humanear, the ear sensitivity not only varying with frequency, but its sensitivity at different frequencies varying as a function of the intensity of the signal. That is, the higher the intensity of the signal, the more sensitive the ear becomes to low frequencies.

Frequency characteristic variations are also caused by a speaker lowering and raising his voice below or above the normal speaking level. When a person speaks in a low tone because he is in abnormally quiet surroundings, such as an actor in a studio, the low and high frequencies become relatively more prominent as compared with a voicecharacteristic of normal intensity. If this low voice were reproduced at normal speech level, it would tend to sound unnatural by being heavy and sibilant. Another cause of accentuation of the low frequencies and the loss of high frequencies is the reverberation characteristic of sound stages. It is well known that practically all acoustic materials are less absorbent at the low frequencies and become increasingly absorbent at the high frequencies. Thus, the reverberation time for the low frequencies is usually longer than for the extreme high tones.

Compression is used in sound recording for several reasons, the chief one being the contraction of a wide range of sound amplitudes into a limited sound track area. This, of course, can be accomplished to a certain degree by an operator, but it is preferred to'do it automatically. It has been found, however, that a predetermined frequency characteristic variation must be used in the compressor unit itself to avoid frequency characteristic changes caused by compression. An equalizer invention for solving this compres- 7 sor problem is disclosed and claimed in Miller Patent No. 2,312,260 of February 23, 1943.

In sound recording for motion pictures, a microphone is the usual element for detecting sound at its original source, various types of microphones being used in accordance with the type of sound source to be recorded. For instance, a pressure, a velocity, or a uni-directional type of microphone is usually employed, the latter being a combination of the pressure and velocity types. Since such microphones vary in their frequency response characteristics, it is necessary to introduce equalization to compensate for this variation. The variations of frequency response characteristics of microphones are usually compensated for in the pre-amplifier to which the microphone is directly connected. This generally requires a different preamplifier for each type of microphone, although microphone equalization could be placed at a later point in the channel. However, if placed after the pro-amplifier, microphones of different types could not be used simultaneously. In the present invention, a portion of the equalization still takes place in the pre-amplifier, while another portion is taken care of at a later point in the channel. This arrangement has several advantages, as will be pointed out hereinafter.

Location of the entire low frequency equalization in the mixing console is likewise an established outgrowth of an early, expedient procedure. It was at one time considered generally appropriate, leading to consistently satisfactory results, to allow the mixer on the set to control the general character, and to some extent also the intelligibility, of the voices to be recorded. With the advent of the compressor, however, some of the functions of the mixer. along this line became curtailed. Moreover, by locating all dialogue equalization in the mixing console, which is always ahead of the usual. compressor. in the circuit, a certain accenti'iation of the low frequencies occurs with increasing voice-level. This accentuation is due to the compressor action, which tends to flatten. out"the si'gna1 char-- acteristic by acting only on those. signal com ponents which are above its threshold. If the low frequency components are severely attenuated ahead of the compressor, so that they lie below the compressor threshold, and hence, must pass through the instrument uncompressed, the resulting" output from the compressor will show anincreased amount or low frequency response. This change in: Voice quality is particularly noticea'ble when an actor delivers a declarn'atory speech. Acoustical studies have proved that decl'amatory speech contains fewer low frequency components than normal: speech, just as intimate, speech contains a preponderance of. low notes.v Compressor action, however, tends-to wipe out? or to eliminate this lack: of lower register during. declamation making: the" voice sound less highpitched;

From the" above; it is realized that if the reproduced sound: isto appear normal during reproduction; the: frequency' characteristic of the sound as recorded should notcorrespond with the frequency characteristic of the original-source of'sound.

The present invention is directed to an electrical transmission circuit which modifies the original frequency characteristic of a signal ina particular manner to provide optimum re-- sults, as explainedhereinafter. Thus, the main features of this invention are the allocation and type of equalization in a sound recording channel.

The principal object of the invention, tliere fare; isto facilitate the recordin of. a sound recurs,

Another object of" theinvention is to provide an improved method of signal detection, transmi-ssi'o'n, and recording.

A further object of the inventionis to provide an improved system of transmitting sound waves.

A further object of the invention is to modify the frequency characteristic of a signal being recorded to obtain improved reproduction thereof.

A still further object of the invention is to provide an improved method of and means for obtaining low frequency and high frequency equalization in a sound recording system.

Although the novel features which are believed to be characteristic of this invention will be pointed out with particularity in the appended claims, the manner of its organization and the mode of its operation will be better understood by reierringto the following description read in conjunction with the accompanying drawings, forming a part hereof, in which:

Fig. 1 is a block diagram showing the elements or units of a sound recording channelembodying the invention together with the respective frequency response characteristics of the units.

Fig. 2 is a graph showing the variations between the characteristics of loud, soft, and normal voices, and

Fig. 3 is a graph illustrating the effect of the invention upon the frequency response characteristic between and 500 cycles.

Referring, now to the drawings, a microphone '51 is a high. quality uni-directional transducer having the frequency response characteristic shown opposite by the solid line. The microphone feeds into a pie-amplifier 6 and then into a mixing consolev '1' including a low frequency variable equalizer before being impressed upon a compressor 8. Following the compressor is a fixed equalizer Hi. a low-pass filter H, and a high-pass; filter l2, after which the signal is amplified in a main amplifier l4 and impressed upon a: standardfilm recorder l5.

As forv the microphone 5, opposite each of the other units is shown itsrespective frequency response characteristic. It will be noted that the pre-amplifi'er characteristic, indicated by the solid line; shows that no equalization is effective up to approximately 1000 cycles, while equalization is applied in the high frequency range to compensate for the droop in the solid line microphone characteristic. The dotted line characteristic of the pie-amplifier shows the equalization necessary to compensate for the dotted linecharacteristic of a microphone. In the past, allhigh.

frequency equalization was applied at this point in the channel to compensate for film and slit losses as wellas high frequency microphone characteristic compensation.

The next element in the. circuit is the mixer con sole 1 which includes a variable low frequency equalizer used to compensate for abnormol signal frequency variations, such as boomy voices, or voices recorded in a boomyset. This equalizer is variable and may be adjusted by the operator as the signal and set conditions vary.

The next element is compressor 8 for contracting a predetermined signal range into the limited range of a sound track area. However, as set forth in the above mentioned Miller patent, since the spectral energy distribution of words and other sounds varies in accordance with the character of the voiced and unvoiced components thereof, it is necessary, in order to make the degree of control of the compressor amplifier independent of the spectral location of these components, to make the products of rectification provide a control voltage whose magnitude varies with the frequency of the predominant components of the signal being transmitted through the variable gain amplifier in a manner corresponding to the inverse characteristic of the average spectral energy distribution curve for words and other sounds. Since equalization occurs in the pre-amplifier and in the mixing console ahead of the compressor, this. equalization must be taken into consideration when designing the equalization for the compressor control circuit.

Another factor which must be taken into consideration in compressor equalization is the variation between the spectral energy distributions. for extremely low and extremely loud speech. To illustrate this variation, reference is made to Fig. 2, wherein three frequency characteristic curves are shown for normal, soft, and loud speech, the curves being matched at 1000' cycles for purposes of comparison. An average of these curves must be taken. The equalization in the compressor must also be based on the variations between the extreme voice characteristics" of men, and the extreme voice characteristics of Women.

Thus, the response characteristic, shown by the solid line in Fig. 1, for the compressor-rectifier equalizer is one in which the average of all the above mentioned factors has been taken into consideration, the dotted curve being the actual frequency response characteristic of the compressor adjusted for a compression ration of 20 into 10 db. It will be noted that in the Miller patent, equalization was only applied from 500 cycles up, while in the present invention, the compressor is equalized over the entire range of audio frequencies. Equalization in the lower range has been found to prevent a certain boominess of the reproduced sound.

The next element in the circuit is a fixed equalizer II) which is designed to have a frequency response characteristic as shown by its curve. This equalizer compensates for the film losses, recorder slit attenuation, and for the increase of low frequencies occasioned by the reproducing level being higher than normal speech level.

By dividing the low frequency equalization between the variable equalizer I and the fixed equalizer H], many beneficial results are obtained. First, the low frequency components are not emphasized as much by the compressor. To understand this effect, reference is made to Fig. 3 wherein the curve A shows the present type of dialogue equalization which attenuates 100 cycles by approximately 12.5 db. Because in the normal voice characteristic, the 100 cycle component is about 2.5 db. below the 500 cycle component, the signal entering the compressor has more nearly the characteristic indicated by the curve B. Curve C shows the frequency characteristic of the compressor output signal when the device is set for a compression ratio of 20 into 10 db. As shown on the figure, the 100 cycle component is now only 1.5 db. below the 500 cycle component, indicating that the 100 cycle component was accentuated db. Now, referring to the other curves in the figure, curve D represents half of the total dialogue equalization which is introduced at 1 in Figure 1. Because in the normal voice characteristic, the 100 cycle component is about 2.5 db. below the 500 cycle component, the signal entering the compressor has more nearly the characteristic indicated by curve E. Again, curve F shows the frequency characteristic of the compressor output signal when the compressor is set for a compression ratio of 20 into db. Since this signal will be subjected to the effect of the second part of the dialogue equalization placed after the compressor at Iii, the resultin signal will have more nearly the resulting characteristic of curve G. Thus, the 100 cycle component is now 10.5 db. below the 500 cycle component, indicating that the 100 cycle component was accentuated only 2 db., as compared with 5 db. when all the dialogue equalization was placed ahead of the compressor.

In view of the above relationship, the incidental low frequency set noises, stage rumble, etc., will not be accentuated as much when part of the dialogue equalization is placed ahead and part after the compressor. Furthermore, if an equalizer, which is of the constant impedance type, is placed after the compressor, it will act as a buffer between the compressor output and the high or low-pass filter inputs with their inherently large impedance variations. Therefore, the disturbing reflections of these impedance variations into the compressor are considerably reduced. By locating all the high frequency compensation after the compressor to equalize for film losses, recording slit'attenuation, etc., over accentuation of sibilants by the compressor is further reduced.

Thus, it is realized that by utilizing fixedequalizationafterthe compressor in both the high and low frequency range, a compressor which compensates for spectral energy distribution distortion over the entire frequency range, variable low frequency dialogue equalization ahead of the compressor, and fixed high frequency equalization ahead of the compressor to take care of microphone variations, a more flexible sound recording channel is obtained. That is, a more desirable overall characteristic is more conveniently and consistently secured to provide better reproduction at high levels in the theater.

The other characteristics shown for the highpass and low-pass filters, amplifier, and recorders, are standard characteristics, since the amplifier and recorder characteristics are flat, while the low and high-pass filters have the necessary cutoifs in accordance with standard practice.

We claim:

1. The method of obtaining a predetermined frequency characteristic of a signal before impression on a sound recorder, comprising translating sound waves into electrical currents, said currents having a frequency characteristic in accordance with the frequency response characteristic of the translator, equalizing the translator frequency response characteristic, applying additional low frequency equalization to compensate for voice variations from normal, compressing said currents while applying equalization to the control signal, and then equalizing for film losses and recorder slit attenuation.

2. The method in accordance with claim 1, in which said control signal equalization for said compressor varies in accordance with the inverse average spectral energy distribution characteristic over the entire frequency range of the signal-being transmitted.

3. The method in accordance with claim 1, in which said translator equalization and said last mentioned equalization extends over the entire signal range.

4. A sound recording system comprising a source of electrical currents corresponding to sound waves, means for equalizing the frequency characteristic of said source, means for equalizing for the character of said sound waves, means for compressing the level range of at least a portion of said currents, a recorder, and means connected between said compressing means and said recorder for equalizing for variations in the frequency characteristic introduced by said recorder, by film processing, and by reproduction,

5. A sound recording system in accordance with claim 4, in which means are included in said compressing means for providing inverse spectral energy equalization over the signal range.

6. A sound recording system comprising a microphone having a certain frequency response characteristic, a recorder having a certain effect on the frequency characteristic of a signal to be recorded, and intermediate transmission means, said means including a compressor having an equalized control current, an equalizer between said compressor and said microphone for equalizing the frequency response characteristic thereof,

and an equalizer between said compressor and said recorder for equalizing for recording, processing, and reproducing variations.

'7. A sound recording system in accordance with claim 6, in which a mixing console is provided frequency. portion of the audio spectrum and said compression equalization is over the entire audio Spectrum l -MICHAEL RETTIN GER.

-KURT SINGER.

REFERENCES CITED The following references are of record in the file of this patent:

UNITED STATES PATENTS Number Name Date 2,154,067 Dimmick Apr. 11, 1939 2,158,268 Baker May 16, -1939 2,367,293 Levinso'n Jan. 16,-1945

Images