U.S. patent number 3,746,792 [Application Number 05/046,345] was granted by the patent office on 1973-07-17 for multidirectional sound system.
Invention is credited to Peter Scheiber.
United States Patent |
3,746,792 |
Scheiber |
July 17, 1973 |
**Please see images for:
( Certificate of Correction ) ** |
MULTIDIRECTIONAL SOUND SYSTEM
Abstract
There is disclosed a sound system for producing at least three
and typically four sound outputs from respectively different
directions from the listener wherein the sound content per se and
the directional information are encoded on a conventional
standardized two-channel record or on a transmission by a
conventional two-channel broadcasting medium such as stereo FM. The
system is maximally compatible with existing systems in the sense
that reproduction on existing two-direction (stereo) or
one-direction (mono) sound systems is completely satisfactory
although, of course, the extent of directionality reproduction is
limited by the inherent characteristics of such existing systems.
In a typical example, the system provides for four directional
sound inputs with equal, 90.degree. separation around a circle. The
four sound inputs are fed to two sound channels, for example stereo
recording or transmission channels. Each of the four sound input
channels is fed in part to each of the two stereo channels but the
polarity and/or amplitude of each sound input channel is different
in each of the stereo channels. Conventional analog computer
electronic circuits may be utilized to transform the four sound
input signals into two stereo channels according to prescribed
formulae. The system provides for reproduction of sound from four
loudspeakers located in the four corners of a room and having
nominal positions with respect to the listener of left front, right
front, left rear and right rear. The two stereo channels are
combined according to different formulae setting forth different
amplitudes and/or polarities to produce four output sound channels.
Assuming these four directions correspond to directions of the four
input sound channels, and sound originating from a particular sound
input channel is reproduced predominantly in the corresponding
loudspeaker. Refinements for the system control the gain for the
respective loudspeakers to permit sound from a particular input
sound channel to be localized to a particular corresponding output
loudspeaker. In generalizations of the system, the number of inputs
and the number of loudspeakers may be greater or less than four
(but always more than two) and the numbers and/or directions of the
input sound channels may not correspond to the numbers and
directions of the outputs feeding the loudspeakers.
Inventors: |
Scheiber; Peter (Peekskill,
NY) |
Family
ID: |
27489051 |
Appl.
No.: |
05/046,345 |
Filed: |
June 15, 1970 |
Related U.S. Patent Documents
|
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
Issue Date |
|
|
697103 |
Jan 11, 1968 |
|
|
|
|
853822 |
Aug 28, 1969 |
|
|
|
|
888440 |
Dec 29, 1969 |
3632886 |
|
|
|
Current U.S.
Class: |
381/23 |
Current CPC
Class: |
H04S
3/02 (20130101) |
Current International
Class: |
H04S
3/00 (20060101); H04S 3/02 (20060101); H04h
005/00 () |
Field of
Search: |
;179/1G,1GP,15ST,1.4ST,1.1TD |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
Other References
"Circuits for Three-Channel Sterophonic Playback Derived from Two
Sound Tracks"-Klipsch IRE Transactions Nov.-Dec. 59. .
"Three-Channel Stereo Playback of Two Tracks Derived from Three
Microphones" Klipsch IRE Transactions March-April 59..
|
Primary Examiner: Claffy; Kathleen H.
Assistant Examiner: D'Amico; Thomas
Parent Case Text
This application is a continuation-in-part of prior copending
application Ser. No. 697,103 filed Jan. 11, 1968 entitled
STEREOPHONIC SOUND SYSTEM; Ser. No. 853,822 filed Aug. 28, 1969
entitled STEREOPHONIC RECORDING AND TRANSMISSION SYSTEM (now
abandoned); and application Ser. No. 888,440 filed Dec. 29, 1969
entitled QUADRASONIC SOUND SYSTEM now U.S. Pat. No. 3,632,886, all
in the name of Peter Scheiber.
A complete system in accordance with the invention would include
multidirectional sound pickup apparatus, encoding apparatus for
converting the multiple channels from the pickup apparatus into
only two separate electronic channels, recording apparatus and
playback apparatus (or transmitting apparatus and receiving
apparatus), decoding apparatus for producing multidirectional sound
electrical signals correlating respectively with the
multidirectional sound input signals, and audio amplifiers and
loudspeakers or equivalent apparatus for producing a
multidirectional sound effect for one or more listeners.
Multidirectional shall herein be interpreted to mean representative
of at least three directions of sound, and in a typical case four
directions of sound. It is permissible and in fact highly desirable
for certain portions of the system described above to be strictly
conventional. For example, transmitting and receiving portions of a
system would typically consist of a conventional stereo FM
broadcast transmitter and an FM broadcast receiver.
For many years the enjoyment of stereo or bidirectional sound
reproduction has been a common reality, at least in this country.
However, such bidirectional sound reproduction apparatus has
serious inadequacies in reproducing music or other audio
entertainment with an effect approaching that obtainable with live
performances.
The desirability of expanding the bidirectional sound techniques to
multidirectional sound techniques has been expounded in prior
copending Scheiber patent applications and elsewhere. The
desirability of multidirectional sound systems over bidirectional
sound systems is intuitively apparent when one appreciates that the
bidirectional sound system can only simulate sound sources existing
within a limited angle of substantially less than 180.degree.,
whereas a multidirectional sound system can simulate sounds
originating from any direction, thus encompassing 360.degree.. The
augmented realism and impact of multidirectional sound over
bidirectional sound is fully borne out by actual experience.
In theory, there is no obstacle to creating multidirectional sound
systems. Thus, a four-direction sound system for magnetic tape can
be extrapolated in an obvious way from stereo tape systems. Instead
of using two magnetic tracks on the magnetic tape, one uses four
tracks. In effect, the stereo system which resulted by effectively
doubling the monaural system is again doubled to produce a
four-direction system. Such a four-track sound system, while
feasible in theory, fails to solve numerous practical difficulties.
The essential difficulty is that one has effectively doubled the
entire system for reproduction of sound. Thus, if one wishes to
transmit by radio, it would require two stereo FM radio stations
rather than the one stereo FM station required to broadcast
bidirectional sound. For magnetic recording, twice as many record
or playback heads are required, twice as many tape tracks, and
twice as much tape. The difficulty in the case of disc recordings
is even greater since it appears that two grooves or tracks on the
disc would be required, or else some complex subterfuge to obtain
an equivalent recording capacity, neither of which has been found
to be practical.
In accordance with the present invention, it has been found that
substantially all the directionality information which can be
reproduced by four loudspeakers and appreciated by a listener can
be recorded and transmitted without doubling and in fact ithout
substantially increasing the basic information-carrying capacity of
two conventional audio channels (for example those of a stereo
record or stereo FM radio broadcast).
In the system according to the invention, the directionality
information is encoded in the amplitude and phase (or polarity)
relationships of each multidirectional input signal in one of the
stereo channels as compared to the other channel. While the
directionality information is not recorded with a precision equal
to that achieved for the frequency components making up the audio
information per se, this is quite unimportant in view of the
inadequacies of reproduction and evaluation of directional
information in the system overall (including the listener). By way
of example, note that in any directional sound system there are a
finite number of loudspeakers, and sounds emanating other than from
those precise positions must be approximated by activating
loudspeakers in different positions with appropriate amplitudes.
Thus a sound which should emanate from directly in front of the
listener must be approximated by sound from both a left front and a
right front speaker.
In the present invention, the basic directional encoding and
decoding arrangement utilizing amplitude and phase relation between
the two stereo channels may be augmented by a gain control
arrangement for the output signals. Such a gain control system
permits the sound to be localized in one loudspeaker to an extent
greater than would be achieved with the basic encoding and decoding
system.
In a preferred embodiment, the invention also provides a system
which is maximally compatible with existing stereo FM broadcasting
and with stereo disc recording. For example, if a disc record in
accordance with the present invention is played and the sound is
reproduced from a monaural amplifier or monaural FM radio receiver,
or from a stereo amplifier or from a stereo FM radio receiver, the
result is highly acceptable monaural or stereo sound reproduction
as the case may be. The reproduction in either case includes all
four sound inputs of the recording but with the rear sound input
diminished in amplitude from the amplitude they would have if
reproduced with a four-loudspeaker system in accordance with the
invention. No directionality distortion is introduced in mono or
stereo reproduction in the sense that the right channels and the
left channels are produced equally in the monaural playback and in
the stereo playback are produced correctly in the best possible
approximation to the four-speaker reproduction.
For simplicity of explanation, it is convenient to think in terms
of a multidirectional sound system with four loudspeakers situated
in the corner of a substantially square room reproducing material
having four input signals corresponding in direction to the four
loudspeakers utilized in reproduction. Such an arrangement is
encompassed in the preferred embodiment of the system. However, a
generalization of the system is presented such that the number of
inputs to the encoder are not limited to four, nor are they limited
as to the direction which is to be represented by an output. In
fact, the direction represented by an input signal is determined by
the values of certain resistors, and it is readily possible to
provide variable resistor networks such that the direction
represented by an input can be varied at will or can be set for any
desired direction. In a similar manner, the direction represented
by the output loudspeakers can be changed from the previously
described 90.degree. separation to some other values. The direction
associated with a loudspeaker output may thus be changed to
accommodate the necessity of placing a speaker in an abnormal
position. However, the position represented by the loudspeaker
output does not necessarily have to conform to its physical
position. Interesting effects can thereby be obtained. It will be
seen that one can increase or decrease the angle which a
distributed sound source such as an orchestra appears to subtend.
One can thus control the signal to the loudspeakers to create the
effect of moving from the rear of a hall where the orchestra
subtends a relatively small angle to the front of the hall (or in
fact the podium) where the orchestra subtends a much larger
angle.
The assignment of any desired direction to a loudspeaker output may
also prove desirable for special situations such as
automobile-installed systems wherein the loudspeaker placement may
best be other than the left front, right front, left rear, right
rear placement usual in a living room or studio.
In addition to providing the advantages described above, it is an
object of the present invention to provide a multidirectional sound
system such that records or transmissions in accordance with the
system may be played on existing standard bidirectional or
monodirectional equipment with completely satisfactory results
comparable to recordings or transmissions produced with only stereo
or mono information content.
It is another object of the present invention to provide a
multidirectional sound system wherein three or more sound input
signals are processed and transmitted over two channels, which two
channel signals are inversely processed to provide three or more
multidirectional output signals correlating to the input
signals.
It is another object of the present invention to provide a
multidirection sound system in which the assigned direction for
sound inputs or outputs may be determined by the assignment of
amplitude values as determined by resistor values or the like with
the result that the effective direction assigned to an input sound
signal or that assigned to an output sound signal can be adjustably
determined within a wide angle.
Claims
What is claimed is:
1. In a multidirectional sound system for encoding at least three
directonal input sound signals on A and B audio channels and
reproducing from the A and B channels at least three directional
output sound signals correlated with the input signals, encoder
apparatus comprising at least three inputs for input sound signals
having respective position agnles associated therewith, first means
for generating an A channel signal connected to at least three of
said inputs, said first means causing the amplitude of each said
input sound signal in said A channel to be substantially
proportional to the cosine of one half the angular difference
between the input position angle and the angle assigned to the A
channel, second means for generating a B channel signal connected
to at least three of said inputs, said second means causing the
amplitude of each said input sound signal in said B channel to be
substantially proportional to the cosine of one half the angular
difference between the input position angle and the angle assigned
to the B channel, the angles assigned to said A and B channels
differing by approximately 180 degrees, decoder apparatus
comprising an A channel input and a B channel input, means for
communicating said A channel and B channel signals to said decoder
apparatus inputs, first, second and third means for generating
first, second and third directional sound output signals each
having a respective position angle associated therewith, each said
means being connected to each of said A and B inputs and causing
the amplitude of each of said A and B inputs in said output sound
signal to be substantially proportional to the cosine of one half
the angular difference between the output position angle and the
angle assigned to the respective A or B input.
2. Apparatus as claimed in claim 1 wherein said A channel and B
channel generating means cause the polarity of at least one of said
input sound signals in said A channel to be opposite to its
polarity in said B channel.
3. Apparatus as claimed in claim 1 wherein said first, second and
third means for generating output signals cause at least one of
said A and B channel signals in one of said outputs to be opposite
to its polarity in at least one other of said outputs and the same
as its polarity in at least another of said outputs.
4. Apparatus as claimed in claim 3 wherein said A channel and B
channel generating means cause the polarity of at least one of said
input sound signals in said A channel to be opposite to its
polarity in said B channel.
5. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels at least three directional
output sound signals correlated with the input signals, encoder
apparatus comprising at least three inputs for input sound signals
having respective position angles associated therewith, first means
for generating an A channel signal connected to at least three of
said inputs, said first means causing the amplitude of each said
input sound signal in said A channel to be substantially
proportional to the cosine of one half the angular difference
between the input position angle and the angle assigned to the A
channel, and second means for generating a B channel signal
connected to at least three of said inputs, said second means
causing the amplitude of each said input sound signal in said B
channel to be substantially proportional to the cosine of one half
the angular difference between the input position angle and the
angle assigned to the B channel, the angles assigned to said A and
B channels differing by approximately 180 degrees.
6. Apparatus as claimed in claim 5 wherein said A channel and B
channel generating means cause the polarity of at least one of said
input sound signals in said A channel to be opposite to its
polarity in said B channel.
7. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels at least three directional
output sound signals correlated with the input signals, decoder
apparatus comprising an A input and a B input, at least three
directional sound signal outputs, first, second and third means for
generating first, second and third directional sound output signals
each having a respective position angle associated therewith, each
said means being connected to each said input and causing the
amplitude of each said input in said output sound signal to be
substantially proportional to the cosine of one half the angular
difference between the output position angle and the angle assigned
to the respective A or B input, the angles assigned to said A and B
inputs differing by approximately 180 degrees.
8. Apparatus as claimed in claim 7 wherein said first, second and
third means for generating output signals cause at least one of
said A and B channel signals in one of said outputs to be opposite
to its polarity in at least one other of said outputs and the same
as its polarity in at least another of said outputs.
9. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels four directional output sound
signals correlated with the input signals, decoder apparatus
comprising an A input and a B input, first, second, third and
fourth means for generating first, second, third and fourth
directional sound output signals each having a respective position
angle associated therewith, at least three of said means being
connected to each said input and causing the amplitude of each said
input in said output sound signal to be substantially proportional
to the cosine of one half the angular difference between the output
position angle and the angle assigned to the respective A or B
input, the angles assigned to said A and B inputs differing by
approximately 180 degrees.
10. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels four directional output sound
signals correlated with the input signals, decoder apparatus
comprising an A input and a B input, first, second, third and
fourth means for generating first, second, third and fourth
directional sound output signals each having a respective position
angle associated therewith, each said means being connected to each
said input and causing the amplitude of each said input in said
output sound signal to be substantially proportional to the cosine
of one half the angular difference between the output position
angle and the angle associated to the respective A or B input, the
angles assigned to said A and B inputs differing by approximately
180 degrees.
11. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels four directional output sound
signals correlated with the input signals, decoder apparatus
comprising an A input and a B input, first, second, third and
fourth means for generating first, second, third and fourth
directional sound output signals, each having a respective position
angle associated therewith, each said means being connected to each
said input and causing the amplitude of each said input in said
output sound signal to be substantially proportional to a function
of the angular difference between the output position angle and the
angle assigned to the respective A or B input.
12. In a multidirectional sound system for encoding at least four
directional input sound signals on A and B audio channels and
reproducing from the A and B channels at least four directional
output sound signals correlated with the input signals, decoder
apparatus comprising an A input and a B input, at least four
directional sound signal outputs, first, second, third and fourth
means coupled to said A and B inputs for generating first, second,
third and fourth directional sound output signals each having a
respective position angle associated therewith, each said means
causing the amplitude of each said input in said output sound
signal to be substantially proportional to the cosine of one half
the angular difference between the output position angle and the
angle assigned to the respective A or B input, the angles assigned
to said A and B inputs differing by approximately 180.degree. .
13. Apparatus as claimed in claim 12 wherein said first, second,
third and fourth means for generating output signals cause at least
one of said A and B channel signals in one of said outputs to be
opposite to its polarity in at least one other of said outputs and
the same as its polarity in at least another of said outputs.
14. In a multidirectional sound system for encoding at least three
directional input sound signals on A and B audio channels and
reproducing from the A and B channels four directional output sound
signals correlated with the input signals, decoder apparatus
comprising an A input and a B input, first, second, third and
fourth means coupled to said A and B inputs for generating first,
second, third and fourth directional sound output signals each
having a respective position angle associated therewith, each said
means causing the amplitude of each said input in said output sound
signal to be substantially proportional to a function of the
angular difference between the output position angle and the angle
assigned to the respective A or B input.
Description
Other objects and advantages will be apparent upon consideration of
the following description in conjunction with the appended drawings
in which:
FIG. 1 is a diagram showing a typical speaker placement in
relationship to a listener and certain desirable phase or polarity
relationships for apparatus according to the invention;
FIG. 2 is a schematic diagram of an encoder section of apparatus
according to the present invention;
FIG. 3 is a schematic diagram of a decoder and reproduction section
of apparatus according to the present invention;
FIG. 4 is a diagram useful in explaining directionality effects
obtained by a typical system in accordance with the invention;
FIG. 5 is a diagram useful in explaining the subjective effect of
reproduction on conventional stereo reproduction equipment of four
direction sound signals produced in accordance with the
invention;
FIG. 6 illustrates the effect of reproduction on monaural
reproduction equipment of four direction sound signals produced in
accordance with the invention;
FIG. 7 is a diagram illustrating the directional effect obtainable
with an alternative form of system in accordance with the present
invention;
FIG. 8 is a diagram illustrating the directional effect of a
further alternative system in accordance with the present
invention;
FIG. 9 is a schematic diagram of a gain control arrangement for a
four-directional system which may be adapted to decoding systems in
accordance with the invention to produce greater localization of
sound from individual ones of the four output loudspeakers;
FIG. 10 is a schematic diagram of an alternative form of gain
control for a four-directional sound system in accordance with the
invention;
FIG. 11 is a schematic diagram of a gain control signal generator
for providing a low frequency gain control signal in recorded or
transmitted material to properly control the gain in apparatus as
illustrated in FIG. 12; and
FIG. 12 is a schematic diagram of a gain control system adaptable
for use in conjunction with the described four-direction sound
system wherein the localization of sound output to various
loudspeakers is controlled by a low frequency signal.
AMPLITUDES AND POLARITIES FOR ENCODING AND DECODING
As previously stated, the present system takes three or more
directional sound input signals and combines them into two
conventional audio information channels (such as utilized in stereo
recording or broadcasting) in a manner to impart multidirectional
sound direction information.
According to the invention, there are essentially two parameters to
be considered in processing the multidirectional sound signals.
These two parameters are phase and amplitude. It is convenient to
limit consideration of phase relationships to only two different
discrete phase relationships, namely zero degrees (in phase) and
180.degree. (out of phase). These phase relations may also be
treated as a simple reversal in polarity or sign.
Considering first the amplitude relationships, it will be noted
that one wishes to have a formula for determining the amplitude
with which a particular input sound signal is to be supplied to the
A channel (left channel) of a conventional stereo recording or
transmission system and the amplitude with which such signal is to
be supplied to the B channel (right channel) of the system.
It will be seen that the appropriate amplitude relation for the A
and B channels can be determined as a function of the direction
assigned to the particular input signal. In this discussion, the
direction assigned to the input signal will be identified by an
angle x.sub.1 for signal No. 1, x.sub.2 for signal No. 2, and so
on.
The angle of the position directly to the right of the listener is
arbitrarily assigned the value of zero degrees and the angular
progression is counterclockwise. Hence the positions, illustrated
in FIG. 1 for example, of right front, left front, left rear and
right rear would have angular position designations of 45.degree.,
135.degree., 225.degree. and 315.degree. respectively. See also
FIG. 4.
Using this convention, the amplitude of an input signal supplied to
the A channel is equal to the amplitude of the input signal
multiplied by the sine of one half the position angle. The equation
for A signal amplitudes is given in Equation 1 of the Appendix
hereinafter.
The amplitude of the signal supplied to the B channel is equal to
the amplitude of the input signal multiplied by the cosine of one
half the position angle. The equation for the B channel signal
amplitudes is given in Equation 2 in the Appendix.
It should be noted that the terms in Equations 1 and 2 all have a
positive sign, but that this is not intended to indicate the
polarity or phase of the signals. The polarity is determined in
accordance with the quadrant in which the direction angle lies as
explained hereinafter.
It is also necessary, of course, to define the appropriate decoding
process to produce output signals correlating with the three or
more input signals and having a desired directional characteristic.
The angular position convention for decoding is the same as for
encoding. Each output signal g.sub.1, g.sub.2 etc. has an amplitude
equal to the A channel amplitude times the sine of one half the
position angle for the output signal plus the amplitude of the B
channel times the cosine of one half the position angle for the
output signal. The decoding equation is given as Equation 3 in the
Appendix. As before, the polarity or sign for the signals is
determined by the quadrant in which the position angle lies rather
than by Equation 3.
An important aspect of the system is the manner in which the phase
or polarity relations for different angular position designations
of output and input are related to the A and B channel signal
polarities. In order to gain some intuitive understanding of the
encoding and decoding criteria described above, it is useful to
note the operation of the system for a particular input signal.
Consider an input signal which is assigned the angular position of
left front or 135.degree.. It can be shown that in accordance with
the above encoding and decoding criteria, and assuming that the
output loudspeakers have angular positions of 45.degree. ,
135.degree., 225.degree. and 315.degree., the output correlating to
the hypothetical input signal will be predominantly from the left
front or 135.degree. loudspeaker.
There will be no output from the right rear speaker and there will
be some output from the right front speaker and the left rear
speaker. In each of the latter two speakers, the power output will
be one half the power (0.7 times amplitude) of the left front
speaker level. The phase or polarity of the signal from the right
front speaker and from the left front speaker are desired to be the
same as are the polarities of the left front and left rear
speakers. This is indicated in FIG. 1 by the notation "in" on the
lines joining the left rear and left front and the right front and
left front speakers.
The use of positive and negative polarities in the system requires
that two adjacent speakers, at least, must be of opposite polarity
or out of phase. According to the preferred embodiment of the
invention, only the two rear speakers have this out-of-phase
condition, this being most compatible with the prevalent situation
in which the principal subject material rarely originates
predominantly from the rear speakers.
Another important consideration is that in standardized FM
stereo-radio broadcasting, a monaural receiver reproduces a signal
for the listener corresponding to the sum of the A channel and the
B channel in equal amplitude. Referring to the decoding
relationship, it will be seen that the position angles which yield
equal A and B amplitudes are 90.degree. and 270.degree.. One
desires that the monaural radio receiver have a position angle
corresponding to front center or 90.degree.. Hence, the A channel
and B channel must be combined with the same polarity or in phase
for the front center or 90.degree. position. This also means that a
four-direction broadcast received on a stereo FM receiver would
produce left front sounds and right front sounds from the left and
right speakers respectively which are in phase (of the same
polarity). This is obviously the desired situation. In other words,
for one-direction and two-direction playback compatibility, the
present invention provides that the front center direction be
represented by A equal to B and of the same polarity, that is, the
position angle x equals 90.degree. and A is at the left and B is at
the right.
In reproduction of four-directional sound, the most important
direction is obviously front center. Due to the predetermined
arrangement of speakers in the four corners, the sound image for a
front center directional sound input is necessarily a ghost image
between the front pair of speakers. It is important that in
addition to the front speakers having equal amplitude for this
situation, that they be in phase. Such a situation prevails in FIG.
1. It is also desirable that this front center sound input, to the
extent that it is reproduced in the rear speakers, be reproduced in
the same polarity or in phase. It will be seen in FIG. 1 that this
situation also prevails.
The situation with respect to those images existing between a front
speaker and a corresponding rear speaker is not so critical but it
is desirable that any out-of-phase or opposite polarity
reproduction of such signal be in an opposite rear speaker and that
is also shown in FIG. 1.
As previously mentioned, one wishes to have any ghost image
existing between two speakers produced in phase in each of the
flanking speakers and this situation is maintained to the maximum
extent possible as shown in FIG. 1 (excepting only the ghost images
in the rear quadrant).
It will be seen that a particular assignment of polarities or phase
relationships for the encoding and decoding formulas must be
observed to achieve the desirable reproduction relationships
depicted in FIG. 1. Referring to Equations 1, 2 and 3, the proper
polarities can be summarized as follows. All sine terms should be
assigned a positive value for left front, right front and left rear
quadrants, and a negative value for the right rear quadrant. All
cosine terms should be assigned a positive value for the left
front, right front and right rear quadrants and a negative value
for the left rear quadrant. The foregoing polarities or signs are
in lieu of the sign for the trigonometric function for the
particular quadrant. It is of interest that the difference between
the assigned value and the trigonometric values from zero through
360.degree. is only different in the fourth quadrant, i.e. the
right rear. If one assigns the right rear quadrant angle values of
from minus 90.degree. to zero rather than 270.degree. to
360.degree., then the algebraic sign of the trigonometric functions
would correspond to the desired values as derived from the analysis
illustrated in FIG. 1.
ENCODER APPARATUS
Suitable apparatus for encoding in accordance with the system
herein described is shown in FIG. 2. It must be noted that the
particular form the apparatus takes is subject to great variation.
As will be seen from Equations 1 through 3, the operation to be
performed is quite simple in that it involves multiplying
respective audio frequency inputs by predetermined constants and
adding or subtracting the products so obtained to derive an encoded
A channel or a B channel signal. Numerous forms of conventional and
readily available electronic analog computer circuits or components
may be utilized to perform this operation.
Referring to FIG. 2, multidirectional signals f.sub.1, f.sub.2,
f.sub.3 and f.sub.4 are obtained from a multisignal source 11. The
multisignal source would typically consist of a multiple track tape
recording produced at a recording session with various microphones
or other audio input devices representing sound input signals to be
assigned angular position values for multidirectional sound
reproduction. The angular position value assigned to a particular
recorded sound signal may correspond to an actual direction from a
listener position in an actual recording session. The assigned
angular position value for a particular sound signal can, however,
just as well be purely arbitrarily assigned to produce a desired
directional sound effect.
A schematic circuit diagram for the encoder is shown within the
dashed outline box 12.
The encoder circuit is shown with four inputs for separate and
distinct sound input signals but it should be noted that more
inputs may be provided, and the number of inputs is not determined
by the number of output loudspeakers (which would normally be four
in number). It will be noted that with more than four inputs, two
or more inputs may be assigned to the same quadrant or to the same
position angle resulting in the encoder serving the function of a
mixing and sound effect control apparatus as well as an encoder.
One sound signal may also be supplied to two encoder inputs for
specific effects.
The function of the encoder of FIG. 2 is to process the signals
f.sub.1, f.sub.2, f.sub.3 and f.sub.4 in accordance with Equations
1 and 2 to arrive at output signals A and B.
A pair of operational amplifiers 13 and 14 together with
appropriate input resistors, feedback resistors and other resistors
utilized in a well-known manner are employed to derive the output
signals A and B. The amplifiers 13 and 14 may, for example, be
Philbrick-Nexus 1011 amplifiers which have the advantage of being
capable of driving 600 ohm characteristic impedance output lines
directly. Numerous other amplifiers may also be used, noting of
course that they must have suitably high gain over the full audio
frequency band for which the encoder is intended. Fifteen cycles to
fifteen thousand cycles would normally be ample frequency
coverage.
Resistors R1 through R8 in FIG. 2 are input resistors, the
resistance values of which may conveniently be utilized to
determine the position angle which is to be assigned to each of the
four multidirectional input signals. The circuit illustrated in
FIG. 2 is intentionally selected so that it is not limited to a
prescribed set of position angles for the respective input signals.
Rather, a formula has been derived which permits the position
angles to be set at different values over a wide range. Since
certain position angles are associated with negative polarity and
other position angles are associated with positive polarity for the
sine and cosine terms of Equations 1 and 2, a limit is imposed to
some extent on the assignment of position angles to the respective
inputs. Accordingly, signal f.sub.1 and signal f.sub.2 may be
located anywhere in the first or second quadrants. Signal f.sub.3
may be located anywhere in the third quadrant, and signal f.sub.4
may be located anywhere in the fourth quadrant.
It may also be noted that for circuit economy, the amplifiers 13
and 14 are provided with three inputs to their negative input
terminal (not counting the feedback) and only a single input to the
positive input terminal. One could also simply provide all inputs
to the amplifier to the one negative input terminal and utilize an
inverter amplifier in series in any of the inputs which one desired
to provide with an opposite polarity.
In the circuit of FIG. 2, resistors R9 and R11 are feedback
resistors, resistors R10 and R12 are ground resistors, and
resistors R13 and R14 are trimming resistors. Resistors R15 and R16
are output resistors.
In accordance with well-known analog computer techniques, the
values of resistors R9 through R12 are selected to make circuit
values fall within a convenient range and to make input impedances
sufficiently high to avoid loading associated circuits. These
values would usually be in the range between ten thousand and
several hundred thousand ohms.
The values of resistors R13 and R14 are selected in accordance with
operational amplifier manufacturer instructions to null DC offset.
Resistors R15 and R16 are small isolating resistors to prevent
loading from effecting operational amplifier stability and may be
of the order of thirty ohms. The values of resistors R1 through R8
depend upon the angular position assigned to corresponding inputs
as set forth in Equations 4 through 11 in the Appendix.
As a specific example of an encoder circuit, one may utilize a
basic configuration of input position angles as illustrated in FIG.
4. As seen from FIG. 4, input f.sub.1 is at 135.degree., input
f.sub.2 is at 45.degree., input f.sub.3 is at 225.degree., and
input f.sub.4 is at 315.degree.. The equations for the amplitudes
of the A channel and B channel signals are given in Equations 12
and 13, and the values for resistors R1 through R16 are as
follows:
R1 = 108.2k
R2 = 261.3k
R3 = 108.2k
R4 = 422.1k
R5 = 261.3k
R6 = 108.2k
R7 = 108.2k
R8 = 422.1k
R9 = 100.0k
R10 = 50.0k
R11 = 100.0k
R12 = 50.0k
R13,r14 are selected to cancel amplifier DC offset
R15,r16 = 30 ohms
It may be noted that if all (four or more) inputs to each
operational amplifier are to the negative terminal and inverters
are provided for inputs desired to be of different sign, then
simple relations (such as for R1 and R5) prevail for all input
resistors and the two resistors associated with each input have
values solely determined by that input's position angle. Variable
resistors can thus be employed which are calibrated to provide any
desired position angle (at least over one quadrant, 90.degree.) for
respective inputs.
DECODER APPARATUS
FIG. 3 illustrates an exemplary decoder schematic circuit diagram
as part of the playback portion of the overall system. Similarly to
the decoder schematic circuit diagram, the circuit illustrated in
FIG. 3 is one way to implement Equation 3. Other known electronic
analog computer circuits could also be used within the scope of the
invention.
In FIG. 3, an encoder material source produces two electrical
signal outputs representing an A channel and a B channel. The
encoder material source could, for example, be a conventional
stereo record player playing a record of material encoded by the
apparatus of FIG. 2. Alternatively, the encoder material source
could be in stereo FM radio receiver receiving encoded material
from a transmitter that had been either encoded from a live
broadcast or had been encoded on a two-channel tape or disc record
for playback over the FM stereo transmitter. In any event, the
audio signal channels A and B will typically be standard stereo
transmission channels.
The A and B outputs of the encoder material source are the sole
inputs to the decoder circuit contained within the dashed box 22 in
FIG. 3.
A plurality of operational amplifiers 23, 24, 25 and 26 are
provided which respectively generate four directional outputs for
four directional loudspeakers.
The A channel and B channel inputs are supplied to the operational
amplifiers with a particular amplitude ratio and polarity
relationship determined by the position angle assigned to the
particular operational amplifier and its associated
loudspeaker.
Input resistors R21 and R22 determine the amplitude ratios of the
channel A signal and channel B signal supplied to amplifier 23,
resistors R23 and R24 serve this function with respect to amplifier
24, resistors R25 and R26 serve this function with respect to
amplifier 25, and resistors R28 and R27 serve this function with
respect to amplifier 26.
R29, R30, R31 and R33 are feedback resistors. R32 and R34 are
ground resistors and R35, R36, R37 and R38 are trim resistors, all
selected and used in accordance with known electronic analog
computer techniques.
Outputs g.sub.1, g.sub.2, g.sub.3 and g.sub.4 from amplifiers 23,
24, 25 and 26, respectively, are each supplied to the corresponding
one of four power amplifiers 27, 28, 29 and 30.
Power amplifiers 27 through 30 feed respective loudspeakers 31
through 34. Loudspeakers 31 through 34 are, of course, arranged for
directional sound effects, for example as illustrated in FIG.
1.
Amplifiers 27 through 30 and loudspeakers 31 through 34 may be of
conventional form. While illustrated as a single loudspeaker, each
of the loudspeakers 31 through 34 may comprise a loudspeaker system
and enclosure for improved audio reproduction. Similarly,
amplifiers 27 through 30 may have controls, indicators and other
features normally associated with audio power amplifiers.
It will be noted that the inputs to amplifier 23 from channel A and
channel B are of the same polarity as is also the case with
amplifier 24. On the other hand, the B input to amplifier 25 and
the A input to amplifier 26 are of the opposite polarity.
With the specific polarity arrangement illustrated, amplifier 23
has an output which may be assigned a position angle anywhere in
the left front or right front quadrant, and the same is true of
amplifier 24. The output of amplifier 25 may be assigned a position
angle anywhere in the left rear quadrant, and the output of
amplifier 26 may be assigned a position angle anywhere in the right
rear quadrant. As previously explained, the position angle assigned
for the operational amplifier output feeding a particular
loudspeaker may or may not correspond to the actual physical
position of the loudspeaker in the listening room.
If the position angles for the loudspeakers 31, 32, 33 and 34 are
desired to correspond to the position angles illustrated in FIG. 1
(and thus to the input position angles illustrated in FIG. 4), the
relative amplitudes of stereo channels A and B in each of the
operational amplifier output signals g.sub.1, g.sub.2, g.sub.3 and
g.sub.4 are readily calculated from Equation 3 in the Appendix, the
results being given in Equations 14 through 17 in the Appendix.
To instrument the Equation 14 through 17 for the FIG. 3 decoder
circuit, the following resistor values (in ohms) are
appropriate:
R21 = 108.2k
R22 = 261.3k
R23 = 261.3k
R24 = 108.2k
R25 = 108.2k
R26 = 201.4k
R27 = 108.2k
R28 = 201.4k
R29, r30, r31 and R33 = 100k
R32 and R34 = 50k
R35, R36, R37 and R38 are selected to cancel amplifier DC offset.
Amplifiers 23, 24, 25 and 26 may be the same as those described
with reference to FIG. 2. Resistors R30 through R34 are selected to
make circuit values fall in a convenient range.
Referring now to the complete system shown and described in FIGS. 2
and 3 where the position angles are as illustrated in FIGS. 1 and
4, the operation can be described in fairly simple terms.
It can be shown that an input signal f.sub.1 appears with greatest
amplitude in the output signal g.sub.1 feeding amplifier 27 and
loudspeaker 31. It also appears at a reduced level in output signal
g.sub.2 and output signal g.sub.3 feeding loudspeakers 32 and 33
respectively. The level of output in speakers 32 and 33 is reduced
by one half in power (3dB) or in amplitude to a level of 0.707.
There is no output in signal g.sub.4 corresponding to an input
signal f.sub.1.
Generalizing from the full power, half power and zero power
relationship described above for the apparatus of FIGS. 1 through
4, it can be stated that output for a given input signal will be a
maximum when the position angle of the output signal is the same as
the position angle of tbe input signal. When the position angle of
the output signal is different (by an angle dx) from the position
angle of the input signal, the output will be reduced by
multiplying its amplitude by a factor equal to the cosine of half
the angle of difference. An expression for the attenuation in
decibels S.sub.dB to which an input signal of a prescribed position
angle is subjected in a particular output signal is given in the
Appendix as Equation 18. This is also the separation in signal
level which is obtainable between two output signals having a
difference in position angle (dx) as prescribed.
It should be noted that while the position angle arrangement for
inputs illustrated in FIG. 4 with inputs at 45.degree.,
135.degree., 225.degree. and 315.degree. is convenient to use as
the basis for a simple explanation, one should consider other
inputs at other position angles to better understand the operation
of the system. An input at 90.degree. or front center is of
especial interest.
A front center input obviously will be fed to loudspeakers 31 and
32 (left front and right front) equally and with slight attenuation
(about 0.7 dB).
A front center input will also appear to a minor extent in outputs
g.sub.3 and g.sub.4 (and in speakers 33 and 34). The level of this
output is quite low, however, being attenuated by approximately 8
dB.
It should be kept in mind that the particular input and output
position angles, presented in FIGS. 1 and 4 as a specific example
and basis for description and explanation, are not the only
possible embodiments nor necessarily the best embodiment for all
purposes. Of course, the four-corner arrangement of loudspeakers as
illustrated in FIG. 1 is a particularly practical one.
MONO AND STEREO COMPATIBILITY
Just as it is useful to consider the possibility of different input
position angles than those of FIG. 4, it is also useful to consider
different output position angles than those illustrated in FIG. 1
and corresponding to FIG. 4.
This is especially useful in considering compatibility with
existing stereophonic and monophonic audio equipment. This existing
equipment will be seen to have a corresponding position angle or
position angles by which one can evaluate the monophonic and
stereophonic reproduction of audio material encoded in accordance
with the invention and particularly in accordance with the
apparatus illustrated in FIG. 2.
Considering first stereophonic reproduction, it is well known, of
course, that in such reproduction the stereo channel A is applied
to a left loudspeaker and the stereo channel B is applied to a
right loudspeaker (in stereo FM broadcast and reception, this
occurs after numerous intermediate steps).
Referring to Equation 3 for decoding in the Appendix, it may
readily be seen that reproduction of the A channel signal alone
without any contribution from the B channel corresponds to a
position angle of 180.degree. (the cosine of one half of
180.degree. is zero). Similarly, the reproduction of the B channel
alone without any contribution from the A channel corresponds to a
position angle of zero degrees (the sine of one half of zero
degrees is equal to zero). From previous descriptions of the
operation of the system, it will then be clear that a conventional
stereo reproduction system will operate to reproduce in its left
loudspeaker the subject matter that would have been reproduced in a
left front and left rear loudspeaker of a four-directional system
in accordance with the invention. Similarly, the right loudspeaker
of a conventional stereo system would reproduce the material which
would have been reproduced in the right front and right rear
loudspeakers of a four-directional system.
Using the same approach to the reproduction of four-directional
material on monaural equipment, for example a monaural FM receiver
receiving a broadcast from a stereo FM broadcast station, it will
be noted that the in-phase and equal combination of the stereo A
channel and B channel reproduced in such a system corresponds to a
90.degree. position angle or a front center position (the sine of
one half of 90.degree. is equal to the cosine of one half of
90.degree.). The effect on the listener of reproduction of
four-directional material on either conventional stereo or
conventional monaural equipment is schematically illustrated in
FIGS. 5 and 6.
Referring to FIG. 5, a pair of speakers 51 and 52 are shown
together with indications of typical sound image locations for
four-directional material received by stereo equipment. Note that
the image for left front four-directional audio material is, in
fact, at left front in the two-speaker reproduction and the right
front image is at right front, and that the left and right
materials are properly balanced. It will be noted that the image
for left front material and for right front material is displaced
inwardly somewhat from the speaker location. This produces no
material adverse effect, however, and is necessary so that the
extreme speaker position available in the two-speaker system be
reserved for the maximum right and left position angles of zero
degrees and 180.degree.. The rear material image from the left
speaker 51 is indicated by a series of small x's. This is a
schematic indication of the fact that the rear or reverberant
channel material will be affected by a slight subjective outward
displacement and spreading effect due to the out-of-phase or
opposite polarity relation of the rear position angle material.
This effect is entirely in keeping with the desired use of these
position angles to create "ambience." Accordingly, the
two-direction reproduction (stereo reproduction) of the
four-direction material produces some emphasis for the front
direction yet does not lose the rear direction material entirely.
Thus reproduction on stereo equipment is nearly exactly what one
would choose to have and provides excellent compatibility.
Referring to FIG. 6, a single speaker 53 is shown representing a
monaural system speaker such as a monaural FM receiver speaker
which might be tuned to a stereo FM broadcast transmitter. In such
circumstances, the sound image obviously can only be in direct line
with the loudspeaker and the relative power levels for various
material will be: left front and right front zero dB and balanced,
left rear and right rear material substantially diminished (-7.6
dB) but still present. Only the material encoded at exactly center
rear (270.degree.) will be eliminated entirely in monaural pickup.
For mono reproduction, very little rear material, if any, is
desired. Mono listening is usually done in less than ideal or even
rather noisy environments, and it is important that primary (front)
direction information should be emphasized as in fact results from
the four-direction material monaural repoduction.
In regard to both monaural and stereo reproduction of the
four-direction material, it should be noted that considerable
control is exercised over the way that the material will be
reproduced in stereo or mono by selection of position angles for
the inputs to the encoder. Being aware of exactly how the material
would be affected in stereo or mono reproduction, one may arrange
that the monaural and stereo reproduction is of excellent quality
as well as the four-directional reproduction, which is of course
the main objective.
ENCODING WITH ALTERNATIVE POSITION ANGLE DESIGNATIONS FOR
INPUTS
FIG. 7 shows alternative direction angle inputs which are
particularly suitable for certain musical material to be encoded in
accordance with the multidirectional system of the present
invention.
In situations such as the concert hall, where the front encoder
inputs will be required to carry the most critical primary program
information, while the rear encoder inputs supply "ambience," or
reverberation, one may wish to enhance the separation between the
front encoder inputs at the expense of reducing the separation for
the rear encoder inputs. This may be accomplished, for example as
shown in FIG. 7, by prescribing a greater difference in position
angle (dx) between the front pair of inputs and less between the
rear pair of inputs.
As shown in FIG. 7, x.sub.1 equal 150.degree., x.sub.2 equal
30.degree., x.sub.3 equal 240.degree. and x.sub.4 equal
300.degree.. The resulting encoding equation for channel A and
channel B is given in the Appendix as Equations 19 and 20. The
encoder resistor values for FIG. 2 which would provide the input
position angles shown in FIG. 7 are given below.
R1 = 103.5k
R2 = 386.4k
R3 = 115.5k
R4 = 259.1k
R5 = 386.4k
R6 = 103.5k
R7 = 115.5k
R8 = 259.1k
R9, r11 = 100.0k
R10, r12 = 50.0k
R13, r14 are selected to cancel amplifier DC offset
R15, r16 = 30 ohms
It should be noted that the use of position angles for inputs as
illustrated in FIG. 7 may or may not involve the use of decoder
output position angles with the same values. Obviously, the decoder
output position angles can likewise be varied without varying the
input position angles.
To obtain output position angles corresponding to FIG. 7, the
following resistor values in ohms may be utilized in FIG. 3.
R31 = 103.5k
R32 = 386.4k
R33 = 386.4k
R34 = 103.5k
R35 = 115.5k
R36 = 136.6k
R37 = 115.5k
R38 = 136.6k
R39,r40,r41,r43 = 100.0k
R42, r44 = 50.0k
R15,r16,r17,r18 are selected to cancel amplifier DC offset
The concept of separating the position angles for either the input
or the output in the system to an angle greater than 90.degree. for
the front input or output signals is, of course, not limited to the
particular position angle designations illustrated in FIG. 7. A
limiting case for such separation is illustrated in FIG. 8 in which
the front position angles have been spread to 180.degree.. The rear
position angles have been set with a negligible separation. The
position angles of FIG. 8, therefore, are for signal 1,
180.degree.; for signal 2, 0.degree.; for signal 3, 269.degree.;
for signal 4, 271.degree.. It is to be noted that signal 3 is still
in the third quadrant and has the polarity relationships designated
for that quadrant, while signal 4 is in the fourth quadrant and has
the polarity relationships designated for that quadrant. If the
extreme case shown in FIG. 8 were used as position angle
designations for a decoder output where the speakers were located
in the four corners of a room, the left front speaker would have
the A channel alone supplied to it, the right front speaker would
have the B channel alone supplied to it, the left rear speaker
would have the A channel minus the B channel at a reduced
amplitude, and the right rear speaker would have the B channel
minus the A channel at a reduced amplitude. Thus the two rear
speakers would have the same content except for being out of phase
with each other. While the arrangement of FIG. 8 would not likely
actually be employed in its exact extreme version, it well
illustrates the manner in which the left front and right front
channel speaker separations can be increased to any desired extent
by designation of appropriate decoder output position angles.
Note that for greatly distorted position angle designations as in
FIG. 8, some adjustment in signal levels in indicated. For example,
the two rear speakers in FIG. 8 are reproducing rear center
material so the volume should be decreased accordingly (or one
speaker omitted).
It should further be noted that notwithstanding the current
preference for placement of loudspeakers in the four corners of a
room or studio so that their relationship to the listeners is left
front, right front, left rear and right rear, it is perfectly
feasible to arrange a four-directional system with speakers
arranged in positions left, right, front and rear or in the centers
of the walls of a listening room or studio. In such an arrangement,
the position angles for both the input and output may correspond
with the loudspeaker location and thus the position angles would be
0.degree., 90.degree., 180.degree.and 270.degree.. The encoding
equations for such an arrangement are given in Equations 21 and 22
in the Appendix, and the decoding equations are given in Equations
23 through 26.
It will be noted that Equations 21 through 26 correspond to
encoding and decoding equations in prior copending Scheiber patent
applications.
It is obvious that numerous circuits other than those illustrated
in FIGS. 2 and 3 can be used to provide the functions of the
encoder and decoder. For example, suitable arrangements of
operational amplifiers or tranformers or combinations thereof may
be used in either or both devices to provide the required
functions. It is possible for a transformer type decoder to be
located preceding or following the power amplifiers. In the latter
case, only two power amplifiers are required for all four channels
which means that existing stereo systems can be adapted merely by
adding the decoder and two speakers at the receiving or playback
station.
GAIN CONTROL APPARATUS
The separation between adjacent speakers provided by the encoder
and decoder embodiments of FIGS. 2 and 3 alone provides the desired
result, i.e. location of a virtual sound source at any place on a
circle around a listener. However, to further emphasize the effect
in respect to highly localized sound sources, it may be desired to
provide substantially unlimited separation between adjacent
speakers for such highly localized sounds. This can be accomplished
in a number of ways, some of which are of particular utility with
the invention and, as such, may be considered to be an improvement
over the basic system of FIGS. 1-4. Four such improvements are
described below with reference to FIGS. 9, 10, 11 and 12.
FIG. 9 is a block diagram of a simplified form of a gain control
circuit which varies the gain of any pair of "diagonal" channels
(e.g. left front and right rear) with respect to the other diagonal
channels. Hereinafter, by diagonal channels is meant the any pair
of channels having an angular difference dx substantially equal to
180.degree. as defined electrically and algebraically with
reference to FIG. 1 but not necessarily corresponding to the
physical placement of the loudspeakers.
Since, in accordance with a basic embodiment of the invention, any
given input will appear in three adjacent speaker channels with
maximum gain in the speaker channel corresponding to the input
channel, the directional effect can be emphasized by decreasing the
gain of the two speakers on either side of the desired speaker. In
the system of FIG. 1, these two diagonal speakers are also placed
in physically diagonal positions. It can also be shown that where
the absolute value of the LF signal is equal to the absolute value
of the RR signal (i.e. the waveforms are identical except for
possibly opposite polarity), the sound source should either be
located at the right front or left rear speaker. Thus, when this
condition exists it is desirable that the gain for the right front
and left rear signals be maximum relative to the gain for the left
front and right rear signals. Similarly, when either the RR or LF
signal is zero or the waveforms in RR and LF are unrelated, the
sound source should be located at the left front or right rear
speaker, in which case the gain for the right front and left rear
should be minimum relative to the gain of the left front and right
rear signals.
The foregoing shows that the separation between the outputs of FIG.
1 (and thus the directional characteristics of the audio output)
can be emphasized by simultaneously varying the gain in each pair
of two diagonal outputs alternatively to controlling the gain in
each of the individual channels. It is also desirable that the gain
in one pair of diagonal channels be accompanied by an appropriate
decrease in the gain of the other diagonal channels. Otherwise, an
increase in gain (for example) to enhance the directional
characteristic of a signal would result in a volume change of the
total audio output as a function of direction. However, by
simultaneously decreasing the power gain in one pair of diagonal
channels (e.g. from 1 to 0) while increasing the power gain in the
other diagonal channels (e.g. from 1 to 2), it is possible to
maintain the total power at the speakers constant, and separation
between adjacent speakers can be increased without changing the
total volume of the system. These functions are performed by the
systems illustrated in FIG. 9.
In FIG. 9, the decoder output is shown at the left. The four
signals LF, RR, RF and LR are coupled to respective variable gain
amplifiers 110LF, RR, RF and LR, which provide the signals for
driving the four speakers as indicated. The LF and RR channels are
also coupled directly through high-pass filters 112LF and 112RR to
absolute value circuits 114LF and 114RR. The absolute value
circuits 114 may comprise full-wave rectifiers the outputs of which
are of the same polarity. These signals, representing the absolute
value of the LF and RR signals, are fed to respective logarithmic
amplifiers 116LF and 116RR, which are well-known devices, providing
output voltages approximately equal to the logarithm of the applied
input voltage over the usable signal voltage range. The outputs of
these amplifiers 116AF and 116BR are coupled through partial
smoothing filters 117A and 117B to the negative and positive
inputs, respectively, of an operational amplifier 118 which
subtracts the two signals providing an output substantially
dependent on log .vertline.LF .vertline. - log .vertline.RR
.vertline. (i.e. log .vertline.LF/RR.vertline. ). This signal is
then fed through another absolute value circuit 120 and an
averaging network 122 (an integrating circuit) to a gain control
generator 123 which controls the gain of the two pairs of variable
gain amplifiers 110LF, 110RR and 110RF, 110LR as a function of the
output of amplifier 118.
As indicated above, where the absolute values of the LF and LR
signals are equal, the gain of amplifiers 110 RF and 110LR should
be maximum and the gain of amplifiers 110LF and 110RR a minimum.
When this condition exists, the output from the operational
amplifier 118 will be equal to zero and the power gain of
amplifiers 110RF and 110LR should be a maximum (e.g. two) while the
gain of amplifiers 110LF and 110RR is a minimum (e.g. zero). At the
other extreme, where either the RR or LF signal is equal to zero or
the waveforms in RR and LF are unrelated, the output of the
amplifier 118 will be a maximum (theoretically infinite but limited
in practice to a definite value, for example, 9 volts). This
maximum voltage causes the gain control generator 123 to provide
output voltages which maximize the gain of amplifiers 110LF and
110RR and minimize the gain of amplifiers 110RF and 110LR.
For conditions between those described above, the gains of the
respective pairs of amplifiers 110 will be appropriately controlled
by generator 123. Mathematically, it can be shown (assuming a FIG.
4 embodiment of the system) that the curve of the required gain
approximates a square root curve to yield constant total acoustical
power output, with the gains in the respective diagonal channels
being equal when the amplitude ratio of LF to RR (of RR to LF) is
about 2.4 and the waveforms are the same. The equations 27 and 28
in the Appendix may be used to determine the (amplitude) gain
control voltages from generator 123 where T is the time constant of
the averaging circuit 122, K is a constant, t is time, V.sub.14 is
the RF and LR control voltage and V.sub.23 is RF-LR control
voltage.
One purpose of the high-pass filters 112A and 112B is to prevent
the passage of low-frequency signals which might otherwise appear
on the inputs to the variable gain amplifiers 110 and possibly
modulate the amplifier inputs. It has further been found desirable
to discriminate against lower frequency signals (at 6 dB per
octave). The filters 112A and 112B also serve this function.
Filters 117A and 117B may have time constants from 100 to 1,000
microseconds and their outputs are in part responsive to the
envelope of their inputs and in part responsive to instantaneous
values. Each type of response is preferred in different situtations
and thus a desirable compromise is achieved by filters 117A and
117B. The averaging circuit 122 should respond to changes in the
output of the amplifier 118 quickly enough so that the ear does not
notice the delay, but not so fast as to pass the actual waveform to
the amplifiers 110. As an example, a 20 millisecond charging rate
has been found satisfactory for practical purposes. to stabilize
the action of the gain control circuits of the decoder, it may be
desirable to mix slightly the LF and RR signals at the encoder
output (or the left and right inputs to the encoder) to minimize
excursions of the LF/RR signal ratio. Conversely, to prevent the
log of this ratio from going to zero, constant phase differences
may be introduced between the respective signals applied to the A
and B channels at the encoder. This has the effect of restricting
gain control action to a relatively narrow range so that the gain
control action will not be audible at the speakers. Such mixing can
be done in proportions which will accomplish the desired result
without materially altering the audio characteristics. Where
extreme channel separation is required, this technique would not be
used.
As noted previously, there are many different ways of controlling
the gain in the respective channels to provide the desired
directional enhancement at the speakers. The embodiment illustrated
and described with reference to FIG. 9 is a relatively inexpensive
way of providing the desired gain.
FIG. 10 shows an alternative gain control arrangement in which the
gain associated with each speaker is determined by a combination of
a gain control element serially connected in the respective speaker
input, and a gain control voltage generator whose output is coupled
to the gain control element. The audio signal in each decoder
output passes through the respective gain control element. Then,
according to the output signal of the control voltage generator,
the signal in the gain control element is either enhanced or
attenuated. When the output signal of the control voltage generator
is at a maximum, the output of the gain control element is at a
maximum, and vice versa.
The gain control elements 203 and 204 are thus controlled by an
output voltage V.sub.1 produced by the control voltage generator
210. The gain control elements 208 and 206 are controlled by an
output voltage V.sub.2 produced by the control voltage generator
212. If desired, separate control voltage generators may be
respectively coupled to each gain control element.
The expressions for each of the control voltages V.sub.1 and
V.sub.2 are dictated by design considerations of the various
control voltage generators which produce these expressions, as well
as by the specific phase, waveform, and level cues present in the
original signals A and B which are to activate the respective
speakers.
For example, a desired acoustical reproduction requires that the
gain associated with the speakers 31 and 34 increases as the ratio
of the intensity levels of the signals g.sub.1 and g.sub.4 diverges
from unity, or their waveforms become increasingly dissimilar. To
achieve this result, the control voltage V.sub.1 applied to the
gain control elements 203 and 204 may be represented by one of
various expressions.
V.sub.1 may be proportional to the average absolute value of the
logarithm of the quotient of the absolute values of g.sub.1 and
g.sub.4. Alternatively, V.sub.1 may be proportional to the average
absolute value of the logarithm of the quotient of the sum and
difference of the absolute values of g.sub.1 and g.sub.4. As a
third alternative, V.sub.1 may be proportional to the average of
the quotient of the sum and difference of the absolute values of
g.sub.1 and g.sub.4. Equations 29-31 in the Appendix represent some
expressions for V.sub.1.In addition to the above or in combination
with the above, similar expressions may be employed in which the
envelopes of g.sub.1 and g.sub.4 are substituted for the
instantaneous signals.
The gain for channels 32 and 33 must obviously be varied in a
complementary manner to that of speakers 31 and 34 so that the
voltage V.sub.2 from control voltage generator 212 may be
proportional to the constant minus the expression for the voltage
V.sub.1. Control voltage V.sub.1 increases as the loudness level
associated with either the g.sub.1 or g.sub.4 signals becomes
stronger with respect to the other, or their waveforms become
increasingly dissimilar.
The gain for speakers 32 and 33, on the other hand, is to increase
as the ratio of the intensity levels of each of the signals g.sub.1
and g.sub.4 approaches unity and as their waveforms become
similar.
GAIN CONTROL SIGNAL APPARATUS
FIGS. 11 and 12 illustrate a further embodiment of a gain control
system employing the basic principles of the invention wherein
subsonic control tones are impressed upon the A and B channels for
the purpose of controlling the gain of the two pairs of diagonal
channels from the decoder 20. In describing the operation of FIGS.
11 and 12, the microphones, speakers, encoder and decoder perform
the same function as previously described and therefore are not
described further.
To facilitate an understanding of this embodiment it is convenient
to refer to power ratios rather than voltage ratios as previously.
The power which can be derived from a given signal is directly
proportional to the square of the voltage level of that signal.
The signal recording means is illustrated in FIG. 11. The outputs
of the right front and left rear microphones 2RF and 2LR are sensed
and coupled to a power-adding circuit 130, while a similar
power-adding circuit 131 sums the power outputs from microphones
2LF and 2RR. These two power-adding circuits are devices which
produce output voltages directly proportional to the total power
which can be derived from the applied input voltages. Their output
voltages are then summed in an adding circuit 132, the output of
which is thus proportional to the total power in the four input
channels.
The outputsof summing circuits 130 and 132 are coupled to a ratio
circuit 134 which, in turn, causes respective A and B modulators
136 and 138 to modulate a 20-cycle (or other subsonic) tone from
oscillator 140. Ratio circuit 134 may be any of a number of
well-known circuits and, for example, may produce a direct output
voltage having an amplitude proportional to the ratio of the
applied input voltages.
Modulators 136 and 138 may be adapted to amplitude-modulate the
20-cycle tone from oscillator 140 with respect to a preselected
level, depending upon the magnitude of the applied control voltage
from the ratio circuit 134. When the A modulator 136 provides a
tone of increased amplitude, the B modulator 138 should be
providing a tone of proportionately decreased amplitude. These
modulated tones are then added to the A and B outputs of encoder 18
to provide the signals which are to be conveyed by the two-channel
transmission path and which, in this particular embodiment, are
indicated as A' and B'.
The receiving end of the system is illustrated in FIG. 12. Two
high-pass filters 142 and 144 are used to separate the audio
control tones from the A and B audio signals on the A' and B'
channels. These A and B signals from the filters 142 and 144 are
coupled to decoder 20 to provide the four output channels described
above.
The control tones from the filters 142 and 144 are coupled to a
gain control generator 146 which controls the gain of variable gain
amplifiers 148RR, RF, LF and LR to increase the gain of one pair of
diagonal channels while appropriately decreasing the gain of the
other pair of diagonal channels. From the preceding discussion of
FIG. 11, it follows that the amplitudes of the control tones will
each be equal to the desired power in their corresponding diagonal
input channels divided by the total power in the system. Each of
these signals varies from a value of zero to one and their sum
should always equal one. Accordingly, since the desired power
ratios (i.e. the power ratios at the microphones) are directly
represented in the control tone signals it is a simple matter for
gain control generator 146 to utilize these known ratios to control
the gain of amplifiers 148RR, LF and 148LR, RF to recreate the same
ratios at the outputs of the amplifiers 148. This will necessarily
enhance the desired signals while deemphasizing these signals which
are not in their corresponding channels. The total power will also
not be varied due to directionality changes. Generator 146 also
serves as a normalizer to maintain the total gain of the four
channels such that the sum of the power in the respective channels
is maintained equal to a constant. This prevents unwanted changes
in the amplitude of the control tone from affecting the volume of
the outputs from the respective loud-speakers.
From the foregoing description of various embodiments of
multidirectional sound systems in accordance with the invention, it
will be seen that a highly effective system is provided capable of
reproducing virtually all essential directional information without
sacrificing fidelity, frequency response or other qualities of the
audio information. Also transmission or recording is possible using
only two conventional stereo channels. It should be appreciated
that the particular apparatus disclosed is not intended to
represent a suitable design for manufacturing economy but is rather
presented for ease of explanation and to show the manner in which
the system can be assembled from well-known existing electronic
analog computer compnents and circuits. In practice, more
economical transistor circuits would be substituted for the
expensive operational amplifier components, the resistors also
would be accorded much more tolerance in resistance values than
indicated in the description, and other practical economies would
be effected.
The gain control features described here are a useful adjunct to
the system, but are not in all cases necessary. Furthermore,
numerous other variations of gain control systems to enhance the
separation between speakers or the localization of sound direction
could be utilized other than the particular ones described here or
in copending patent applications.
More elaborate gain control systems could utilize analysis circuits
similar to the ones described herein but duplicated or triplicated
so that each analysis circuit would serve to analyze a different
frequency band within the overall audio frequency band of the
system.
Gain controls for adjustment of the front to rear power ratio
(rather than the diagonal pair power ratio) are also potentially
useful. Such adjustment can be made on the basis that inequalities
between front and rear power should generally be increased to tend
to restore the power ratios to those of the input signals.
In addition to those variations and modifications to the system
described or suggested herein, numerous other variations and
modificationswill be apparent to those skilled in the art. The
invention is to be understood not to be limited to the specific
illustrations described and rather is to include those variations
and modifications within the ordinary skill of the art.
APPENDIX
1. A=f.sub.1 sin x.sub.1 /2 +f.sub.2 sin x.sub.2 /2 . . . + f.sub.n
sin x.sub.n /2
2. B=f.sub.1 cos x.sub.1 /2 + f.sub.2 cos x.sub.2 /2 . . . +
f.sub.n cos x.sub.n /2
3. g.sub.n =A sin x.sub.n /2 + B cos x.sub.n /2
4. R1 = [R9/sin (x.sub.1 / 2)]
5. R2 = [R9/sin (x.sub.2/ 2)]
6. R3 = [R9/sin (x.sub.3 /2 )]
7. R4 = [{ 1 + sin (x.sub.1 /2 ) + sin (x.sub.2 /2 ) + sin (x.sub.3
/2 ) - sin (x.sub.4 / 2) } .sub.R10 /sin(x.sub.4 /2)]
8. R5 = [R11/cos(x.sub.1 /2 )]
9. R6 = [R11/cos(x.sub.2 /2 )]
10. R7 = [R11/-cos(x.sub.4 /2 )]
11. R8 = [{ 1 + cos (x.sub.1 /2 ) + cos (x.sub.2 /2 ) + cos
(x.sub.3 /2) - cos (x.sub.4 /2 )} .sub.R12 /-cos (x.sub.3 /2 )]
12. A = 0.9239f.sub.1 + 0.3827f.sub.2 + 0.9239f.sub.3 -
0.3827f.sub.4
13. B = 0.3827f.sub.1 + 0.9239f.sub.2 - 0.3827f.sub.3 +
0.9239f.sub.4
14. g1 = 0.9239A + 0.3827B
15. g2 = 0.3827A + 0.9239B
16. g3 = 0.9239A - 0.3827B
17. g4 = -0.3827A + 0.9239B
18. S.sub.dB = 20 log.sub.10 cos (dx/2)
19. A = 0.9659f.sub.1 + 0.2588f.sub.2 + 0.8660f.sub.3 -
0.5000f.sub.4
20. B = 0.2588f.sub.1 + 0.9659f.sub.2 - 0.5000f.sub.3 +
0.8660f.sub.4
21. A = f.sub.1 + 0.707f.sub.3 + 0.707f.sub.4
22. B = f.sub.2 + 0.707f.sub.3 - 0.707f.sub.4
23. g1 = A
24. g2 = B
25. g3 = 0.707A + 0.707B
26. g4 = 0.707A - 0.707B ##SPC1##
29. V.sub.1 = K .vertline. log (G.sub.1 / G.sub.4) .vertline.
g.sub.1 = .vertline.g.sub.1 .vertline.; G.sub.4 = .vertline.g.sub.4
.vertline.
30. V.sub.1 = K.vertline.log (G.sub.1 -G.sub.4 / G.sub.1
+G.sub.4).vertline .
31. V.sub.1 = K.vertline.G.sub.1 -G.sub.4/ G.sub.1 +G.sub.4
.vertline.
* * * * *