The present invention relates to audio signal processing and, in particular, to an apparatus
and a method for edge fading amplitude panning for 3D loudspeaker setups.
After the progression from stereo to 5.1 surround sound, the move towards 3D audio can
be regarded as the next step in the evolution of movie and home cinema sound systems.
A greater number of loudspeakers can extend the listening area and improve the spatial
resolution of the reproduced sound field. However, a greater number of loudspeakers also
means a greater demand, because more loudspeakers need to be placed where they are
supposed to be. In a domestic environment like a living room it can be difficult to place
them according to the specification. In practice, the placement and the number of involved
loudspeakers is a compromise between sound quality, costs, aesthetics, spatial
limitations, and also domestic/social aspects (see [20]).
Object-based audio scenes do not require a specific loudspeaker configuration like
channel-based content and thus have less demands on the placement of the
loudspeakers. The rendering process involves a panning method where the object's
sound signal is played back by more than one loudspeaker (see [7]).
According to the prior art, for creating auditory events between the loudspeakers of a 3D
speaker setup, Vector Base Amplitude Panning (VBAP) is a widely used method, which
can be regarded as an extension to the tan-law (see [17], [5]). While this approach has
proven its suitability for daily use, it is not idea! in all situations.
In the following VBAP is briefly described. VBAP uses a set of N unit vectors l.v
which point at the loudspeakers of the 3D speaker set. A panning direction given by a
Cartesian unit vector p is defined by a linear combination of those loudspeaker vectors
according to formula (1):
where g„ denotes the scaling factor that is applied to ln. In 3 , a vector space is formed
by 3 vector bases.
Formula (1) can generally be solved by a matrix inversion, if the number of active
speakers and thus the number of non-zero scaling factors is limited to 3 . Practically, this is
done by defining a mesh of triangles between the loudspeakers and by choosing those
triplets for the area in between. This leads to the solution
-i n n = h P (2)
where {n-„ n2, n denotes the active loudspeaker triplet.
Finally, a normalization that ensures power normalized output signals results in the final
panning gains a . . . , aN
a - n
| | [ 1 . . n ]T
3
VBAP exhibits particular properties. The vector arithmetic based concepts of VBAP are in
relation to the sound field which is created by the involved loudspeakers. The base vector
that corresponds to a certain loudspeaker, e.g., Gerzon's velocity vector (see [9]),
coincides with the particle velocity that can be measured under free field conditions at the
listener position. A linear combination of the sound fields created by two or more
loudspeakers results in the linear combination of the particle velocity.
VBAP reproduces under free field conditions the particle velocity at the sweet spot that
results from a sound source at the panning position.
As the human auditory system senses the sound pressure instead of the particle velocity
(see [4]) and further involves directional filtering and cognitive processes, there is actually
no direct relation between the underlying vector arithmetic and human localization.
However, sum localization works fairly well for small angles between horizontally arranged
loudspeakers in the frontal or rear area [6] For angles significantly larger than 90°,
loudspeakers at the side, or vertically arranged loudspeaker positions, the sum
localization is less convincing (see [21], [10], [15]).
Fig. 19 illustrates the VBAP panning gains for a common 5.1 surround setup (see [13]).
Between the two rear speakers at 110° and 250s, rather flat curves and a Sow level
difference for a wide angular range are observed. For an angular range where sum
localization is not really working, VBAP results in even smaller level differences than for a
smaller opening angle where sum localization is working. The reason for this behavior is
the great opening angle between the vector bases.
In Fig. 20, a generalized VBAP method using an imaginary loudspeaker (light gray) and a
downmix is depicted.
For a 3D loudspeaker setup, VBAP always uses 3 base vectors depending on the chosen
triangulation. If the 3D setup consists of two or more height layers stacked on top of each
other with loudspeakers at the same azimuth angles, then there is no preference for a
certain triangulation. For each section between two speakers of a layer, there are two
possibilities for subdividing the rectangle between the middle and the upper layer
speakers into two triangles. This arbitrary choice introduces an asymmetry even for
perfectly symmetric setups. To illustrate this property, let us take a 5.1 setup as an
example that has been extended by four height speakers above the M30, M-30, M 110,
and M-1 0 speakers i.e., U30, U-30, U 110, and U-1 10 [14]. Between the middle and the
upper layer surround speakers, the subdivision into the two triangles can either be defined
by the diagonal M 110 ®U-1 10 or by the diagonal U 110 M-1 10. The same holds for the
area above/between the upper layer loudspeakers. Whatever choice is made, it breaks
the left-right symmetry. As a consequence, an audio object that moves from the upper
front right to the upper rear left would sound different then if it would move from upper
front left to upper rear right - despite the symmetry of the loudspeaker setup.
The object of the present invention is to provide improved concepts for amplitude panning.
The object of the present invention is solved by an apparatus according to claim 1, by a
method according to claim 26 and by a computer program according to claim 27, by an
apparatus according to claim 28, by a method according to claim 45 and by a computer
program according to claim 46.
An apparatus for generating four or more audio output signals is provided. The apparatus
comprises a panning gain determiner and a signal processor. The panning gain
determiner is configured to determine a proper subset from a set of five or more
loudspeaker positions, so that the proper subset comprises four or more of the five or
more loudspeaker positions. Moreover, the panning gain determiner is configured to
determine the proper subset depending on a panning position and depending on the five
or more loudspeaker positions. Furthermore the panning gain determiner is configured to
determine a panning gain for each of the four or more audio output signals by determining
said panning gain depending on the panning position and depending on the four or more
loudspeaker positions of the proper subset. The signal processor is configured to
generate each audio output signal of the four or more audio output signals depending on
the panning gain for said audio output signal and depending on an audio input signal.
Moreover, a method for generating four or more audio output signals is provided. The
method comprises:
Determining a proper subset from a set of five or more loudspeaker positions, so
that the proper subset comprises four or more of the five or more loudspeaker
positions, wherein determining the proper subset is conducted depending on a
panning position and depending on the five or more loudspeaker positions.
Determining a panning gain for each of the four or more audio output signals by
determining said panning gain depending on the panning position and depending
on the four or more loudspeaker positions of the proper subset. And:
Generating each audio output signal of the four or more audio output signals
depending on the panning gain for said audio output signal and depending on an
audio input signal.
Furthermore, a computer program for implementing the above-described method when
being executed on a computer or signal processor is provided.
Moreover, an apparatus for generating four or more audio output signals is provided. Each
loudspeaker position of four or more loudspeaker positions is associated with exactly one
of the four or more audio output signals, and wherein each of the four or more audio
output signals is associated with exactly one of the four or more loudspeaker positions.
The apparatus comprises a panning gain determiner, and a signal processor. The panning
gain determiner is configured to determine, for each audio output signal of the four or
more audio output signals, a group of associated loudspeaker positions, being associated
with said audio output signal, depending on the loudspeaker position of each of the four or
more audio output signals and depending on a panning position, so that said group of
associated loudspeaker positions comprises the loudspeaker position being associated
with said audio output signal and at least two further loudspeaker positions of the four or
more loudspeaker positions, wherein at least one of the four or more loudspeaker
positions is not comprised by said group of associated loudspeaker positions. Moreover,
the panning gain determiner is configured to calculate, for each audio output signal of the
four or more audio output signals, the panning gain for said audio output signal depending
on the panning position and depending on the loudspeaker positions of the group of
associated loudspeaker positions being associated with said audio output signal. The
signal processor is configured to generate each audio output signal of the four or more
audio output signals depending on the panning gain for said audio output signal and
depending on an audio input signal. The group of associated loudspeaker positions being
associated with a first one of the four or more audio output signals is not equal to the
group of associated loudspeaker positions being associated with a different second one of
the four or more audio output signals.
Furthermore, a method for generating four or more audio output signals is provided. Each
loudspeaker position of four or more loudspeaker positions is associated with
exactly one of the four or more audio output signals, and wherein each of the four
or more audio output signals is associated with exactly one of the four or more
loudspeaker positions. The method comprises:
Determining, for each audio output signal of the four or more audio output signals,
a group of associated loudspeaker positions, being associated with said audio
output signal, depending on the loudspeaker position of each of the four or more
audio output signals and depending on a panning position, so that said group of
associated loudspeaker positions comprises the loudspeaker position being
associated with said audio output signal and at least two further loudspeaker
positions of the four or more loudspeaker positions, wherein at least one of the four
or more loudspeaker positions is not comprised by said group of associated
loudspeaker positions,
Calculating, for each audio output signal of the four or more audio output signals,
the panning gain for said audio output signal depending on the panning position
and depending on the loudspeaker positions of the group of associated
loudspeaker positions being associated with said audio output signal. And:
Generating each audio output signal of the four or more audio output signals
depending on the panning gain for said audio output signal and depending on an
audio input signal.
The group of associated loudspeaker positions being associated with a first one of the
four or more audio output signals is not equal to the group of associated loudspeaker
positions being associated with a different second one of the four or more audio output
signals.
Furthermore, a computer program for implementing the above-described method when
being executed on a computer or signal processor is provided.
The provided concepts provide a requirement-driven concept for amplitude panning.
In the following, embodiments of the present invention are described in more detail with
reference to the figures, in which:
Fig. 1 is an apparatus according to an embodiment,
Fig. 2 illustrates 17 loudspeakers at 17 loudspeaker positions,
Fig. 3 illustrates an example for a determination of a proper subset of loudspeaker
positions according to an embodiment dependent on a first panning
position,
Fig. 4 illustrates another example for a determination of a proper subset of
loudspeaker positions according to an embodiment dependent on a second
panning position,
Fig. 5 illustrates the determination of two proper subsets of loudspeaker positions
according to an embodiment depending on the first and the second panning
position,
Fig. 6 illustrates the determination of two proper subsets of loudspeaker positions
according to an embodiment depending on the first and a third panning
position,
Fig. 7 illustrates five loudspeaker positions and a panning position,
Fig. 8 illustrates a first triangle-subdivision of a polygon defined body dependent
on a first loudspeaker position according to an embodiment,
Fig. 9 illustrates a second triangle-subdivision of the polygon defined body
dependent on a second loudspeaker position according to an embodiment,
illustrates a third triangle-subdivision of the polygon defined body
dependent on a third loudspeaker position according to an embodiment,
illustrates a fourth triangle-subdivision of the polygon defined body
dependent on a fourth loudspeaker position according to an embodiment,
illustrates a fifth triangle-subdivision of the polygon defined body dependent
on a fifth loudspeaker position according to an embodiment,
illustrates triangle-subdivisions of another polygon-defined according to an
embodiment, wherein the polygon-defined body is a quad,
illustrates a triangle-subdivision of a further polygon-defined according to
an embodiment, wherein the polygon-defined body is a hexagon,
illustrates a subdivision of a further polygon-defined according to an
embodiment, wherein the polygon-defined body is a octagon, which is
subdivided into quads,
illustrates a determination of a panning gain based on distances according
to an embodiment,
illustrates a determination of a panning gain based on distances according
to another embodiment,
illustrates a system according to an embodiment,
illustrates the VBAP panning gains for a common 5.1 surround setup,
depicts a generalized VBAP method using an imaginary loudspeaker and a
downmix,
illustrates VBAP triangles in spherical coordinates for a 5.1+4 setup,
illustrates panning gains for a stereo setup,
Fig 23 illustrates a top view of an angular deviation between VBAP and linear
cross-fading,
Fig. 24 illustrates a subdivision of polygon-defined body into triangles according to
an embodiment,
Fig. 25 indicates trajectories reproduced by test signals in a listening test,
Fig. 26 shows the average and the 95% confidence interval of test results for a first
listening test where the timbre was rated,
Fig. 27 illustrates a difference plot for the first listening test where the timbre was
rated,
Fig. 28 shows the test results for a second test where the location accuracy and
smoothness of movement was rated,
Fig. 29 illustrates a difference plot for the second listening test where the location
accuracy and smoothness of movement was rated,
Fig. 30 shows the test results for a third test where the source extension and focus
was rated,
Fig. 3 1 illustrates a difference plot for the third listening test where the source
extension and focus was rated,
Fig. 32 illustrates the results for the overall quality, and
Fig. 33 illustrates a difference plot for the results for the overall quality.
Fig. 1 illustrates an apparatus for generating four or more audio output signals according
to an embodiment. The apparatus comprises a panning gain determiner 110 and a signal
processor 120.
The panning gain determiner 110 is configured to determine a proper subset from a set of
five or more loudspeaker positions, so that the proper subset comprises four or more of
the five or more loudspeaker positions, wherein the panning gain determiner 110 is
configured to determine the proper subset depending on a panning position and
depending on the five or more loudspeaker positions.
Moreover, the panning gain determiner 110 is configured to determine a panning gain for
each of the four or more audio output signals by determining said panning gain depending
on the panning position and depending on the four or more loudspeaker positions of the
proper subset.
The signal processor 120 is configured to generate each audio output signal of the four or
more audio output signals depending on the panning gain for said audio output signal and
depending on an audio input signal.
A proper subset of a set of five or more loudspeaker positions is a subset of the five or
more loudspeaker positons which does not comprise at least one of the five or more
loudspeaker positions.
As described the panning gain determiner is configured to determine a proper subset from
a plurality of five or more loudspeaker positions, so that at least four loudspeaker positions
are comprised by the subset.
This is explained with reference to Figs. 2 - 6 .
Fig. 2 illustrates 17 loudspeakers at 17 loudspeaker positions 201 - 217. The 17
loudspeaker positions 201 - 217 define five pentagons 221 , 222, 223, 224 and 225. In
particular, the pentagon 221 is defined by a polygon with the vertices 201 , 202, 203, 204
and 205. The pentagon 222 is defined by a polygon with the vertices 201 , 205, 206, 207
and 208. The pentagon 223 is defined by a polygon with the vertices 201 , 208, 209, 210
and 2 11. The pentagon 224 is defined by a polygon with the vertices 208, 212, 213, 214
and 209. And, the pentagon 225 is defined by a polygon with the vertices 209, 214, 215,
216 and 217.
In Fig, 2 it may be assumed that the loudspeaker positions are positions within a twodimensional
coordinate system.
For example, the horizontal axis 231 of the two-dimensional coordinate system may, e.g.,
indicate an azimuth angle Q of the loudspeaker position, and the vertical axis 232 of the
coordinate system may, e.g., indicate an elevation angle f of the coordinate system.
Thus, in all loudspeaker positions that are described only by an azimuth angle or an
elevation angle may be positions (assumed to be) located on a sphere in the real threedimensional
world.
Or, for example, the horizontal axis 231 of the coordinate system may, e.g., indicate an
abscissa (x-axis) coordinate value of the loudspeaker position, and the vertical axis 232 of
the coordinate system may, e.g., indicate an ordinate (y-axis) of a Cartesian coordinate
system. For example, in the real-dimensional world, all loudspeakers may be located in a
plane.
n Fig. 2 , the bodies defined by the polygons are convex. For example, the body defined
by the polygon with the vertices 201 , 202, 203, 204 and 205 is convex. Moreover, for
example, the body defined by the polygon with the vertices 201 , 205, 208, 207 and 208 is
convex.
Moreover, the five polygons which define the five pentagons do not enclose any other
loudspeaker position which does not belong to the respective polygon. For example, the
polygon with the vertices 201 , 202, 203, 204 and 205 does not enclose any of the
loudspeaker positions 206 - 217.
In Fig. 3 , a panning position 241 is indicated. The loudspeaker arrangement shall now
reproduce an audio input signal, as if the source emitting the audio input signal would be
located at the panning position.
The panning gain determiner 110 of Fig. 1 may, e.g., be configured to determine one of
the polygons mentioned-above that enclose the panning position to determine the subset
of loudspeaker positions. In the example of Fig. 3 , this is the (subset-specific) polygon
with the vertices 201 , 202, 203, 204 and 205. Thus the loudspeaker positions 201 , 202,
203, 204 and 205 are the (only) elements of the proper subset of loudspeaker positions.
Vice-versa, the polygon that defines the subset is subset-specific for the subset and can
thus be referred to as subset-specific polygon.
The panning gain determiner is now configured to determine a panning gain for each of
the audio output signals depending on the panning position and depending on the
loudspeaker positions of the (preselected) subset.
After the loudspeaker positions o the proper subset have been determined, it is not
necessary to further consider the other loudspeaker positions for determining the panning
gains.
Embodiments are based on the finding that to reproduce the audio input signal as if t
would originate from the panning position 241 , it is sufficient that only the loudspeakers
201 , 202, 203, 204 and 205 output an output signal. The other loudspeakers are not
needed.
As each audio output signal is generated for a particular loudspeaker position (or, in other
words, for a loudspeaker which is associated with, or, e.g., positioned at, that particular
loudspeaker position), it is sufficient to generate only the audio output signals for the
loudspeakers at the loudspeaker positions of the proper subset, to reproduce an audio
output signal, as if it would be emitted from the panning position.
Thus, for generating the audio output signals, only one panning gain for each of the audio
output signals is needed to reproduce the audio input signal as if emitted from the panning
position. Moreover, for determining the panning gain, as the audio input signal is panned
between the loudspeakers associated with the loudspeaker positions of the proper subset,
only the panning position and the loudspeaker positions of the proper subset have to be
taken into account.
Therefore, these embodiments are advantageous, as only a reduced number of
loudspeaker positions has to be considered, what reduces complexity.
Moreover, embodiments are based on the finding that at least four loudspeaker positions
shall be in the subset, as at least four speakers should be employed to represent an audio
input signal in the panning position. Embodiments are also based on the finding that a
reproduction of the audio input signal by only three speakers or less has disadvantages
compared to using four or more speakers, in particular, when the panning position moves,
as will be described further below.
Therefore, the subset is a proper subset and does therefore not comprise a l existing
loudspeaker positions, but also, the subset comprises four or more loudspeaker positions.
Fig. 4 again illustrates the 17 loudspeakers at the 1 loudspeaker positions 201 - 217. In
Fig. 4 , a new panning position is located at position 242. The new panning position 242 of
Fig. 4 is different from the old panning position 241 of Fig. 3 . The reason for this, may for
example be, that in the recording scene, a person which emits sound waves causing the
audio input signal may have moved, so that at a later point in time, the panning position
also moves from position 241 to position 242.
Or, panning position 242 may relate to the same point-in-time but to a further audio input
signal. For example, the panning position 241 may relate to a first audio input signal which
may comprise the sound part of a violin in an orchestra. The panning position 242 may
relate to a second audio input signal which may comprise the sound port of a trumpet in
the orchestra. Then, in the reproduction scene, panning position 241 indicates that the
violin shall be virtually positioned at panning position 241 , and the panning position 242
indicates that the trumpet shall be virtually positioned at panning position 242. Thus, in an
embodiment, the audio input signal relating to the violin is only reproduced by the
loudspeakers at loudspeaker positions 201 , 202, 203, 204 and 205 and the further audio
input signal relating to the trumpet is only reproduced by the loudspeakers at loudspeaker
positions 208, 212, 213, 214 and 209 (see Fig. 5). Thus, according to an embodiment,
panning gains for amplifying or attenuating the audio input signal representing the sounds
from the violin are only calculated for the loudspeakers at loudspeaker positions 201 , 202,
203, 204 and 205. And gains for amplifying or attenuating the further audio input signal
representing the sounds from the trumpet are only calculated for the loudspeakers at
loudspeaker positions 208, 212, 213, 214 and 209.
In this example, referring to the audio input signal representing the violin as a s and
referring to the audio input signal representing the trumpet as ais , then gains g g2 1,
93,1 , 9 ,1 and g5 for the loudspeakers at loudspeaker positions 201 , 202, 203, 204 and
205, respectively, are calculated by the panning gain determiner 110, and the signal
processor 120 applies the calculated panning gains g ·,, g2, , g3 ,i , g . and g on the audio
input signal ais 1 to obtain the audio output signals aosi, aos 2, aos 3 aos and aos 5 for the
loudspeakers at the loudspeaker positions 201 , 202, 203, 204 and 205, respectively, for
example, according to:
Likewise, gains g 8 2, g 2 2 , 3,2 9i4,2 and g for the loudspeakers a t loudspeaker positions
208, 212, 213, 214 and 209, respectively, are calculated by the panning gain determiner
110, and the signal processor 120 applies the calculated panning gains g , g12,2 , g 3 2,
9 i and g 2 on the audio input signal ais 2 to obtain the audio output signals aos , aos 1 ,
aos 3 , aos 4 and aos for the loudspeakers at the loudspeaker positions 208, 212, 213,
214 and 209, respectively, for example, according to:
aos 3 = g 3 2 ais 2
In particular, according to an embodiment, the audio input signal comprises a plurality of
audio input samples. The signal processor 120 may, e.g., be configured to generate each
audio output signal of the four or more audio output signals by multiplying each of one or
more of the audio input samples of the audio input signal with the panning gain for said
audio output signal to obtain one or more audio output samples of the audio output signal.
Or, in another embodiment, the audio input signal comprises a plurality of audio input
samples, and the signal processor 120 is configured to generate each audio output signal
of the four or more audio output signals by multiplying each of one or more of the audio
input samples of the audio input signal with a square root of the panning gain for said
audio output signal to obtain one or more audio output samples of the audio output signal.
In some cases, more than one audio output signal shall be reproduced by the same
loudspeaker. For example, in Fig. 6 , the panning position 243 relating to an audio input
signal ais3 is located within a pentagon defined by the polygon with the vertices 201 , 208,
209, 210 and 2 11. Then, panning gains g 3, g8 3, g 3. 9io ,3 and g11 3 relating to ais 3 are
calculated by the panning gain determiner 110, and the signal processor 120 applies the
calculated panning gains g1 3, g8 3, g9,3 , g 0 3 and g 3 on the audio input signal ais3. To
obtain the audio output signals, the signal processor 120 may, for example, apply the
following formulas:
aos = g. . aisi + g 3 ais3
aos 0
= 9l0,3 ais 3
aosn = gn, 3 ais3
In more general, if an audio output signal shall reproduce portions of more than one audio
input signal, the signal processor 120 may, e.g., be configured to obtain such an audio
output signal by applying the respective gains on the respective audio input signals and by
combining the respectively amplified or attenuated audio input signals. For example, in
Fig. 1, the calculated panning gain g is applied on aisi to obtain the amplified or
attenuated ais-,, and the calculated panning gain g 3 is applied on ais 3 to obtain the
amplified or attenuated g 3 · ais3. Then aisi and g 3 · ais3 are combined.
Thus, the provided concepts can be applied to more than one audio input signal.
Correspondingly, according to an embodiment, the audio input signal may, e.g., be a first
audio input signal, wherein the panning position is a first panning position, wherein the
panning gain is a first input-signal-dependent panning gain, and wherein the proper
subset is a first proper subset.
The panning gain determiner 110 may, e.g., be configured to determine one or more
further proper subsets from a set of five or more loudspeaker positions, so that each of the
one or more further proper subsets comprises four or more of the five or more
loudspeaker positions. Moreover, the panning gain determiner 110 may, e.g., be
configured to determine each of the one or more further proper subsets depending on one
of one or more further panning positions and depending on the five or more loudspeaker
positions,
Moreover, the panning gain determiner 110 may, e.g., be configured to determine one or
more further input-signal-dependent panning gains for each of the four or more audio
output signals by determining each of the one or more further panning gains depending on
one of the one or more further panning positions and depending on the four or more
loudspeaker positions of one of the one or more further proper subsets. The signal
processor 1 0 may, e.g., be configured to generate each audio output signal of the four or
more audio output signals depending on the first input-signal-dependent panning gain for
said audio output signal, depending on the one or more further input-signal-dependent
panning gains for said audio output signal, depending on the audio input signal, and
depending on the one or more further audio input signals.
As a side remark, it is mentioned, that in the following, the panning position is sometimes
also called a panning direction. The term panning direction originates from that for
example, in an azimuth, elevation coordinate system, the panning position in the twodimensional
coordinate system is, in the real three-dimensional setup a direction
information pointing from a central point, e.g., from a sweet spot to the direction of the
loudspeaker.
In the following, another aspect of embodiments is described. This aspect relates to how
panning is realized between the loudspeaker positions of the determined subset, for
example between the loudspeaker positions 208, 212, 213, 214 and 209 of Figs. 4 and 5 .
However, it should be noted that according to some embodiments, no preselection of a
subset takes place. Instead, for example, audio output signals are generated for the
loudspeakers at all loudspeaker positions 208, 212, 213, 214 and 209 to simulate that an
audio output signal, for example an audio output signal ais2 originates from a panning
position, e.g., a panning position 242. Also, such embodiments are covered.
Fig. 7 illustrates the setup, showing the 5 loudspeaker positions 208, 212, 213, 214 and
209 of the loudspeakers and the panning position 242. According to embodiments, to
determine the panning gains for the to obtain audio output signals to be output at the
loudspeaker positions 208, 212, 2 13 , 214 and 209, the following concepts are applied:
Fig. 8 illustrates panning gain determination for the audio output signal for loudspeaker
position 209. The body enclosed by the polygon with the vertices 208, 212, 2 13 , 214 and
209 is subdivided into three triangles, namely a first triangle with vertices 209, 208, 212, a
second triangle with vertices 209, 212, 213, and a third triangle with vertices 209, 2 13 ,
214, so that the subdivision of the body resulted in triangles that have loudspeaker
position 209 (for which the panning gain is determined) as vertex.
As the panning position is comprised by the second triangle with the vertices 209, 212,
213, according to embodiments, the panning gain for the audio output signal for
loudspeaker position 209 is then calculated depending on loudspeaker positions 209, 212,
2 13 , and not by the remaining loudspeaker positions 208 and 214. This simplifies
computation and heips to save processor time compared to using all loudspeaker
positions when calculating the panning gain associated with the audio output signal to be
generated for loudspeaker position 209.
Thus, by subdividing the body enclosed by the polygon, the panning gain determiner has
determined a group of associated loudspeaker positions comprising the loudspeaker
positions 209, 212, 213, wherein the group of associated loudspeaker positions is
associated the audio output signal for the loudspeaker at loudspeaker position 209 and
determines which of the loudspeaker positions are taken into account when calculating
the panning gain to obtain the output signal for (associated with) the loudspeaker position
209.
Vice versa, the group of associated loudspeaker signals defines a triangle that is groupspecific
for the group of associated loudspeaker signals. In more general, the triangle 209,
212, 213 can be considered as a group-specific polygon with the vertices 209, 212, 213.
Likewise, Fig. 9 illustrates panning gain determination for the audio output signal for
loudspeaker position 208. The body enclosed by the polygon with the vertices 208, 212,
213, 214 and 209 is subdivided into three triangles, namely a first triangle with vertices
208, 212, 213, a second triangle with vertices 208, 213, 214, and a third triangle with
vertices 208, 214, 209, so that the subdivision of the body resulted in triangles that have
loudspeaker position 209 (for which the panning gain is determined) as vertex. As the
panning position is comprised by the first triangle with the vertices 208, 212, 213,
according to embodiments, the panning gain for the audio output signal for loudspeaker
position 208 is then calculated depending on loudspeaker positions 208, 212, 213, and
not by the remaining loudspeaker positions 209 and 214. Thus, by subdividing the body
enclosed by the polygon, the panning gain determiner has determined a group of
associated loudspeaker positions comprising the loudspeaker positions 208, 212, 213,
wherein the group of associated loudspeaker positions is associated the audio output
signal for the loudspeaker at loudspeaker position 208.
Similarly, Fig. 10 illustrates that the group of associated loudspeaker positions, being
associated with the audio output signal for the loudspeaker at loudspeaker position 212,
comprises the loudspeaker positions 212, 213, 214 and the panning gain to obtain said
audio output signal is calculated depending on these loudspeaker positions 212, 213, 214.
Likewise, Fig. 11 illustrates that the group of associated loudspeaker positions, being
associated with the audio output signal for the loudspeaker at loudspeaker position 213,
comprises the loudspeaker positions 2 13 , 208, 212 and the panning gain to obtain said
audio output signal is calculated depending on these loudspeaker positions 213, 208, 212.
Similarly, Fig. 12 illustrates that the group of associated loudspeaker positions, being
associated with the audio output signal for the loudspeaker at loudspeaker position 214,
comprises the loudspeaker positions 214, 212, 213 and the panning gain to obtain said
audio output signal is calculated depending on these loudspeaker positions 214, 212, 213.
According to embodiments, the triangle that encloses the panning position defines the
group of associated loudspeaker positions.
If the panning position is exactly located on an edge of two of the triangles, some
embodiments, for example, choose one of the two triangles for calculating the panning
gain. Other embodiments, for example, calculate a first intermediate panning gain for a
first one of the two triangles and further calculate a second intermediate panning gain for
a second one of the two triangles, and then calculate the average of the first and the
second intermediate panning gain as the final panning gain.
For subdividing the body defined by the polygon, (here the polygon with the edges 208,
212, 213, 214, 209, which here defines a pentagon) it is preferable that the body is
convex.
Moreover, it is preferred that the body defined by the polygon is subdivided into triangles,
such that a triangle does not enclose loudspeaker positions different from the loudspeaker
positions that define the vertices of the triangle.
According to some embodiments, the polygon with the loudspeaker positions as vertices,
does not define a pentagon, but defines any other kind of body with four or more vertices,
for example, a quad, a hexagon, etc.
Fig. 13 illustrates the subdivision of a quad for the audio output signals for loudspeakers
at each of the loudspeaker positions 301 , 302, 303, 304 and for a panning position 305.
Fig. 14 illustrates the subdivision of a hexagon with vertices 401 , 402, 403, 404, 405, 406
for the audio output signal for the loudspeaker at the loudspeaker position 401 and for a
panning position 407. The group of associated loudspeaker positions for the audio output
signal for the loudspeaker at the loudspeaker position 401 comprises the loudspeaker
positions 401 , 403 and 404.
The sub-bodies in which the body defined by the polygon is subdivided do not have to be
triangles. Fig. 15 illustrates an example according to an embodiment, wherein the octagon
with the vertices 501 , 502, 503, 504, 505, 506, 507 and 508 is subdivided into three
quads, when the group of associated loudspeaker positions for the audio output signal for
loudspeaker position 501 shall be determined, namely a first quad with the vertices 401 ,
402. 403 and 404, a second quad with the vertices 401 , 404, 405 and 406 and a third
quad with the vertices 401 , 406, 407, 408. As panning position 409 is enclosed by the
quad 401 , 404 405, 406, the panning gain determiner calculates the panning gain
associated with loudspeaker position 401 dependent on the panning position 409, and
depending on the loudspeaker positions 401 , 404, 405, 406 of the group of associated
loudspeaker positions for the audio output signal for loudspeaker position 401 .
In general the panning gain determiner 110 is configured to determine a group-specific
polygon which encloses the panning position. Such a polygon is group specific for the
group of associated loudspeaker signals.
These concepts are based on the finding that complexity is reduced when less than all
loudspeaker positions are taken into account.
Moreover, these concepts are based on the finding, that by determining a gain factor for
each audio output signal for each of the loudspeaker positions creates a more realistic
sound impression compared to only determining gain factors and thus audio output
signals for the loudspeaker positions of a single triangle. Instead, embodiments determine
gain factors for each loudspeaker position of the subset, although by only taking for each
of the gain factors the loudspeaker positions of a gain-factor-specific triangle into account.
However, as the corresponding triangles (or, more generally: sub-bodies) for determining
the panning gains for the audio output signals differ for at least some of the audio output
signals, this ensures, that all loudspeaker positions are taken into account for determining
at least one of the gain factors. This is advantageous compared to always taking the same
triangle into account for determining all panning gains.
In the following, another aspect of the invention is described. Here, it is explained, how the
panning gain for an audio output signal for a loudspeaker at a loudspeaker position may,
for example, be determined depending on the panning position and depending on the
loudspeaker positions of the group of associated loudspeaker positions.
Fig. 16 illustrates a corresponding example showing the loudspeaker positions 501 , 502,
503, 504, 505 and panning position 506. As the panning position 508 is located within the
triangle of the loudspeaker positions 501 , 503 and 504, only the loudspeaker positions
501 , 503 and 504 belong to the group of associated loudspeaker positions, and only these
loudspeaker positions 501 , 503, 504 are taken into account for determining the panning
gain for the audio output signal for loudspeaker position 501 {and not the loudspeaker
positions 502 and 505, which do not belong to this group of associated loudspeaker
signals).
Line 5 11 indicates a first distance being a shortest distance between the panning position
507 and a first straight ine through the two further loudspeaker positions 503, 504 of the
group of associated loudspeaker positions.
Line 512 indicates a second distance being a shortest distance between the loudspeaker
position 501 (for the audio output signal of which, the panning gain is determined) and a
second straight line 515 through the panning position, wherein said second straight line is
parallel to said first straight line 510.
The panning gain determiner 110 may, for example, be configured to determine the
panning gain depending of the ratio of the first distance 5 11 and a sum of the first distance
5 11 and the second distance 512.
For example, assuming that in Fig. 16 the first distance 5 11 is 0.6 and the second
distance 512 is 0.2, then, the panning gain 0 may, e.g., be calculated to be
0.6 0.6 _.
p = = = 0.75
0.6 + 0,2 0.8
This reflects that the loudspeaker position 501 is closer to the line 515 than loudspeaker
positions 503 and 504 and thus, the panning gain p is closer to 1 than to 0 .
Fig. 17 illustrates another embodiment, wherein the group of associated loudspeaker
positions comprises four loudspeaker positions 601 , 602, 603 and 604. The panning
position is indicated by 605. The panning gain for the audio output signal for loudspeaker
position 601 shall be determined. Mathematical concepts of the state of the art may be
employed to determine a curve 608 through the loudspeaker positions 602, 603, 604. In
Fig. 17, a dashed straight line 610 through loudspeaker position 601 and panning position
605 is illustrated. The intersection of the dashed straight line 610 and curve 608 defines
intersection point 609. A first distance 6 11 is defined by the distance between panning
position 605 and intersection point 609. A second distance 612 is defined by the distance
between panning position 605 and loudspeaker position 601 .
Again, the panning gain determiner 110 may, for example, be configured to determine the
panning gain depending of the ratio of the first distance 5 11 and a sum of the first distance
5 1 and the second distance 512.
Assuming that in Fig. 17, the first distance 6 is 0.25 and that the second distance 612
is 0.3, the panning gain may, e.g., be
0.25 025
p = = = 0 .435
1 0.25 + 0.3 0.55
The panning gain ¾o is slightly below 0.5 and this reflects that the loudspeaker position
601 is slightly farer away from the panning position 605 than the intersection point 609.
As already mentioned, in some embodiments, no group of associated loudspeaker
positions for determining each of the panning gains is determined. Instead, all
loudspeaker positions of the proper subset are taken into account for calculating each
gain.
In such an embodiment, each loudspeaker position of the four or more loudspeaker
positions of the proper subset is associated with exactly one of the four or more audio
output signals, and wherein each of the four or more audio output signals is associated
with exactly one of the four or more loudspeaker positions of the proper subset. The
panning gain determiner 110 may, e.g., be configured to calculate, for each audio output
signal of the four or more audio output signals, the panning gain for said audio output
signal depending on a panning position and depending on the loudspeaker position of
each of the four or more audio output signals.
Different panning gains may, e.g., be determined for different points-in-time. According to
such embodiments, the panning gain determiner 110 may, e.g., be configured to
determine the panning gain for each audio output signal of the four or more audio output
signals for a first point in time as a first time-dependent panning gain for said audio output
signal. Moreover, the panning gain determiner 110 may, e.g., be configured to determine
a further panning gain for each audio output signal of the four or more audio output
signals for a different second point in time as a second time-dependent panning gain for
said audio output signal, said second time-dependent panning gain being different from
the first time-dependent panning gain for said audio output signal.
Moreover, different panning gains may, e.g., be determined for different frequencies.
According to such embodiments, the panning gain determiner 0 may. e.g., be
configured to determine the panning gain for each audio output signal of the four or more
audio output signals for a first frequency as a first frequency-dependent panning gain for
said audio output signal. Moreover, the panning gain determiner 0 may, e.g., be
configured to determine a further panning gain for each audio output signal of the four or
more audio output signals for a different second frequency as a second frequencydependent
panning gain for said audio output signal, said second frequency-dependent
panning gain being different from the first frequency-dependent panning gain for said
audio output signal.
Fig. 18 illustrates a system according to an embodiment. The system comprises five or
more loudspeakers, and an apparatus as described above with reference to Fig. 1. The
apparatus is indicated by reference sign 100. Furthermore, the system of Fig. 18
comprises 17 loudspeakers at the loudspeaker positions 201 - 217.
Each of the loudspeakers is associated with exactly one loudspeaker position 201 - 217
of the set of five or more loudspeaker positions.
Each of the four or more audio output signals is associated with exactly one loudspeaker
position of the loudspeaker positions 201 - 217 of the proper subset. Moreover, each of
the four or more audio output signals is associated with exactly one of the loudspeaker
positions 201 - 21 of the proper subset.
The system is configured to output each audio output signal of the four or more audio
output signals by the loudspeaker which is associated with the same loudspeaker position
as said audio output signal.
In an embodiment, the system may, e.g., be configured to output none of the four or more
audio output signals by any of the four or more loudspeakers not being associated with
the same loudspeaker position as said audio output signal.
According to an embodiment, the system may, e.g., be configured to not output any of the
four or more loudspeaker signals by at least one of the five or more loudspeakers.
As already mentioned above, not a l embodiments require that the panning gain
determiner 110 conducts a pre-selection of a proper subset of loudspeaker positions, as
described with reference to Figs. 3 - 6 .
In such embodiments, each loudspeaker position of four or more loudspeaker positions is
associated with exactly one of the four or more audio output signals, and wherein each of
the four or more audio output signals is associated with exactly one of the four or more
loudspeaker positions.
Moreover, in such embodiments, the panning gain determiner 110 of the apparatus of Fig.
1 is configured to determine, for each audio output signal of the four or more audio output
signals, a group of associated loudspeaker positions, being associated with said audio
output signal, depending on the loudspeaker position of each of the four or more audio
output signals and depending on the panning position, so that said group of associated
loudspeaker positions comprises the loudspeaker position being associated with said
audio output signal and at least two further loudspeaker positions of the four or more
loudspeaker positions, wherein at least one of the four or more loudspeaker positions is
not comprised by said group of associated loudspeaker positions.
Moreover, in such embodiments, the panning gain determiner 110 is configured to
calculate, for each audio output signal of the four or more audio output signals, the
panning gain for said audio output signal depending on the panning position and
depending on the loudspeaker positions of the group of associated loudspeaker positions
being associated with said audio output signal.
Furthermore, in such embodiments, the signal processor 120 is configured to generate
each audio output signal of the four or more audio output signals depending on the
panning gain for said audio output signal and depending on an audio input signal. The
group of associated loudspeaker positions being associated with a first one of the four or
more audio output signals is not equal to the group of associated loudspeaker positions
being associated with a different second one of the four or more audio output signals.
Each of the concepts, implementations and configurations described above with reference
to Figs. 1 - 17 may be employed in such an apparatus.
In the following, specific embodiments of the provided polygon-based panning concepts
for 3D loudspeaker setups are presented in more detail.
The provided concepts relate to 3D loudspeaker setups, as the 3D loudspeaker setup
may be projected in the above-described two-dimensional coordinate system.
Embodiments provide Edge Fading Amplitude Panning (EFAP) concepts for 3D
loudspeaker setups. Similar to other panning methods like Vector Base Amplitude
Panning (VBAP), it can be used to create phantom sources between the loudspeaker
positions. The proposed method features symmetric panning gains for symmetric
loudspeaker setups, N-wise panning by using polygons instead of triangles, and a better
behavior for large opening angles between loudspeakers while involving a computational
complexity that is in the same order of magnitude as VBAP,
A solution would require the usage of polygons instead of triangles as boundary, resulting
in N-wise panning. While VBAP supports only triangles due to its fundamental principle, it
can be generalized to yield N-wise panning as illustrated in Fig. 2 . In doing so, an
imaginary loudspeaker [1] is added in the middle of the polygon and its VBAP gain is then
downmixed to its neighbors - a solution that is simpler than previously proposed solutions
[2].
In surround productions, dual balance panners are widely used for positioning mono
signals. For 3D productions, such a panner can easily be extended by an additional slider
that adds height information. However, controlling the object's direction in 3D space is
probably more crucial than controlling the source extension or auditory source width.
Hence, using a dual balance panner for controlling the azimuth and the elevation angle of
an object in combination with a slider for automating the source extension, is a worthwhile
alternative. If such a user interface is employed, then the vector arithmetic of VBAP
results in a property that is illustrated in Fig. 2 1.
Fig 2 1 illustrates VBAP triangles in spherical coordinates for a 5.1+4 setup (squares:
loudspeaker positions).
The squares mark the loudspeaker positions of the setup that was already used in the
previous example. The solid lines result from the vectorial linear combination of the
loudspeaker pairs which specify the edges of the VBAP triangles. The apparent geometric
distortion of the triangles can be explained as follows: The triangles are the subdivided
surfaces of the polyhedron which is defined by the loudspeakers at constant radius By
projecting the triangle edges onto the sphere surface, one yields the azimuth and the
elevation angle as part of their spherical coordinates. Consequently, if the user wanted an
object to be pair-wise panned between the U 110 and the U-1 10 speaker located at 35°
elevation, he would have to follow a trajectory that goes beyond 60° elevation. For a
trajectory with constant elevation of 35°, VBAP would result in significant amplitudes for
the M 10 and M-1 10 loudspeaker channels.
Some of the provided embodiments aim to
1. keep an approach with reduced computation requirements (like VBAP)
2 . realized as amplitude panning ( ike VBAP)
3 . in a particular embodiment, employ power normalized gains (like VBAP)
4 realize native support for N-wise panning defined by polygons instead of triangles
5 . Pairwise panning along the polygon
6 . For small angles (< 60°) and 2 active speakers, approximate tan-law (2D VBAP)
7. For large angles and 2 active speakers achieve sufficient level difference to avoid
panning where sum localization doesn't work
8 . achieve smooth transition of the gains between the involved loudspeakers
9 . wherein some embodiments realize calculation directly in the spherical coordinate
system (see, for example, Fig 21).
Panning concepts are provided that conform to these requirements. 2D considerations are
extended for 3D setups.
At first, 2D considerations are described.
In the 2D case, the directional parameters are reduced to the azimuth angle. While the
fourth design goal is not relevant to the 2D case, the sixth and seventh are of special
importance. A simple solution that features the wanted properties can be found by
computing linear cross-fading gains as an intermediate result,
,
9 n = - -
« 0 (4)
where a denotes the opening angle between the involved loudspeaker pair and a„
denotes the angle between the respective loudspeaker and the panning direction.
Fig. 22 illustrates panning gains for a stereo setup (solid: VBAP; dotted: linear crossfading;
dashed: power normalized cross-fading). In particular, Fig. 22 depicts for an
opening angle of 80° these cross-fading functions and the target curves which are given
by the 2D VBAP panning gains
If in a second step the energy normalization 3) is applied to the linear cross-fading gains,
like for VBAP, one can observe that the result closely approximates the given target
curves.
Fig. 23 illustrates a top view of an angular deviation between VBAP (+) and linear crossfading
(□ ) . In particular, Fig. 23 illustrates the underlying approximation principle for the
60° example. The loudspeakers are the boundary of the shown angular range. The
crosses mark the given target directions which consist of a set of equi-angular
intermediate positions. If the two base vectors are multiplied by the corresponding crossfading
gains, the results are obtained that are marked by the squares. The angular
deviations between the target directions and the result of the cross-fading approach are
illustrated by the solid lines. From this geometric consideration it can be concluded a) that
the approximation is the closer the smaller the loudspeaker opening angle is and b) that
an opening angle of 80° can still be considered as being small.
The cross-fading gain only depends on the ratio between the panning angle and the
opening angle between the loudspeakers. Hence, a greater opening angle results in the
dashed graph shown in Fig. 22 scaled along the x-axis (azimuth angle). This is a desired
property as it complies with the seventh design goal.
Power normalization may be conducted, e.g., by employing the formula:
Now, a 3D concept is provided.
While the parameter space is one-dimensional in the 2D case and only consists of the
azimuth angle, it is two-dimensional in the 3D case and is spanned by the azimuth and the
elevation angle. By specifying the mesh / loudspeaker polygons in this parameter space,
compliance with the ninth design goal is achieved and the geometry distortion which can
be observed for VBAP's Euclidean domain is avoided. This can either be done manually
or by means of an algorithm like the Quick-Hull algorithm which outputs a triangle mesh
[3]. In the latter case, triangles can be combined to polygons, if their vertices are located
within the same plane or at least within a certain tolerance range.
Fig. 24 illustrates a subdivision of polygon-defined body into triangles according to an
embodiment. n particular, Fig, 24 illustrates a contour plot of the panning gain for the
topmost loudspeaker of an exemplary polygon. The x-axis indicates the azimuth angle,
the y-axis indicates the elevation angle, squares indicate loudspeaker vertices, solid lines
indicate gain contour, dashed lines indicate polygon edges, and arrows indicate normal
vectors.
The linear cross-fading method can be transferred to the 3D case by defining linear crossfading
functions between the loudspeakers of a polygon. Fig. 24 illustrates this by a
contour plot for the topmost loudspeaker which is part of a polygon defined by five
loudspeakers. The loudspeaker directions, which are the vertices of the polygon shown by
the dashed line, are marked by the squares. The gain for the topmost loudspeaker as a
function of the panning direction within the polygon is shown by means of the solid
contour lines.
In order to compute the cross-fading gain for a loudspeaker, the polygon first needs to be
sub-divided into triangles specified by the loudspeaker vertex and the edges of the
polygon. This sub-division is indicated in Fig. 24 by the dotted lines. The triangle in which
the panning direction P = is located can be determined by computing two
coefficients l and m ,
[C , m ] = [b - a , c - a] _ (p - a) . (5
where = Fa] denotes the direction of the loudspeaker for which the cross-fading
gain is computed, and where = and c = ¾ denote the remaining
vertices of the triangle. The panning direction p is located inside the triangle, if all of the
following conditions are fulfilled:
l > 0 6
³ 7
l + m < 1 )
For this triangle, the normal vector is then computed, e.g., according to
' = [Ob - . (9)
n =
i ' ( 10)
This normal vector then allows for computing the cross-fading gain as follows:
= l - n (p - a ) f
It should be noted that the sub-division into triangles and the computation of (9), 0), and
( 1 1 must be performed for each loudspeaker of the polygon.
The final panning gains are then obtained by applying the energy normalization (3) to the
cross-fading gains.
According to an embodiment, as a first step 2D crossfading is conducted, e.g., by applying
the formula
gn = 1 n (p - a„)
And, in some embodiments, as a second step, power normalization is conducted, e.g., by
applying the formula
A special feature of the used coordinate system is the existence of the poles at ±90°
elevation. As a pole may not be located within a polygon, a method like the generalized
VBAP approach needs to be applied to solve this issue. In doing so, an additional vertex
is added at ±90° elevation and the polygons which contain the poles are split. After
computing the panning gains for this extended set of loudspeakers, the gains for the
imaginary pole loudspeakers are downmixed to their physical neighbors.
Furthermore, as the poles are not points in the azimuth-elevation parameter space but
lines, it is reasonable for the computation of the cross-fading gains to set the azimuth
angle of the pole vertices to the azimuth angle of the panning direction.
All normal vectors besides those for the pole vertices can be pre-calculated as we as the
inverse matrices which are needed for the determination of the polygon/triangle n which p
is located. Consequently, the computational complexity for the determination of the
panning gains during runtime is considerable low.
If the panning direction coincides with the position of one of the loudspeakers, then only
this loudspeaker is active while two or more loudspeakers are active in between. The
varying source extension as a consequence of the varying number of active loudspeakers
can be compensated by means of Multiple Direction Amplitude Panning (MDAP) exactly in
the same way as it is done for VBAP [ 18],
Some localization studies have shown that the Vector Base Intensity Panning (VBIP)
method [16] based on Gerzon's Energy Vector [9] results in a smaller deviation between
the panning direction and the perceived source location, especially at higher frequencies.
This is a behavior which is to some extend predictable by a binaural model [8]. In general,
both methods can be combined into a frequency-dependent panning method as
suggested by Pulkki [19]. The same principle can be applied to the proposed method by a
frequency-dependent exponentiation of the cross-fading gains ( 1 1 .
As already described above, according to some embodiments, instead of applying the
gain, e.g., the gain of formula ( 11) on the samples of an audio input signal, a square root
of the gain, e.g., a square root of the gain of formula ( 1 1) may, e.g., be applied on the
samples of the audio input signal.
In order to assess the performance of the proposed panning concepts, a listening test was
performed where four different object trajectories were investigated: "front right" to "upper
front left", "rear right" to "upper rear left", "front left" to "upper side left", and "side left" to
"upper front left". Fig. 25 indicates these trajectories.
The conducted listening test was not a MUSHRA test. The "Re signal is a reference with
regard to all quality features besides location accuracy. The test signals shall reproduce
one of the trajectories. Participants were encouraged to slightly move their head within
±30° azimuth/elevation angle. The timbre, the location accuracy / smoothness of
movement, the source extension / focus, and the overall quality of all test signals were
judged and commented.
Each test item contained a single object at constant velocity which was rendered with an
elevation angle that was linearly interpolated between 0° and 35° and an azimuth angle
that was linearly interpolated as follows: Trajectory I (front): -30° to 30°; trajectory I I
(back): 110° to 110°; trajectory III (front-left): 30° to 90°; and trajectory IV (front-left): 90°
to 30°.
For the generation of the test signals, three kinds of mono signals were used, which were
then rendered along the four trajectories, namely 1: "Speech"; 2 : "Pink Noise"; and 3 :
"Beat".
In order to reduce the influence of the short-term memory, short stimuli were chosen. The
"Speech" signal was a 6.7s long sentence from a female speaker. The "Pink Noise" signal
contained 8s of stationary pink noise. The "Beat" signal also lasted 8s and contained a
beat of a woodblock and a castanet struck in turn at 160bpm. The three input signals were
manually adjusted to similar loudness.
Each of the 12 test items was rendered using the following panning concepts, namely 1:
"efap" (the proposed concepts); 2 : "vbap A"; and 3 : "vbap B".
All methods involved the shown loudspeaker sets. The two VBAP variants only differed by
the triangulation i.e., the diagonal within the rectangular loudspeaker arrangement,
whereas the diagonal of "vbap A" coincided with the trajectories I , II, and III.
Due to the difficulty to provide a proper reference for the rendered signals, the input signal
played back over the center speaker was used as tonal reference.
The participants of the conducted listening test, which took place in the ITU-R BS.1 116-1
compliant sound lab "Mozart" at Fraunhofer IIS, used a conventional MUSHRA software
that was configured to leave out the hidden references and lower anchors (see [ 1 1 , [12]).
In total 4 listening test instructions were handed out to the participants in written form
which only differed by the highlighted quality. The loudspeakers were marked by the same
labels which were also used within the instructions. The test participants were asked to
exclusively grade the quality features / attributes "timbre", "localization accuracy /
smoothness of movement", "source extension / focus", and "overall quality" of the
presented stimuli, where each test was conducted on a different day.
Table 1 illustrates test material used during the training phase.
No. .Signal Trajectory Panning Method
1 "Speech" (reference)
2 "Speech" 1 (from) vbap A
3 "Speech" , ront e ft ) "efap"
4 "Speech" 1 1 (back) "vbap "
5 "Pink" (reference)
6 "Pink" (hack) fap"
7 "Pink" 1 ront.) "vbap "
8 "Pink" V ( c t- ) "vbap A"
9 "Beat" - (reference)
ί "Beat" I V f b k'it "vbap B" '
"B at (back) "vbap A"
"Beat" I (front) "efap"
(Table 1)
In the following, the test results are presented.
Fig. 26 shows the average and the 95% confidence interval of the test results for the first
listening test where the timbre was rated.
It details the results for all combinations of input signals ("Pink", "Speech", and "Beat") and
trajectories (I, II, III, and IV). The average over all conditions indicates that the timbre of
the VBAP output is slightly closer to the reference than the output of the EFAP method.
This observation is confirmed by the difference plot shown in Fig. 27.
Fig. 27 illustrates a difference plot for the first listening test where the timbre was rated.
The given comments revealed that the EFAP output featured a slightly stronger bass
boost. This s an expected behavior as the incoherent summation which is the basic
assumption for the power normalization, no longer holds at low frequencies.
Hence, a greater number of loudspeakers causes a greater bass boost effect which can
be compensated by means of an equalizer [19].
Fig. 28 shows the test results for a second test where the location accuracy and
smoothness of movement was rated.
The corresponding difference p ot shown in Fig. 29 reveals that the EFAP method results
in a smoother movement / better location accuracy than VBAP.
Some subjects gave the feedback that the VBAP trajectories were partly too low and then
quickly moved to the upper loudspeaker at the end.
This is an observation which can be explained by the previously mentioned geometric
distortion which results in stronger gains for the middle layer loudspeakers.
Fig. 30 shows the test results for the third test where the source extension and focus was
rated.
The corresponding difference p ot is shown in Fig. 3 1. It can be observed that EFAP
performs either equally well or slightly worse than VBAP with respect to the source
extension. This observation can be explained by the fact that "vbap A" mostly resulted in
pair-wise panning and thus caused a smaller perceived source extension compared to the
other triangulation variant or EFAP.
The results for the overall quality are shown in Fig. 32 and Fig. 33. While some subjects
clearly favored one of the test candidates, the results are totally balanced on average.
In embodiments, symmetric panning gains for symmetric setups by N-wise panning
defined via polygons are realized.
The listening test, compared the provided concepts with VBAP, gives evidence that the
proposed concepts result in a better location accuracy. The greater number of active
loudspeakers stabilizes the position and trajectory of the phantom source, but it also
produces a slightly stronger bass boost and a slightly greater source extension.
While some subjects preferred the improved spatial accuracy, others put more emphasize
on the timbre, resulting in a balanced overall preference. The proposed concepts are
beneficial in applications where the location accuracy and smoothness of movement is of
importance. This property is further improved by frequency-dependent exponentiation of
the calculated cross-fading gains while the timbre could be compensated by means of
equalization.
Although some aspects have been described in the context of an apparatus, t is clear that
these aspects also represent a description of the corresponding method, where a block or
device corresponds to a method step or a feature of a method step. Analogously, aspects
described in the context of a method step also represent a description of a corresponding
block or item or feature of a corresponding apparatus.
The inventive decomposed signal can be stored on a digital storage medium or can be
transmitted on a transmission medium such as a wireless transmission medium or a wired
transmission medium such as the Internet.
Depending o certain implementation requirements, embodiments of the invention can be
implemented in hardware or in software. The implementation can be performed using a
digital storage medium, for example a floppy disk, a DVD, a CD, a ROM, a PROM, an
EPROM, an EEPROM or a FLASH memory, having electronically readable control signals
stored thereon, which cooperate (or are capable of cooperating) with a programmable
computer system such that the respective method is performed.
Some embodiments according to the invention comprise a non-transitory data carrier
having electronically readable control signals, which are capable of cooperating with a
programmable computer system, such that one of the methods described herein is
performed.
Generally, embodiments of the present invention can be implemented as a computer
program product with a program code, the program code being operative for performing
one of the methods when the computer program product runs on a computer, The
program code may for example be stored on a machine readable carrier.
Other embodiments comprise the computer program for performing one of the methods
described herein, stored on a machine readable carrier.
In other words, an embodiment of the inventive method is, therefore a computer program
having a program code for performing one of the methods described herein, when the
computer program runs on a computer.
A further embodiment of the inventive methods is, therefore, a data carrier (or a digital
storage medium, or a computer-readable medium) comprising, recorded thereon, the
computer program for performing one of the methods described herein.
A further embodiment of the inventive method is, therefore, a data stream or a sequence
of signals representing the computer program for performing one of the methods
described herein. The data stream or the sequence of signals may for example be
configured to be transferred via a data communication connection, for example via the
Internet
A further embodiment comprises a processing means, for example a computer, or a
programmable logic device, configured to or adapted to perform one of the methods
described herein.
A further embodiment comprises a computer having installed thereon the computer
program for performing one of the methods described herein.
In some embodiments, a programmable logic device (for example a field programmable
gate array) may be used to perform some or al of the functionalities of the methods
described herein. In some embodiments, a field programmable gate array may cooperate
with a microprocessor in order to perform one of the methods described herein. Generally,
the methods are preferably performed by any hardware apparatus.
The above described embodiments are merely illustrative for the principles of the present
invention. t is understood that modifications and variations of the arrangements and the
details described herein will be apparent to others skilled in the art. It is the intent,
therefore, to be limited on y by the scope of the impending patent claims and not by the
specific details presented by way of description and explanation of the embodiments
herein.
References:
[1] "SpaceMap" Software by Meyer Sound. URL: http://www.rrieyersound.com
/products /d-mitri/spacemap.htm.
[2] Ando, A. and Hamasaki, K. Sound Intensity Based Three-Dimensional Panning. In
126th AES Convention, Munich, Germany, May 2009.
[3] Barber, C. B., Dobkin, D. P., and Huhdanpaa, H. The QuickHull Algorithm for
Convex Hulls. ACM Trans. Math. Softw., 22(4):469-483. 1996.
[4] Bekesy, G. v. Experiments in Hearing McGraw-Hill, New York, NY, USA, 1960.
[5] Bennett, J . C , Barker, K., and Edeko, F. O. A New Approach to the Assessment of
Stereophonic Sound System Performance. J . Audio Eng. Soc, 33(5):314-321 ,
1985.
[6] Blauert, J. Spatial Hearing, The Psychophysics of Human Sound Localization. MIT
Press, Cambridge, MA, second edition, 1997.
[7] Craven, P. G. Continuous Surround Panning for 5-Speaker Reproduction. In AES
24th Int. Conference, Banff, Alberta, Canada, Jun. 2003.
[8] Frank, M. Localization Using Different Amplitude-Panning Methods in the Frontal
Horizontal Plane. n EAA Joint Symposium on Auralization and Ambisonics, Berlin,
Germany, Apr. 2014.
[9] Gerzon, M. A. General Metatheory of Auditory Localisation. In 92nd AES
Convention, Vienna, Austria, Mar. 1992.
[10] Gretzki, R. and Silzle, A. A New Method for Elevation Panning Reducing the Size
of the Resulting Auditory Events. In EAA Symposium "TecniAcustica 2003", Bilbao,
Spain, 2003.
[ 1 1] ITU-R. Rec. ITU-R BS.1 116-1 , Methods for the Subjective Assessment of Small
Impairments in Audio Systems Including Multichannel Sound Systems, 2003.
[12] ITU-R. ec ITU-R BS. 1534-1 , Method for the Subjective Assessment of
intermediate Quality Level of Coding Systems, 2003.
[ 13] ITU-R. Rec. ITU-R BS.775-1 , Multichannel Stereophonic Sound System with and
without Accompanying Picture, 2006.
[14] ITU-R. Rep. ITU-R BS. 2266-1 , Framework of Future Audio Representation
Systems, 2013.
[15] Lee, H. The Relationship Between Interchannel Time and Level Differences in
Vertical Sound Localization and Masking. In 131st AES Convention, New York,
NY, USA, Oct. 201 1,
[16] Pernaux, J.-M., Boussard, P., and Jot, J.-M. Virtual sound source positioning and
mixing in 5.1 implementation on the real-time system genesis. In First COST-G6
Workshopon Digital Audio Effects (DAFx-98), Barcelona, Spain, 1998.
[17] Pulkki, V. Virtual Sound Source Positioning Using Vector Base Amplitude Panning.
J. Audio Eng. So , 45(6):456 466, 1997.
[18] Pulkki, V Uniform Spreading of Amplitude Panned Virtual Sources. n IEEE
Workshop on Applications of Signal Processing to Audio and Acoustics, New
Paltz, NY, USA, Oct. 1999.
[19] Pulkki, V., Karjalainen, M., and Vaiimaki, V. Localization, Coloration, and
Enhancement of Amplitude-Panned Virtual Sources. In AES 16th Int. Conference,
Rovaniemi, Finland, 1999
[20] Silzle, A . 3D Audio Quality Evaluation: Theory and Practice. In International
Conference on Spatial Audio (ICSA), Erlangen, Germany, Feb. 2014.
[21] Theile, G. and Plenge, G. Localization of Lateral Phantom-Sources. In 53rd AES
Convention, Berlin, Germany, 1978.

Claims
An apparatus for generating four or more audio output signals, comprising:
a panning gain determiner ( 1 10), and
a signal processor (120),
wherein the panning gain determiner ( 1 10) is configured to determine a proper
subset from a set of five or more loudspeaker positions, so that the proper subset
comprises four or more of the five or more loudspeaker positions,
wherein the panning gain determiner ( 1 10) is configured to determine the proper
subset depending on a panning position and depending on the five or more
loudspeaker positions,
wherein the panning gain determiner ( 1 10) is configured to determine a panning
gain for each of the four or more audio output signals by determining said panning
gain depending on the panning position and depending on the four or more
loudspeaker positions of the proper subset, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals depending on the panning gain for
said audio output signal and depending on an audio input signal.
An apparatus according to claim 1,
wherein each loudspeaker position of the four or more loudspeaker positions of the
proper subset is associated with exactly one of the four or more audio output
signals, and wherein each of the four or more audio output signals is associated
with exactly one of the four or more loudspeaker positions of the proper subset,
wherein the panning gain determiner ( 10) is configured to determine, for each
audio output signal of the four or more audio output signals, a group of associated
loudspeaker positions, being associated with said audio output signal, depending
on the loudspeaker position of each of the four or more audio output signals and
depending on the panning position, so that said group of associated loudspeaker
positions comprises the loudspeaker position being associated with said audio
output signal and at least two further loudspeaker positions of the four or more
loudspeaker positions of the proper subset, wherein at least one of the four or
more loudspeaker positions of the proper subset is not comprised by said group of
associated loudspeaker positions,
wherein the panning gain determiner ( 0) is configured to calculate, for each
audio output signal of the four or more audio output signals, the panning gain for
said audio output signal depending on the panning position and depending on the
loudspeaker positions of the group of associated loudspeaker positions being
associated with said audio output signal, and
wherein the group of associated loudspeaker positions being associated with a first
one of the four or more audio output signals is not equal to the group of associated
loudspeaker positions being associated with a different second one of the four or
more audio output signals.
An apparatus according to claim 2 , wherein, for each of the four or more audio
output signals, the panning gain determiner ( 1 10) is configured to determine the
group of associated loudspeaker positions, being associated with said audio output
signal, so that said group of associated loudspeaker positions comprises exactly
three of the four or more loudspeaker positions of the proper subset.
An apparatus according to one of the preceding claims, wherein the panning
position and each loudspeaker position of the set of five or more loudspeaker
positions each indicates a position within a two-dimensional coordinate system.
An apparatus according to claim 4 , wherein the panning position and each
loudspeaker position of the set of five or more loudspeaker positions each
indicates an azimuth angle and an elevation angle within the two-dimensional
coordinate system.
An apparatus according to claim 4 or 5 , wherein the panning gain determiner ( 1 10)
is configured to determine the proper subset such that a subset-specific polygon
exists, wherein the vertices of the proper subset-specific polygon are the
loudspeaker positions of the proper subset, and wherein the panning position is
enclosed by the subset-specific polygon.
, An apparatus according to claim 8 , wherein the subset-specific polygon does not
enclose any of the five or more loudspeaker positions which is not comprised by
the proper subset
An apparatus according to claim 6 or 7 , wherein a body defined by the subsetspecific
polygon is convex
, An apparatus according to one of claims 4 to 8 , wherein, for each of the four or
more audio output signals, the panning gain determiner ( 1 10) is configured to
determine the group of associated loudspeaker positions, being associated with
said audio output signal, such that a group-specific polygon exists, wherein the
vertices of the group-specific polygon are the loudspeaker positions of said group
of associated loudspeaker positions, and wherein the panning position is enclosed
by said group-specific polygon.
0, An apparatus according to claim 9 , wherein the group-specific polygon, being
determined for each of the four or more audio output signals, does not enclose any
of the five or more loudspeaker positions which is not comprised by the proper
subset,
1, An apparatus according to claim 9 or 10 , wherein the group-specific polygon, being
determined for each of the four or more audio output signals, a body defined by
said group-specific polygon is convex.
2, An apparatus according to one of claims 4 to 11, wherein, for each of the four or
more audio output signals, the panning gain determiner ( 1 10) is configured to
determine the group of associated loudspeaker positions, being associated with
said audio output signal, so that the group of associated loudspeaker positions
comprises exactly the loudspeaker position being associated with said audio
output signal and two further loudspeaker positions of the proper subset, and so
that the panning position is located within a triangle or on an edge of said triangle
in the two-dimensional coordinate system, wherein each loudspeaker position of
the group of associated loudspeaker positions indicates a vertex of said triangle in
the two-dimensional coordinate system.
3 An apparatus according to claim 12, wherein, for each audio output signal of the
four or more audio output signals, the panning gain determiner ( 1 10) is configured
to select a first loudspeaker position, a second loudspeaker position, and a third
loudspeaker position of the four or more loudspeaker positions as the loudspeaker
positions of the group of associated loudspeaker positions, being associated with
said audio output signal, if in
l and m satisfy
l >
m > 0
C + m < 1
wherein l is a first real number, wherein m is a second real number,
wherein a is a first vector having two first vector components, wherein a indicates
the first loudspeaker position, wherein a first one of the two first vector components
indicates a first coordinate value of the first loudspeaker position within the twodimensional
coordinate system, and wherein a second one of the two first vector
components indicates a second coordinate value of the first loudspeaker position
within the two-dimensional coordinate system,
wherein b is a second vector having two second vector components, wherein b
indicates the second loudspeaker position, wherein a first one of the two second
vector components indicates a first coordinate value of the second loudspeaker
position within the two-dimensional coordinate system, and wherein a second one
of the two second vector components indicates a second coordinate value of the
second loudspeaker position within the two-dimensional coordinate system,
wherein c is a third vector having two third vector components, wherein c indicates
the third loudspeaker position, wherein a first one of the two third vector
components indicates a first coordinate value of the third loudspeaker position
within the two-dimensional coordinate system, and wherein a second one of the
two third vector components indicates a second coordinate value of the third
loudspeaker position within the two-dimensional coordinate system,
wherein p is a fourth vector having two fourth vector components, wherein p
indicates the panning position, wherein a first one of the two fourth vector
components indicates a first coordinate value of the panning position within the
two-dimensional coordinate system, and wherein a second one of the two fourth
vector components indicates a second coordinate value of the panning position
within the two-dimensional coordinate system,
14. An apparatus according to claim 12 or 13 , wherein the panning gain determiner
( 1 10) is configured to calculate, for each audio output signal of the four or more
audio output signals, the panning gain for said audio output signal depending on a
first distance being a shortest distance between the panning position and a first
straight line through the two further loudspeaker positions of the group of
associated loudspeaker positions, and depending on a second distance, being a
shortest distance between the loudspeaker position being associated with said
audio output signal and a second straight line through the panning position,
wherein said second straight line is parallel to said first straight line,
15 . An apparatus according to claim 1 or 13 ,
wherein the panning gain determiner ( 1 10) is configured to calculate, for each
audio output signal of the four or more audio output signals, the panning gain
depending of a ratio of a first distance and a sum of a first distance and the second
distance,
wherein the first distance indicates a shortest distance between the panning
position and a first straight line through the two further loudspeaker positions of the
group of associated loudspeaker positions, and
wherein the second distance indicates a shortest distance between the
loudspeaker position being associated with said audio output signal position and a
second straight line through the panning position, wherein said second straight line
is parallel to said first straight line
1 An apparatus according to claim 13,
wherein, for each of the four or more audio output signals, the panning gain
determiner ( 1 10) is configured to determine said panning gain according to the
formula
= 1 - n (p - a )
wherein g is said panning gain,
wherein n is a fifth vector being defined according to
n
- a )
wherein n' is a sixth vector being defined according to
wherein indicates the first coordinate value of the loudspeaker position, being
assigned to the second audio output signal,
wherein indicates the second coordinate value of the loudspeaker position,
being assigned to the second audio output signal,
wherein indicates the first coordinate value of the loudspeaker position, being
assigned to the third audio output signal, and
wherein indicates the second coordinate value of the loudspeaker position,
being assigned to the third audio output signal.
An apparatus according to one of the preceding claims,
wherein the audio input signal comprises a plurality of audio input samples, and
wherein the signal processor 120) is configured to generate each audio output
signal of the four or more audio output signals by multiplying each of one or more
of the audio input samples of the audio input signal with the panning gain for said
audio output signal to obtain one or more audio output samples of the audio output
signal
18 An apparatus according to one of claims 1 to 16,
wherein the audio input signal comprises a plurality of audio input samples, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals by multiplying each of one or more
of the audio input samples of the audio input signal with a square root of the
panning gain for said audio output signal to obtain one or more audio output
samples of the audio output signal.
An apparatus according to claim ,
wherein each loudspeaker position of the four or more loudspeaker positions of the
proper subset is associated with exactly one of the four or more audio output
signals, and wherein each of the four or more audio output signals is associated
with exactly one of the four or more loudspeaker positions of the proper subset,
wherein the panning gain determiner ( 1 10) is configured to calculate, for each
audio output signal of the four or more audio output signals, the panning gain for
said audio output signal depending on the panning position and depending on the
loudspeaker position of each of the four or more audio output signals.
An apparatus according to one of the preceding claims,
wherein the panning gain determiner ( 1 10) is configured to determine the panning
gain for each audio output signal of the four or more audio output signals for a first
point in time as a first time-dependent panning gain for said audio output signal,
and
wherein the panning gain determiner ( 1 10) is configured to determine a further
panning gain for each audio output signal of the four or more audio output signals
for a different second point in time as a second time-dependent panning gain for
said audio output signal, said second time-dependent panning gain being different
from the first time-dependent panning gain for said audio output signal.
An apparatus according to one of claims 1 to 19,
wherein the panning gain determiner ( 1 10) is configured to determine the panning
gain for each audio output signal of the four or more audio output signals for a first
frequency as a first frequency-dependent panning gain for said audio output signal,
and
wherein the panning gain determiner 1 10) is configured to determine a further
panning gain for each audio output signal of the four or more audio output signals
for a different second frequency as a second frequency-dependent panning gain
for said audio output signal, said second frequency-dependent panning gain being
different from the first frequency-dependent panning gain for said audio output
signal.
An apparatus according to one of the preceding claims, comprising:
wherein the audio input signal is a first audio input signal, wherein the panning
position is a first panning position, wherein the panning gain is a first input-signaldependent
panning gain, and wherein the proper subset is a first proper subset,
wherein the panning gain determiner ( 1 10) is configured to determine one or more
further proper subsets from a set of five or more loudspeaker positions, so that
each of the one or more further proper subsets comprises four or more of the five
or more loudspeaker positions,
wherein the panning gain determiner ( 1 10) is configured to determine each of the
one or more further proper subsets depending on one of one or more further
panning positions and depending on the five or more loudspeaker positions,
wherein the panning gain determiner ( 1 10) is configured to determine one or more
further input-signal-dependent panning gains for each of the four or more audio
output signals by determining each of the one or more further panning gains
depending on one of the one or more further panning positions and depending on
the four or more loudspeaker positions of one of the one or more further proper
subsets, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals depending on the first input-signaldependent
panning gain for said audio output signal, depending on the one or
more further input-signal-dependent panning gains for said audio output signal,
depending on the audio input signal, and depending on the one or more further
audio input signals.
A system comprising;
five or more loudspeakers, and
an apparatus according to one of the preceding claims,
wherein each of the five or more loudspeakers is associated with exactly one
loudspeaker position of the set of five or more loudspeaker positions,
wherein each of the four or more audio output signals is associated with exactly
one loudspeaker position of the four or more loudspeaker positions of the proper
subset, and wherein each of the four or more audio output signals is associated
with exactly one of the four or more loudspeaker positions of the proper subset,
wherein the system is configured to output each audio output signal of the four or
more audio output signals by the loudspeaker which is associated with the same
loudspeaker position as said audio output signal.
A system according to claim 23, wherein the system is configured to output none
of the four or more audio output signals by any of the four or more loudspeakers
not being associated with the same loudspeaker position as said audio output
signal.
A system according to claim 23 or 24, wherein the system is configured to not
output any of the four or more loudspeaker signals by at least one of the five or
more loudspeakers.
26. A method for generating four or more audio output signals, comprisingdetermining
a proper subset from a set of five or more loudspeaker positions, so
that the proper subset comprises four or more of the five or more loudspeaker
positions, wherein determining the proper subset s conducted depending on a
panning position and depending on the five or more loudspeaker positions,
determining a panning gain for each of the four or more audio output signals by
determining said panning gain depending on the panning position and depending
on the four or more loudspeaker positions of the proper subset, and
generating each audio output signal of the four or more audio output signals
depending on the panning gain for said audio output signal and depending on an
audio input signal.
A computer program for implementing the method of claim 26 when being
executed on a computer or signal processor.
An apparatus for generating four or more audio output signals, wherein each
loudspeaker position of four or more loudspeaker positions is associated with
exactly one of the four or more audio output signals, and wherein each of the four
or more audio output signals is associated with exactly one of the four or more
loudspeaker positions, wherein the apparatus comprises:
a panning gain determiner ( 1 10), and
a signal processor (120),
wherein the panning gain determiner ( 1 10) is configured to determine, for each
audio output signal of the four or more audio output signals, a group of associated
loudspeaker positions, being associated with said audio output signal, depending
on the loudspeaker position of each of the four or more audio output signals and
depending on a panning position, so that said group of associated loudspeaker
positions comprises the loudspeaker position being associated with said audio
output signal and at least two further loudspeaker positions of the four or more
loudspeaker positions, wherein at least one of the four or more loudspeaker
positions is not comprised by said group of associated loudspeaker positions,
wherein the panning gain determiner ( 1 10) is configured to calculate, for each
audio output signal of the four or more audio output signals, the panning gain for
said audio output signal depending on the panning position and depending on the
loudspeaker positions of the group of associated loudspeaker positions being
associated with said audio output signal, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals depending on the panning gain for
said audio output signal and depending on an audio input signal,
wherein the group of associated loudspeaker positions being associated with a first
one of the four or more audio output signals is not equal to the group of associated
loudspeaker positions being associated with a different second one of the four or
more audio output signals.
An apparatus according to ciaim 28, wherein, for each of the four or more audio
output signals, the panning gain determiner 1 10) is configured to determine the
group of associated loudspeaker positions, being associated with said audio output
signal, so that said group of associated loudspeaker positions comprises exactly
three of the four or more audio output signals.
An apparatus according to claim 28 or 29, wherein the panning position and each
loudspeaker position of the four or more loudspeaker positions each indicates a
position within a two-dimensional coordinate system.
An apparatus according to claim 30, wherein the panning position and each
loudspeaker position of the four or more loudspeaker positions each indicates an
azimuth angle and an elevation angle within the two-dimensional coordinate
system.
An apparatus according to claim 30 or 3 1, wherein, for each of the four or more
audio output signals, the panning gain determiner ( 1 0) is configured to determine
the group of associated loudspeaker positions, being associated with said audio
output signal, such that a group-specific polygon exists, wherein the vertices of the
group-specific polygon are the loudspeaker positions being associated with the
associated output signals of said group of associated loudspeaker positions, and
wherein the panning position is enclosed by said group-specific polygon.
An apparatus according to claim 32, wherein the group-specific polygon, being
determined for each of the four or more audio output signals, does not enclose any
of the four or more loudspeaker positions which is associated with an audio output
signal which is not comprised by said group of associated loudspeaker positions.
34. An apparatus according to claim 32 or 33, wherein a body defined by the groupspecific
polygon, being determined for each of the four or more audio output
signals, is convex.
An apparatus according to one of claims 30 to 34, wherein, for each of the four or
more audio output signals, the panning gain determiner ( 1 10) is configured to
determine the group of associated loudspeaker positions, being associated with
said audio output signal, so that the group of associated loudspeaker positions
comprises exactly said audio output signal and two further audio output signals of
the four or more audio output signals, and so that the panning position is located
within a triangle or on an edge of said triangle in the two-dimensional coordinate
system, wherein the loudspeaker position of each of the audio output signals of the
group of associated loudspeaker positions indicates a vertex of said triangle in the
two-dimensionai coordinate system.
An apparatus according to claim 35, wherein, for each of the four or more audio
output signals, the panning gain determiner ( 1 10) is configured to select a first
loudspeaker position, a second loudspeaker position and a third loudspeaker
position as the loudspeaker positions of the group of associated loudspeaker
positions, being associated with said audio output signal, if in
[A, m ] = [b - a . c - a] 1 (p - a) .
land m satisfy
l > o
m > 0
l + m < 1
wherein l is a first real number, wherein is a second real number,
wherein a is a first vector having two first vector components, wherein a indicates
the first loudspeaker position, wherein a first one of the two first vector components
indicates a first coordinate value of the first loudspeaker position within the twodimensional
coordinate system, and wherein a second one of the two first vector
components indicates a second coordinate value of the first loudspeaker position
within the two-dimensional coordinate system,
wherein b is a second vector having two second vector components, wherein b
indicates the second loudspeaker position, wherein a first one of the two second
vector components indicates a first coordinate value of the second loudspeaker
position within the two-dimensional coordinate system, and wherein a second one
of the two second vector components indicates a second coordinate value of the
second loudspeaker position within the two-dimensional coordinate system,
wherein c is a third vector having two third vector components, wherein c indicates
the third loudspeaker position, wherein a first one of the two third vector
components indicates a first coordinate value of the third loudspeaker position
within the two-dimensional coordinate system, and wherein a second one of the
two third vector components indicates a second coordinate value of the third
loudspeaker position within the two-dimensional coordinate system,
wherein p is a fourth vector having two fourth vector components, wherein p
indicates the panning position, wherein a first one of the two fourth vector
components indicates a first coordinate value of the panning position within the
two-dimensional coordinate system, and wherein a second one of the two fourth
vector components indicates a second coordinate value of the panning position
within the two-dimensional coordinate system.
37, An apparatus according to claim 35 or 36, wherein the panning gain determiner
( 1 10) is configured to calculate, for each audio output signal of the four or more
audio output signals, the panning gain for said audio output signal depending on a
first distance being a shortest distance between the panning position and a first
straight line through the two further loudspeaker positions of the group of
associated loudspeaker positions, and depending on a second distance, being a
shortest distance between the loudspeaker position being associated with said
audio output signal and a second straight line through the panning position,
wherein said second straight line is parallel to said first straight line.
38. An apparatus according to claim 35 or 36,
wherein the panning gain determiner ( 1 10) is configured to calculate, for each
audio output signal of the four or more audio output signals, the panning gain
depending of a ratio of a first distance and a sum of a first distance and the second
distance,
wherein first distance indicates a shortest distance between the panning position
and a first straight line through the two further loudspeaker positions of the group
of associated loudspeaker positions, and
wherein the second distance indicates a shortest distance between the
loudspeaker position being associated with said audio output signal position and a
second straight line through the panning position, wherein said second straight line
is parallel to said first straight line.
An apparatus according to claim 36,
wherein, for each of the four or more audio output signals, the panning gain
determiner ( 10) is configured to determine said panning gain according to the
formula
g = 1 - n ( p - a )
wherein g is said panning gain,
wherein n is a fifth vector being defined according to
n (b - a )
wherein n ' is a sixth vector being defined according to
wherein indicates the first coordinate value of the loudspeaker position, being
assigned to the second audio output signal,
wherein ( indicates the second coordinate value of the loudspeaker position,
being assigned to the second audio output signal,
wherein Qc indicates the first coordinate value of the loudspeaker position, being
assigned to the third audio output signal, and
wherein c indicates the second coordinate value of the loudspeaker position,
being assigned to the third audio output signal.
An apparatus according to one of claims 28 to 39,
wherein the audio input signal comprises a plurality of audio input samples, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals by multiplying each of one or more
of the audio input samples of the audio input signal with the panning gain for said
audio output signal to obtain one or more audio output samples of the audio output
signal.
An apparatus according to one of claims 28 to 39,
wherein the audio input signal comprises a plurality of audio input samples, and
wherein the signal processor (120) is configured to generate each audio output
signal of the four or more audio output signals by multiplying each of one or more
of the audio input samples of the audio input signal with a square root of the
panning gain for said audio output signal to obtain one or more audio output
samples of the audio output signal.
An apparatus according to one of claims 28 to 4 1,
wherein the panning gain determiner ( 1 10) is configured to determine the panning
gain for each audio output signal of the four or more audio output signals for a first
point in time as a first time-dependent panning gain for said audio output signal,
and
wherein the panning gain determiner ( 1 10) is configured to determine a further
panning gain for each audio output signal of the four or more audio output signals
for a different second point in time as a second time-dependent panning gain for
said audio output signal, said second time-dependent panning gain being different
from the first time-dependent panning gain for said audio output signal.
An apparatus according to one of claims 28 to 4 1,
wherein the panning gain determiner ( 1 10) is configured to determine the panning
gain for each audio output signal of the four or more audio output signals for a first
frequency as a first frequency-dependent panning gain for said audio output signal,
and
wherein the panning gain determiner 1 0) is configured to determine a further
panning gain for each audio output signal of the four or more audio output signals
for a different second frequency as a second frequency-dependent panning gain
for said audio output signal, said second frequency-dependent panning gain being
different from the first frequency-dependent panning gain for said audio output
signal.
An apparatus according to one claims 28 to 43, comprising:
wherein the audio input signal is a first audio input signal, wherein the panning
position is a first panning position, wherein the group of associated loudspeaker
positions for each audio output signal of the four or more audio output signals is a
first group of associated loudspeaker positions for said audio output signal, and
wherein the panning gain is a first input-signal-dependent panning gain,
wherein the panning gain determiner ( 1 10) is configured to determine, for each
audio output signal of the four or more audio output signals, one or more further
groups of associated output signals, being associated with said audio output
signal, wherein the panning gain determiner ( 1 10) is configured to determine each
further group of the one or more further groups depending on the loudspeaker
position of each of the four or more audio output signals and depending on a
further panning position, so that said further group of associated loudspeaker
positions comprises said audio output signal and at least two further audio output
signals of the four or more audio output signals, wherein at least one of the four or
more audio output signals is not comprised by said further group of associated
loudspeaker positions,
wherein the panning gain determiner ( 10) is configured to calculate, for each
audio output signal of the four or more audio output signals, one or more further
input-signal-dependent panning gains for said audio output signal depending on
the further panning position and depending on the loudspeaker position of each
associated output signal of one of the one or more further groups of associated
output signals, and
wherein the signal processor 120) is configured to generate each audio output
signal of the four or more audio output signals depending on the first input-signaldependent
panning gain for said audio output signal, depending on the one or
more further input-signal-dependent panning gains for said audio output signal,
depending on the first audio input signal, and depending on one or more further
audio input signals.
A method for generating four or more audio output signals, wherein each
loudspeaker position of four or more loudspeaker positions is associated with
exactly one of the four or more audio output signals, and wherein each of the four
or more audio output signals is associated with exactly one of the four or more
loudspeaker positions, wherein the method comprises:
determining, for each audio output signal of the four or more audio output signals,
a group of associated loudspeaker positions, being associated with said audio
output signal, depending on the loudspeaker position of each of the four or more
audio output signals and depending on a panning position, so that said group of
associated loudspeaker positions comprises the loudspeaker position being
associated with said audio output signal and at least two further loudspeaker
positions of the four or more loudspeaker positions, wherein at least one of the four
or more loudspeaker positions is not comprised by said group of associated
loudspeaker positions,
calculating, for each audio output signal of the four or more audio output signals,
the panning gain for said audio output signal depending on the panning position
and depending on the loudspeaker positions of the group of associated
loudspeaker positions being associated with said audio output signal, and
generating each audio output signal of the four or more audio output signals
depending on the panning gain for said audio output signal and depending on an
audio input signal,
wherein the group of associated loudspeaker positions being associated with a first
one of the four or more audio output signals is not equal to the group of associated
loudspeaker positions being associated with a different second one of the four or
more audio output signals.
46.A computer program for implementing the method of claim 45 when being
executed on a computer or signal processor