6
Channel sphere surround
Recent times, multi channel
(surround) sound recording is not unusual. It is said that reproducibility of
the sound field is satisfactory.
Really I try hearing, but I feel there are no ceiling and no floor. (I cannot understand, whether living in the room or outdoors). I think it is the unnatural playback sound. Depending on the channel of the top and bottom added (6 channel sphere surround), the solution was tried.@
Here, problem is the
expedient. Regarding to the multi channel systems generally,
a) The equipment and material which can handle the multi channels.
b) The compatibility which utilizes sound source property.
c) Arrangement, and interference of speaker.
And
so on it is the case that it is problem, but this time, the result which was
examined, until recently the equipment and material and being diversion
possible, it made the matrix system sound source property it is many, a) b)
probably can supply fault sufficiently.
If it is matrix system,
reappearance and to be cheapness and easy, until now, objective appraisal to be
little, it is the reason which also value of experiment being large
selects.
As
for encoding system of the matrix, CBS SONY SQ4ch system with the Sansui QS4ch
system , in addition, the rear exists monaural (3 ch) Dolby system, and so
on various systems.
The
among these, normal position possibility, not be located on 4 speaker centers,
by the fact that it expands, the QS4ch which has the possibility where it can
handle the only top and bottom sound field information was designated as
transmission system of 6 channel system.
I produced a decoder with
the QS4ch/SQ4ch/QS6ch logic this time and evaluated those methods and a
difference of having logic or not.
The commentary of the decoder with the
QS4ch/SQ4ch/QS6ch logic
Each encode / decode
expression are as follows.
E The encode
calculating formula of the QS4ch method
L = 0.92 * LF { 0.38 * RF { 0.92 * LR * i { 0.38 * RR * i
R = 0.38 * LF { 0.92 * RF | 0.38 * LR * i | 0.92 * RR * i
E The decode calculating
formula of the QS4ch method
LF = @L { 0.414 * R
RF = @R { 0.414 * L
LR = |L * i { 0.414 * R * i
RR
= R * i | 0.414 * L * i
E The encode
calculating formula of the SQ4ch method
L = LF | 0.707 * LR * i { 0.707 * RR
R = RF | 0.707 * LR { 0.707 * RR * i
E The decode
calculating formula of the SQ4ch method
LF = L
RF = R
LR = 0.707 * L * i | 0.707 * R
RR = 0.707 * L | 0.707 * R * i
The
method that I adopted
LF
= @L
RF
= @R
LR
= |0.5 * L * ( 1|i ) | 0.5 * R * ( 1{i )
RR
= @0.5 * L * ( 1{i ) { 0.5 * R * ( 1|i )
E The encode
calculating formula of QS6ch methods
L = 0.92 * LF { 0.38 * RF { 0.92 * LR * i { 0.38 * RR * i { 0.65 * ( 1{i ) * UP { 0.65 * ( 1|i ) * DN
R = 0.38 * LF { 0.92 * RF | 0.38 * LR * i | 0.92 * RR * i { 0.65 * ( 1|i ) * UP { 0.65 * ( 1{i ) * DN
E The decode
calculating formula of QS6ch methods
LF = @L { 0.414 * R
RF
= @R { 0.414 * L
LR
= |L * i { 0.414 * R
* i
RR
= @R * i | 0.414 * L
* i
UP
= |0.707 * L * i { 0.707 * R
DN
= 0.707 * L | 0.707 * R * i
The
method that I adopted
LF
= @L {
0.414 * R
RF
= @R {
0.414 * L
LR
= |L * i {
0.414 * R * i
RR
= @R * i | 0.414
* L * i
UP
= 0.5 * L * ( 1|i ) { 0.5 *
R * ( 1{i )
DN = 0.5 * L * ( 1{i ) {
0.5 * R * ( 1|i )
@@
EEqualizer circuit
The RIAA equalizer circuit
is a Kaneda-style initial version (simple). @The sorting of first grade
FET is required.
EMatrix decode circuit
EQS4ch
method@F @
Add
with@a factor of 1 / γ 2 (0.414) between the forward channel. Subtract with@ a factor
of 1 / γ 2 (0.414) between the rear channel. And, each +90 degrees,-90 degrees
phase-shifted rear channel. In addition, localization of both lateral side will
be improved by the phase-shifting each +60 degrees, -60 degrees.
ESQ4ch method@F
If
operate at between 0 and 90 degrees as the standard, to decode the backward
channel, phase will be asymmetric. In order to prevent this, calculate with
each -45 degrees and +45 degrees.
EQS6ch
method@F
Left
and right channels of the front and rear is QS4ch system. Up and down channel
is rear channel of SQ4ch system.@
ESound source direction distinction circuit
EQS4ch method@F
It
is direction distinction from the front left and the backward right volume
ratio, the front right and the backward left volume ratio.
ESQ4ch method@F
It
is direction distinction from the volume ratio of front right and left, the
volume ratio of the rear right and left and the volume ratio of center front
and rear.
EQS6ch method@F
It is direction
distinction from the front left and the backward right volume ratio, the front
right and the backward left volume ratio, the volume ratio of the top and
bottom.
As for the volume ratio, is
Diode the band pass of two signals ; after logarithm conversion, both waves
rectification for a difference of both. @In addition, signal is summed before
taking the ratio, but that the symmetry is lost, so the separation is worse.
(Study is required because becomes asymmetrical when summing)@By the addition
between two signals and the minute noise addition, I prevent false distinction
in the minute signal beforehand.
EDirectionality emphasis circuit
EQS4ch logic method@F
I add a crosstalk
ingredient occurring to the channel of both sides of the sound source direction
as a reverse correlative person in charge and cancel it.
ESQ4ch logic method@F
A sound source
does 90 degrees phase shifting of the crosstalk ingredient of the backward
right and left channel in the case of front right and left channel, the front in
the case of the rear and I add it as a reverse correlative person in charge and
cancel it.
EQS6ch logic method@F
I add the front
left and the backward right, the front right and the backward left crosstalk
ingredient as a reverse correlative person in charge, and a sound source
cancels it in the case of the top and bottom. In addition, I add the crosstalk ingredient of the top and
bottom as a reverse correlative person in charge, and a sound source cancels it
in the case of circumference. iBut,
as for this QS6ch logic, a sound source does not cancel the crosstalk with a
simple model at the time of truth right truth leftj
Comparison
of each method
An audition room
is Western-style room 10 tatamis. The speaker is FE87 of Shinano. The all
around speaker is height of the room corner ear.
The sound source is a sound.
The hearing of the sound image when I let circumference (zero degree, +-22.5
degrees, +-45 degrees, +-90 degrees, +-112.5 degrees, +-135 degrees, 180
degrees) do normal position with a bread pot. And it is the level of each
method and crosstalk, a calculation value of the phase (counterclockwise
direction gets worse).
The
audition comparison of four channel methods
EDiscrete method
@I hear the rear approaches forward
when turning to the front, and right and left do not do normal position at all.
It is clogged up and understands that I cannot recognize a normal position
position definitely (at a limit of the human ability) even if it is a discrete
method pro-temporary law, ideal.
EQS4ch method
@Case without the logic, rear sound
field approach forward, and the sound image of the just beside moves forward,
and normal position (the rear in particular) becomes sweet. Even if I get
narrow, the sound field becomes it and similar sound field / normal position in
comparison with the discrete method. It is right and left asymmetry, but does
not understand it only for the feeling that I heard closely.
@When there was a logic, the separation improvement of
the corner was mainly seen. However, unlike the SQ method, which will be
described later, because there is a practical separation performance even
without logic in the QS system, there is no significant difference due to the
presence or absence of logic.
ESQ4ch (symmetric improvement) method
@Without logic, in the standard
system, therefore topologically asymmetrical, front center of the sound image
will move to the right degree 10. Since have a scheme in consideration of
symmetry time, improvement in left-right symmetry was seen. Hardly separating
the front and rear, the rear of the sound image is also very unclear, the
reproducibility of the sound field can say that very low. In addition, since
there is very strong phase relationship between the channels, will not be able
to recognize the placement of the instrument by moving slightly the position of
the head.
@If the logic is there, dramatic separation
improvement before and after the corner was seen. Either forward or backward,
the only direction the sound is small is monaural. Compared with QS4ch logic
system, sound image will hear it slightly stable. However, the sound source is
not a single direction, if that can not be direction discrimination, separation
is significantly worse. Placement of the instruments it will not be able to
recognize most. Therefore,
also be equipped with a logic in SQ method,
it will not be able to play properly when it is recorded
in the way.
@Symmetrical improvement method can improve localization
of the forward and backward. However, it can be seen that the asymmetry between
the front and rear of the left and right to increase, it was found that there
is a problem. (Since the original asymmetry of the front can be relaxed by the
logic) Future, symmetrical improvement scheme will consider in the direction to
cancel.
EQS6ch method
@When there
is no logic, crosstalk increases with increase of the sound source, the sound
field will be narrowed considerably. But was able to get up and down the effect
of sound field. (However, separation is better than SQ system) Front and back
do not do a reverse aspect reason, normal position of the sound of the bottom.
When there was a logic, the
separation improvement of the corner, improvement of the separation of the top
and bottom were seen. The sound of
the bottom does normal position, too. However, 6ch system discontinuity exists
in the space sphere. Lower - +45 degree phase shifter and rear + -90 degree phase
shifter is not connected. Therefore, there is a basic problem of noise can not
be avoided by the sound that you want to move, and measures must be taken in
the future, and challenge was remaining practical..
Future
problem, a summary
@By a natural audio and past
recording material and compatibility were high, and it was inexpensive, and, by
this experiment, a beginning of the many channels reproducing system
realization that I could build was seen. I repeat experiments such as the
improvement of the circuit, the straight record and want to perform further
inspection in future.