US2312260A - Electrical compression system - Google Patents

Description

Feb. 23, 1943.

B. F. MILLER 2312,260

ELECTRICAL COMPRESSION SYSTEM Fled May 28, 1941 2 Sheets-Sheet 1 Feb. 23, 1.943. B. MILLER 2 ELECTRICAL COMPRESSION SYSTEM Filed May 28, 1941 2 Sheets-Sheet 2 Patented Feb. 23, 1943 ELEOTRICAL COMPRESSION SYSTEM Burton F. Millei, Los Angeles, Calli., asslgnor to Radio Corporatlon of America, a corporatlon of Delaware Application May 28, 1941, Sel'il No. 395,563

13 Claims.

This lnvntion relatesto electrical current transmission systems and particularly to transmission systems wherein the amplitudes of the currents are decreased or increased in amounts dependent upon their original amplitudes, such amplitude variations being commonly known as compresson and expansion, respectively.

In the art of signal current transmission, particularly in the audio range of 'crequencies, it

has beeniound that by reducing the higher amplitudes of certain signals to a greater degree than their lower amplitudes, beneficial results are obtained. This is particularly true where the audio signals are reproduced in auditoriums or theatres by public address or sound motion picture film systems.

It is well known that an original sound signal source may have a level or amplitude range of the order of 40 to 80 db., but it has frequently been found undesirable to reproduce such 21 volume range in theatres and homes. For instance, in the reproduction of sound in theatres, the minimum level of reprduction is determined by the level of the theatre background noise produced by the audience or by theatre apparatus whlch may be operating, such as air-conditioning systems, etc. Therefore, the minimum sound reproduction level must be set at a higher value than the theatre noise level or it would otherwise be masked and. rendered unintelligible. Thus, a certain minimum level is fixed, and the reproducin-g system must be set to reproduce this minimum sound level for the lowest level passages recorded. this minimum sound level usually being higher than the lowest level of the original sound.

'Iherefore, since the lowest level of reproduction is generally righer than the lowest level of the original signa1, it is obvious that with a linear relationship existing between reproduced levels and the original levels the highest level of reproduction will exceed the highest level of the original signal by a corresponding ratio. It has been found that such amaximum signal level is entirely too loud for theatre or auditorium reproduction purposes since it is not only annoying to the listeners, but may even be painful. It is t hus desirable to introduce into the original signal channel means of compressing the volume range of the signal being transmitted.

In sound pictures this compression may be introduced either during the recording or the reproduction of the film, while in public address systems, it may be introduced at some point between the microphone and-the loudspeaker. is also applicable to radio broadcast systems.

Thus, the use of compression a 3complished by varying the relative amplitudes of the signal in accordance with their original levels has been adopted, particularly in sound recording for 1110-- tien pictures, the variation in level being accomplishea by the use of variable gain ampli- Signal flets in the signal transmisslon circuit. level control is usually obtained by rectifying a portion of the original signal and. applying the products of rectification to control the gein of one Of the signal amplifiers. Such a compressor system is disclosed and claimed in U. S. Patent No. 2225683 of September 9 1941. The operation of such a compressor system is usually made dependent upon the short-period average value of the signals being transmitted. Thus, the variaton in the gain of the amplifier is proportional to the short-period energy content of the signaling currents passing therethrough.

The term short-period as herein used, refers to the approximate time interval required for the transmisslon of any essentially constant amplitude porti0n of the signal wave train, and may be of the order of one to ten milliseconds In general, since the spectral energy distribution of complex sounds is non-uniform, the degree of compression obtained at any instant of time is actually dependent upon the amplitude of the predominant frequency or frequencies being transmitted. For example, certain words are formed of a predominant vowel portion and a less prominent sibilant portion, as in the case of the word same, where the s portion of the word is an unvoiced component, while the remainder of the word ame is a voiced component. That is, the s portion is iormed by alr passing through the teeth and lips, while the ame portion is produced by the vocal cords. It has been found that in the compression of electrical currents corresponding to such a word, the small amount of energy of the unvoiced component is insuflicient to decrease the gein of the controlled amplifler to the same degree as the ame or voiced p0rtion of the word. Thus, the

resultis that the s becomes predominant upon reproduct ion in comparison with the remalnng portion of the word. The normal type of compression system, therefore, produces a signal whose spectral energy distribution is no longer similar to that of the original sound, and which exhibits a spectral energy distribution characteristic which is a function of the signal level being transmitted. This is 'true regardless of the order of the voiced and unvoiced components within the word, and constitutes a basic form of signal distortion.

In the past it has frequently been necessary in sound recording practice to manually paint over the sound track portions of the sound record representing the unvoiced components of the signal to reduce their excessive amplitudes with respect to the amplitudes of the voiced comreproduction of the original signal.. 'I'his was a very tedious. time-consuming, and costly operation. The present invention solves this problem by providing essentially uniform compression of all portions of words or phrasesregardless of their energy content or spectral energy distribution.

It is well known that the general spectra] energy distribution curves tor speech and musical sounds indicate that the bulk of the energy of such sounds extends throughout the lower portion of the audio spectrum, while such sounds as the unvoiced components of speech, lying in the upper frequency portion of the audio spectrum, contain relatively small amounts of energy. Therefore, a compressor was devised wherein the variable gain amplifler was selectively controlled with respect to the spectra] energy distribution of the signals being compressed.

The principai object of the invention, therefore, is to improve the action of compressors of electrical signals.

Another object of the invention is to so compress audio signals as to substantially retain their original spectra] energy distribution.

A further object of the invention is to provide a signal transmission system wherein the signals are increased or decreased in amplitude at a certain uniform variation with respect to their original amplitudes.

A further object the invention is to vary the gain of an amplifler in such a manner as to retain substantially the same spectra] energy distribution in the compressed signal as existed in the original signal.

Although the novel eatures which are believed to be characteristic of this invention are pointed out with particularity in the appended claims. the manner of its organization and the mode of its operation wil] be better understood by refening to the following description reed in conjunctio n with the accompanying drawings forming a part thereof, in which Fig. 1 is a diagrammatic arrangement of an eiectrical current transmission system embodying the invention;

Fig. 2 is a. detailed schematic drawing 01 the variable gain amplifier and contro] circuits i'or the system of Fig. 1; and

Fig. 3 is a graph illustrating the action of the system of Figs. 1 and 2.

Referring now to Fig. 1, a pickup 4, which may be either a microphone, 111m or disc phonograph or other source of signal current, is cormected to an amplfier 5 which reeds a variable gain amplifler 6. The output of the variable gain amplifier 6 is impressed upon a power ampliiler 1 and then on either a recorder, loudspeaker, or a transmission system For instance, 11' the system is a sound recording system, the pickup 4 will be a microphone, while the output of the power amplifler wil] be impressed on a sound recorder which may be monitored by a loudspeaker. If the invention is -emboclied in a public address system, the pickup unit 4 may he 8" microphone and the unit 2, a loudspeaker, while the invention may be embodied in a phonograph reproducing system wherein the pickup 4 may be either a iilm soundhead or a disc phonograph pickup unit. Also corinected to the output of the variable -gain ampliiier is a selective control device II ieeding a rectiiler II, the output of the rectiiier being connected to the variable gain ampliiler 6 through a timing contro1 unit i2. Although the rectifler is shown connected to the output of the variablegain ampliiier, it is to be understood that it may also be ied tromthe input thereto.

Reterring now to Fig. 2 in which the circuit details of the variable gain ampliiler 0. selective control unit IO, reotiiler Il, and timing control unit l2 are illustrated. it will be observed that the genera] arrangement of the system is similar to that shown in the aboie-mentioned U. 8. Patent No. 2225683 of September 9, 1941. The input to the variable gein ampliiier is at terminals l5 which are connected to the primary of a push-pull transi'ormer IS, the primaries and secondaries of which are shunted by resistances I], 18 and l8 to provide suitable terminations. The secondary of transiormer IG is connected to the control grids oi variable mu tubes 2] and 22. The plates o! tubes 2] and 22 are coupled through a resistance capacity netwerk comprising resistances 24, 25, 28. 2"], 28 and 29, and condensers 20 and 22, to the grids of triode ampliiier tubes 2] and 22. The plates of ampli-- iier tubes 2! and 22 are connected to the primary of a transiormer 24, the secondary of which is shown connected to output terminals 25.

Normal grid bias tor the variable mu tubes 2! and 22 is obtained trom the voltage drop across a resistance 31 through which flow the plate and sereen grid currents of tubes 2! and 22 and the bleeder current of a voltage divider composed of resistances I], 28 and 29, resistances 3],

32 and 29 providing the correct potentie] to the screen grids of tubes 2! and 22. Resistance 4' and condenser 48 form a decoupling filter or 35 netwerk to prevent feedback due to a plate 4o tube circuits. Condenser 41 is a plate power voltage supply common to tubes 2], 22, M and 22. Condenser 45 and resistance 39 form a decoupling filter or network betweenthe scren grid circuit of tubes 2I and 22 and the other supply by-pass condenser. Grid bias ior tubes 3] and 22 is obtained trom the voltage drop across a resistance 48 due to plate current flow in these tubes.

The portion of the circuit :Iust described is the variabie gain ampliiier unit shown at 6 in Fig. 1, the gain of which is varied in a manner now to be described. The voltage generated across the secondary of transformer 34 is impressed on a selective control unit including a transformer 50, the secondary of which is shnted by a potentiometer resistance 5| having a slider 52 tor fixing the initial voltage impressed upon the grid oi-a triode amplifler 54. Also connected in the secondary circuit of transformer 50 is a selecting or energy discriminating netpass condenser ior resistance 65, while a resistance 61 and a condenser 6! form a decoupling filter or networlr ior the plate circuit of tube 54.

For obtaining an initial positive bias on the cathodes of rect iiier 84 with respect to its plates, a voltage divider comprising resistancs 10, 'I] and 12 is employed, resistance 1I being a potentiometr with an adjustable slider 'Il ior adiusting the level at whichconipression is initiated.

The rectiiied current i'r om the rectiiier 84 flows over a conductor II through a resistance 'II in series with a resistance 4I shunted by a condenser 42. The voltage drop across resistance ll controls the negative biasing voltage applied to the control grids of varlable mu tubes 2| and 22. In the compressor disclosed in the abovementioned U. S. patent, this bias voltage vari ed in accordance with the average voltage changes across the secondary of transformer 34 and thus produced variations in gain or transmission in accordance with the short-period average value of the signal levels, as explained above. Such gain variations of tubes 2! and 22 were thus based on the short-period average energy content of the signals since they were the direct result of a linear rectification of thesignal currents by the rectifler 64. However, as mentioned above, since the spectral energy distribution of words and other sounds varies in accordance with the character of the voiced and unvoiced components thereoi', it is necessary, in order to make the degree of control of the. amplifler 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 dstribution curve for words and other sounds.

To explain how this is ac-complished, reference is made to Fig. 3 in which there are illustrated spectral energy distribution characteristics of the voiced and unvoiced components of typical signals. The three solid-line curves a, b and represent three typical signals and show that the voiced components thereof lie in the lower frequency range taken, or purposes of illustration, as trom zero to approximately 1000 cycles, while the unvoiced components 11e in a range between 1000 cycles and 10,000 or above. In general, therefore, the frequencies of the sibilants lie at the high end of the audio spectrum, while the vowel components lie at the 1ow end.

Now, under the action of prior compressors. the compressed wave form of any one of signals,

a, b or c will be as illustrated by the dotted curve d. This is caused by the fact that there is insufficient energy in the unvoiced component range to reduce the gain of the variabie gain amplifier to the same extent as the gain is reduced by the voiced component range. There is thus introduced a very marked degree of distortion in the recording and reproduction of the signals, the effect being very noticable by the strong emphasis placed upon the ss and other form of sibilants. This gives a whistling character to the sibilants when reproduced.

To enablethe compression to be properly proportionate over the entire frequency range, the circuit including elements 60 and 6! is provided,

these elements serving to accentuate the amplias shown by the curve e, this curve being approximately the complement of the average signal curves a, b and 0 in the unvoiced component range. Thus, instead of compression occurring in accordance with curve d, the signals wiil be compressed in the manner shown by the combinatlon curve d --l. It is to be understood that under certain conditions of operation, it may be desirable that curve e complement the average of curves a, b and 0 throughoutthe entire frequency spectrum.

To illustrate a typical characteristic curve e, the following data indicates the marmer in which the relative grid to cathode slgnal voltage applied to tube 54 varies with irequency when constant voltage is appliedto transiormer 50:

Reiative Frequcncy voltgze in db.

As mentioned above, the variation in the volt age drop across resistance 4l due to the rectified current flowing therein produces the variation in bias on tubes 2! and 22. The resistatrce H is shunted by a condenser 42, which condensor has a double function. It serves as a by-pass condenser for the alternating components of the rectified current which varies the bias and also determines, in conjunction with resistances 4I and 16, the internal resistance of the iectifier, and the impedance of the secondary of the rectifier output transformer, the rapidity with which voltage changes across the resistance 4! can take place. That is, a certain time constant for the controi circuit is determined by this combination.

Although it is known that a compressed record may be expanded during reproduction to obtain the original sound volume range, the present application of compression serves to reduce the sound volume range through linear reproduction of the compressed record.

Therefore, the present invention may not be necessary when the signal is to be expanded dur ing reproduction since the reverse spectral energy distribution distortion is obtainable in the expanding reproducer. However, by the use of the invention in compression, the recorded record may be reproduced either in a linear manner with a standard reproducer or with a simple expanding reproducerembodying a complementary selective control netwerk. Thus, the present invention permits the record to be reproduced in either a linear or expanding type of system, whereas without the invention the record could be faithfully reproduced only through a complementary expander. Another lfeature of the present invention is that it permits a compressed record to be reproduced in a standard linear reproducing system withoutspectral energy distribution distortion, there being no type of equalization as such which wiliremove this distortion. Thus such a distorted compressed record must be reproduced by a complementary expander system in order to faithfully convert the signals to their original amplitude relationship.

Aithough the invention has been shown asso ciated;with a variable gain ampiifler, it is to be understood that it may also beembodied in any type of variable transmission system using other types of variable transmission elements.

What I claim as my invention is: 1. In en lectricel current transmission system in which the amplitudes of the current are automatically varled.in accordance with their original relative va.lues the combination o! means for generating en electrical current corresponding to a. signal, a variable transmission element for sa.id current, means for varying the relative transmission of sa.id currents through said transmission element. said varying meaus including a rectifler 3. The method of proportionately varying the amplitude of electrical currents corresponding to 9. signa.l having voiced and unvoiced components which normally r'eceive different degrees of amplification, comprising passing the signal through 9. signal channel including en amplifier, deriving a. control signal from the signal channel having an a.verage spectral energy distribution characteristic varying in accordance with the inverse average spectral energy distribution characteristic of the signal, and utilizing said controi signal to vary the gein of said amplifier to provide en output signal whose spectral energy distribution chara.cteristic corresponds to the energy distribution characteristic of the input signal.

4. The method of variably transmitting signals in accordance with their amplitudes, said signals having an energy distribution dependent upon the frequency of the components of sa.id signals, comprising passing the signals through a signa.l channel including an amplifier, deriving control signals from sa.id signal channel direotly proportional to the short-period avera.ge energy content of said signals, said control signals having respective spectral energy distribution characteristics varying in accordance with the inverse average spectra! energy distribution characteristics of the. signals being transmitted, and utilizing sa,id control signals to vary the gain ofsaid amplifler to obtain correSpondence between the, respective spectral nergy distribution characteristics of the input and output signals.

5. A system for transmitting signaling eurrents at cmpressed amplitudes comprising a. transmission line for said currents, a variab1e gain amplifler in said line, means ior rectifying a. portion of said signaling currents, means for connecting the input and output of said rectifying means to said variabie gain amplifier, and means connected in the input to said rectifying means for obtaining a, rectifled current froni said signaling currents in accordance with the inverse average spectral energy distribution of said signalling currents to provide an output signal whose spectral energy distribution substantially corresponds to the spectral energy distribution of the input signal.

6. A system in accordance with claim 5 in which said last-mentioned menas comprises a. frequency discriminating filter.

7. A system in acc0rdanoe with claim 5 in which ineens are provided in the output of zuid rectiiylng means or 'controlling the ra.te of impressin of sidrectiiied currents on said variable gein amplifler.

8. An electrioal current compressor system 101. audio signals comprising a. va.riable gein amplifler, a rectiiier, means tor connecting the input. and output circuits of said rectifler to said variablegain amplifier, means connected in the input circuit; ot said rectifier fox varying the relative amplitude of impression of sa.id audio signais on said rectifler, said last-mentioned mea.ns varying the ratio of rectified control signal magnitude to average rectifier signal input magnitude in nocordance with the inverse a.verage spectral energy distribution chara.cteristlc of sa.id signals to provide output signals whose spectral energy distribution characteristics substantially correspond to the spectral energy distribution characteristics of respective input signals.

9. An electrical current compressor system in accordance with claim 8 in which e. timing circuit is provided in the output cirouit-otsaid rectifier.

10. An electrical current compressor system in accordance with claim 8 in which said last-mentioned means includes a corrective network having a transmission characteristic approxixnating the inverse a.vrage spectral energy distribution characteristic of said signals.

11. An electricl current compressor system ifor audio signals oomprising a source of electrical signals, a variable gein amplifler or variably amplifying said signals, a. rectifier for said signals, said rectifier producing a directcurrent whose amplitude is proportional to the short-period average energy distribution of said signals, means tor connecting the output -of said rectifier to said variable gein amplifler to vary the-gein thereof in inverse proprtion to the amplitude of said direct current, a.ncl means in the input of said rectifler for varying the impression of said signals on said rectifler in a relationship whereby the output of said. rectiiier varies with the frequency of the predominant components of said signals being transmitted through said variable gain amplifler in accordance with the inverse characteristic of the average spectral energy distribution curve of said signals to provide an output signa-l whose spectral energy distribution substantially corresponds to the spectral energy distribution of th input signal. 12. In an electrical signal variable gain amplifler systern in which the spectral energy distribution of the output signal does not correspond with the spectral energy distribution of the input signal, the method of pmduoing a correspondence between the input and output spectra,l energy disspectral energy distn'bution of the input signal.

13. The method in accordance with claim 12 in which the control signal varies the control of said variable gein amplifier ata, predeterinined rate.

Images