US3292116A

US3292116A - Dynamic speech equalizing system having a control circuit that separates and compares the high and low frequency energy

Info

Publication number: US3292116A
Application number: US353406A
Authority: US
Inventors: Charles E Walker; Joseph F Goveia
Original assignee: Hazeltine Research Inc
Current assignee: Hazeltine Research Inc
Priority date: 1964-03-20
Filing date: 1964-03-20
Publication date: 1966-12-13
Anticipated expiration: 1983-12-13

Description

c. E. WALKER ET AL 3,292,116

Dec. 13, 1966 DYNAMIC SPEECH EQUALIZING SYSTEM HAVING A CONTROL CIRCUIT THAT SEPARATES AND COMPARES THE HIGH AND LOW FREQUENCY ENERGY 2 Sheefs-Sheet 1 Omx mmg

Filed March 20, 1964 Dec. 13, 1966 Filed March 20, 1964 RELATIVE INTENSITY LEVEL (db) c. E. WALKER ET AL 3,292,116

DYNAMIC SPEECH EQUALIZING SYSTEM HAVING A CONTROL CIRCUIT THAT SEPARATES AND COMPARES THE HIGH AND LOW FREQUENCY ENERGY 2 Sheets-Sheet :J

Gdb/OCTAVE |60.

IOO

FREQUENCY (cps) FIG. 2

United States Patent O DYNAMIC SPEECH EQUALIZING SYSTEM HAV- ING A CONTROL CIRCUIT THAT SEPARATES AND COMPARES THE HIGH AND LOW FRE- QUENCY ENERGY Charles E. Walker, Norweil, and Joseph F. Goveia, North Easton, Mass., assignors to Hazeltine Research, Inc., a corporation of Illinois Filed Mar. 20, 1964. Ser. No. 353,406 4 Claims. (Cl. S33-18) The present invention is directed to voice communications systems and more particularly to systems for improving the intelligibility of a transmitted vocal message.

Statistical analyses of human voice power spectrum densities indicate that most of the vocal energy is concentrated at frequencies below 500' cycles per second and, at higher frequencies in which the greatest amount of intelligence is transmitted, the spectral density decreases at a significant rate. This presents a problem where machinery or other low-frequency noise-producing elements in a broadcasting area have a tendency to dest-roy the intelligibility by masking high-frequency energy content of the measure. Speech processing techniques, such as pre-emphasis, compression amplifiers and speech clipping, are presently used to increase the modulation index or percentage of energy transmitted at different frequencies in an attempt to form a signal which will best convey an intelligible message. Pre-emphasis circuits presently used are passive networks with a fixed attenuation slope and corner frequency based on the statistical characteristics of many voice spectrums. Since each person has a different power spectrum, manual controls for the adjustment of the speech amplifying circuits are necessary to increase the intelligibility of the vocal message. However, manual adjustment of the speech amplifying circuits is not efficient because the trial and error method of selecting the correct amount of amplification is required since the exact power density spectrum of the speaker is not known at the start of transmission.

It is an object of the present invention to provide a new and improved speech equalizing circuit whereby the intelligibility of the transmitted vocal message is automatically increased to its optimum value.

It is another object of the present invention to provide an efficient and inexpensive method of increasing the high-frequency vocal energy in proportion to the ratio between the highand low-frequency energy.

In accordance with the present invention there is provided a dynamic speech equalizing system for improving the intelligibility of a transmitted vocal message including first means for providing an adjustable increase in the relative energy content of the high-frequency portion of an input signal in the vocal frequency spectrum, second means responsive to the input signal for providing output signals representative of the relative energy contents of the highand low-frequency portions of the input signal, and third means responsive to the out-put signals for controlling the operation of the first means so that the relative energy content of the high-frequency portion is increased in relation to the relative energy contents of the highand low-frequency portions of the input signal.

For a better understanding of the present invention, together with other and further objects thereof, reference is had to the following description, taken in connection with the accompanying drawings, and its scope will be pointed out in the appended claims.

Referring to the drawings:

FIG. 1 is a schematic of a dynamic speech equalizing system constructed in accordance with the present invention, and

FIG. 2 shows a statistical analysis of a large number Patented Dec. 13, 1966 ICC of human voice power spectrum densities together with curves of use in vthe explanation of the operation of the FIG. 1 system.

General description Referring now to FIG. 1 there is represented one embodiment of a dynamic speech equalizing system for increasing the intelligibility of a transmitted vocal message constructed in accordance wtih the present invention. As illustrated, the system of FIG. 1 includes a source 11 of transmitted signals in the vocal frequency spectrum, which may be a microphone, coupled to equalization control circuit 17 and equalization circuit 13 through bandpass preamplifier 12 which has a frequency response between to 5000 cycles per second. The sharp low-frequency cutoff of 100 cycles per second is chosen in order to eliminate spurious low-frequency noise which serves only to destroy tlhe intelligibility of the signal and the high-frequency cutoff of 5000 cycles per second is selected since signals above this range contain negligible intelligence.

Equalization circuit 13 includes first means, equalizing circuit 16, for providing an adjustable increase in the relative energy content of the high-frequency portion of the input signal. As illustrated, equilization circuit 13 further comprises bandpass filter 14 having -a frequency response between 30() and 3000 cycles per second coupled to the output of bandpass preamplifier 12 and amplifier 15 coupled to the output of bandpass filter 14. The frequency response of bandpass filter 14 is :restricted to 300 to 3000 cycles per second since most of the intelligible energy is located in this frequency range. The output of amplifier 15 is coupled to equalizing circuit 16 which com-v prises four selectively controlled pre-emphasis

networks

16a, 16b, 16e and 16d, each network comprising a capacitor serially coupled through a resistor and Zener diode to ground. As illustrated, the values of the components in each network are such, that each network has a positive rising frequency slope characteristic of 6 db per octave.

Equilization control circuit 17 includes second means responsive to an input signal in the vocal frequency spectrum for providing Output signals representative of the relative energy contents of the highand low-frequency portions of said input signal. As illustrated, second means comprises low-frequency bandpass filter 18, having a frequency response between 100-500 cycles per second, coupled through amplifier 20 to detector and smoothing circut 22. Detector and smoothing circuit 22 is of such construction that it produces a substantially D.C. signal representative of the relative energy content of the lowfrequency portion of the input signal.

Second means further comprises high-frequency bandpass filter 19, having a frequency response between 500- 5000 cycles per second, coupled through amplifier 21 t0 detector and smoothing circuit 23. Detector and smoothing circuit 23 is of such construction that it produces a substantially D.C. signal representative of the relative energy content of the high-frequency portion of the input signal.

Equalization control circuit 17 further includes third means responsive to the output signals from the diode detector and

smoothing circuits

22 and 23 for controlling the operation of the first means so that the lrelative energy content of the high-frequency portion is increased in relation to the relative energy contents of the highand low-frequency portions of the input signal. As illustrated, lthird means comprises D.C. voltage comparator 24 coupled to emitter-follower circuits 25a- 25d. The D.C. voltage comparator 24 is of such construction that when the output signals from detector and `

smoothing circuits

22 and 23 are applied to its input termina-ls simultaneously, it produces an output signal representative of the ratio between the rela-tive energy contents of the lowand high-frequency signals. Bmitterfollowers 25a-25d, respectively, are coupled to pre-emphasis networks 16a16d, respectively, in equalization circuit 13 and control the operation of those networks, selecting any or all of the pre-emphasis networks depending upon the ratio between the relative energy contents of the lowand high-frequency signals.

As illustrated, the system of FIG. 1 further comprises amplifier 26 coupled to the output of equalization circuit i13 for additional iamplification of the adjusted signal. The output of amplifier 26 is coupled to fourth means, clipping circuit 27, which is responsive to the adjusted signal output from the first means for increasing the effective speech power of the adjusted signal. Clipping circu-ift 27 is of such construction that it provides 2() to 30 db of hard clipping on normal speech variations. This clipping increases -the effective yspeech power and minimizes the effects of variations in distance between the speaker and microphone and the variations in sound intensity created by this variation. Bandpass filter 28 is connected to the output of clipping circuit 27 and has a high-frequency cutoff of 3000 cycles per second in order to eliminate the high-frequency components in the adjusted signal generated during clipping.

'Ihe system of FIG. 1 further comprises fifth means, single sideband filter 3i), coupled to the fourth means through bandpass filter 28 and balanced modulator 29 for restricting the voice communication channel to a predetermined bandwidth for use in single sideband communications. As illustrated, single sideband filter has a frequency lresponse between 380 and 1750 cycles per second since this is a range which is utilized extensively in single sideband communications.

Operation of the FIG. I dynamic speech equalizing .system Considering now the operation of the dynamic speech equalizing system of FIG. 1, a message is spoken into microphone 11 and converted int-o electrical energy. Bandpass preamplifier 12 amplifies the signal and restricts it to a frequency range between U-S000 cycles per second. The signal is thereupon applied simultaneously to equalization control circuit 17 and equalization circuit 13.

In the operation 4of' equalization control circuit 17, the vocal signal input is divided into its low-frequency portion (10U-500 cycles per second) and its high-frequency portion (50G-5000 cycles per second) by

bandpass filters

18 and 19, respectively. The low-frequency and high-frequency signals `are then amplified by

amplifiers

20 and 21 respectively before being detected and averaged by diode detector and smoothing

4circuits

22 and 23, respectively. Both D.C. voltages are then applied to D.-C. voltage comparator 24 which produces `an output representative of the ratio between the relative energy contents of the lowand high-frequency signais. The output from D.C. 'voltage comparator 24 is then applied simultaneously to emitter-

follower circuits

25a, 25h, 25e and 25d whose respective output voltages selectively control the operation of

pre-emphasis networks

16a, 16b, 16e and 16d in equalizing circuit 16. Thus, depending upon the ratio between the relative energy contents of the lowand high-frequency signals one, two, three or all four emitter-follower circuits will produce output voltage signals that will exceed the breakdown voltages of the zene-r `diodes in pre-emphasis networks :16a-16d in equalizing circuit 16 and thereby change the over-all slope of equalizing circuit 16 in 6 db per octave incremental steps.

In the operation of equalization circuit 13, the vocal signal input bandwidth is restricted to a range between 300 tand 3000 cycles per second. Referring now to FIG. 2, there is represented a `statistical analysis of a large number of human vocal power spectrum densities which show that the greatest amount of vocal energy is concentrated lbetween 0 4and 500 cycles per second and, that over this frequency range, the relative intensity of the signal remains reasonably constant. Although, by selecting the low-frequency cutoff of 300 cycles per second, some of the vocal power will be lost, the increase in the relative high-frequency energy by equalizing circuit 16 will more than compensate for the high vocal power, negligibly intelligent, low-frequency signals. The high-frequency cutoff of 30010 cycles per second is selected in bandpass filter 14 because the vocal energy contained in .frequencies above 3000 cycles per second contain negligible intelligence` Equalization circuit 13 further comprises amplifier 15 and equalizing circuit 16 coupled to the output of bandpass filter 14. The output signal from baudpass filter 14, having a frequency bandwidth from 300 to 3000 cycles per second, is then amplified by amplifier 15. The magnitude and slope of the amplified signal is dependent upon which of the four selectively controlled

pre-emphasis networks

16a, 16b, 16C and 16d, respectively, each having positive rising frequency slope characteristics of 6 db per octave, are included in the output. Depending upon the proportion between the high-frequency energy content and the low-frequency energy content of the input signal, as determined by equalization control circuit 17, the output signal from amplifier 15 is varied in positive 6, 12, 18, or 24 db per octave increments. Referring now to FIG. 2, it can be seen that from 100 to 500 cycles per second the relative intensity level remains reasonably constant and from 500 to 3000 cycles per second, wherein the greatest amount of intelligible energy lies, the spectral density decreases at a rate of approximately l2 to 24 db per octave. Therefore, the selection of

pre-emphasis networks

16a, or 16a and 1Gb, or 16a, 16b and 16C, or 16a, 16h, 16C and 16d by equalization control circuit 17 equalizes the decrease in the relative intensity level of the intelligible high-frequency energy and has negligible effect on the low-frequency energy whose spectral density remains nearly constant. The addition of one or

more pre-emphasis networks

16a, 1619, 16e and 16d to the output of amplifier 15 is accomplished by exceeding the breakdown voltages of the Zener diodes in

pre-emphasis networks

16a, 16h, 16e and 16d and thereby connecting each respective network to ground. In the analysis depicted in FIG. 2, the four positive rising slopes are shown breaking at the same frequency based on the values of the components given infra. It should be understood that by a different selection of values for the components in the pre-emphasis networks each slope can be made to break at a different frequency than the other slopes and also have different characteristics. l

The output signal from equalization circuit 13 is then applied to amplifier 26 which adjusts the level of the signal to drive clipping amplifier 27. Clipping amplifier 27 is adjusted to provide 2() to 30 db of clipping on normal speech in order to increase the effective speech power and to remove amplitude variations peculiar to certain human vocal characteristics. The output of clipping amplifier 27 is coupled to balanced modulator 29 through bandpass filter 28 which has a high-frequency cutoff of 3000 cycles per second in order to eliminate high-frequency components generated during clipping. The output of balanced modulator 29 is coupled to single sideband filter 30 which has a frequency bandwidth of 380 to 1750 cycles per second in order to achieve more effective communication for any given size and weight of equipment.

From the foregoing description, it is seen that the present invention embodies a system wherein the intensity of the high-frequency vocal energy of a transmitted vocal message is increased in proportion to the ratio between the high-frequency energy and the amount of low-frequency vocal energy contained in the message itself as well as the amount of low-frequency energy introduced by undesirable sources. Equalization circuit 13 contains four selectively controlled

pre-emphasis networks

16a, 16h, 16C and 16d which will vary the intensity of the high-frequency vocal energy anywhere from 6 to 24 db per octave but have negligible effect on the intensity of the low-frequency vocal and spurious energy. Equalization control circuit 17 compares the relative energy contents of the low-frequency and the high-frequency portions of the input signal and determines to what extent the intensity of the high-frequency energy should be increased to achieve optimum intelligible transmission.

While applicants do not wish to be limited to any particular set of circuit constants, the following have proved useful in the pre-emphasis networks of FTG. l.

Pre-emphasis network 16a:

Resistor ohms 200 Capacior microfarad .266

Zener diode type 1N746AM Pre-emphasis network 161;:

Resistor ohms-- 2,000

Capacitor microfarad .0266

Zener diode type 1N750AM Pre-emphasis network 16C:

Resistor ohms 20,000

Capacitor microfarad .00266 Zener diode type 1N754AM Pre-emphasis network 16d:

Resistor ohms 200,000

Capacitor microfarad .000266 Zener diode type lN757AM While there has been described what is at present considered to be the preferred embodiment of this invention, it will be obvious to those skilled in the art that various changes and modifications may be made therein without departing from the invention and it is, therefore, aimed to cover all such changes and modifications as fall within the true spirit and scope of the invention.

What is claimed is:

1. A dynamic speech equalizing system for improving the intelligibility of a transmitted vocal message comprising:

rst equalization means comprising a plurality of selectively controlled pre-emphasis networks, each network having a positive rising frequency slope characteristic, for adjusting the relative energy content of the high-frequency portion of an input signal in the vocal frequency spectrum in incremental steps;

second means responsive to said input signal for separating the high-frequency energy content from the low-frequency content and averaging said separated highand low-frequency contents for providing output signals, the amplitudes of which are representative of the relative energy contents of the highand low-frequency portions of said input signal;

and third means responsive to said output signals for comparing their respective amplitudes and controlling the operation of said first equalization means so that the relative energy content of said high-frequency portion is increased in incremental steps in relation to the relative energy contents of said highand lowfrequency portions of said input signal.

2. A dynamic speech equalizing system Vfor improving 6 the intelligibility of a transmitted vocal message according to claim l, which further comprises:

fourth means comprising clipping circuitry responsive to the adjusted vocal signal output from said first equalization means for increasing the effective speech power of said adjusted signal;

and fifth means coupled to said fourth means for restricting the voice communication channel to a predetermined bandwidth for use in single sideband communications.

3. A dynamic speech equalizing system for improving the intelligibility of a transmitted vocal message comprising:

a source of transmitted signals in the vocal frequency spectrum;

first equalization means comprising a plurality of selectively controlled pre-emphasis networks, each network having a positive rising frequency slope characteristic of 6 db per octave for adjusting the relative energy content of the high-frequency portion of said input signal in 6 db per octave steps;

Second means including a plurality of bandpass filters responsive to said input signal for separating the high-frequency energy content from the low-frequency energy content and a plurality of detector and smoothing circuits for averaging said separated highand low-frequency energy contents for providing output signals, the amplitudes of which yare representative of the relative energy contents of the highand lowfrequency portions of said input signal;

and third means comprising a D.C. voltage comparator circuit responsive to said output signals for comparing the respective amplitudes and controlling the operation of said first equalization means so that the relative energy content of said high-frequency portion is increased by 6 db per octave, l2 db per octave, 18 db per octave, or 24 db per octave steps in accordance with the relative energy contents of said highand low-frequency portions of said input signal.

4. A dynamic speech equalizing system for improving the intelligibility of a transmitted vocal message according to claim 3, which further comprises:

fourth means responsive to the adjusted vocal signal output from said first equalization means for providing 20 to 30 db of clipping of voltage variations in said adjusted signal;

and fifth means comprising a single sideband filter coupled to said fourth means for restricting the voice communication channel to a predetermined bandwidth for use in single sideband communications.

References Cited by the Examiner UNITED STATES PATENTS 1,743,132 l/l930 Green 333-18 X 2,606,971 8/1952 Scott 333-18 3,126,449 3/1964 Shrman l79-l.8 3,176,073 3/1965 Samuelson et al. 179-1 HERMAN KARL SAALBACH, Primary Examiner.

P. L. GENSLER, Assistant Examiner,

Claims

1. A DYNAMIC SPEECH EQUALIZING SYSTEM FOR IMPROVING THE INTELLIGIBILITY OF A TRANSMITTED VOCAL MESSAGE COMPRISING: FIRST EQUALIZATION MEANS COMPRISING A PLURALITY OF SELECTIVELY CONTROLLED PRE-EMPHASIS NETWORKS, EACH NETWORK HAVING A POSITIVE RISING FREQUENCY SLOPE CHARACTERISTIC, FOR ADJUSTING THE RELATIVE ENERGY CONTENT OF THE LIGHT-FREQUENCY PORTION OF AN INPUT SIGNAL IN THE VOCAL FREQUENCY SPECTRUM IN INCREMENTAL STEPS; SECOND MEANS RESPONSIVE TO SAID INPUT SIGNAL FOR SEPARATING THE HIGH-FREQUENCY ENERGY CONTENT FROM THE LOW-FREQUENCY CONTENT AND AVERAGING SAID SEPARATED HIGH- AND LOW-FREQUENCY CONTENTS FOR PROVIDING OUTPUT SIGNALS, THE AMPLITUDES OF WHICH ARE REPRESENTATIVE OF THE RELATIVE ENERGY CONTENTS OF THE HIGH- AND LOW-FREQUENCY PORTIONS OF SAID INPUT SIGNAL;