The present invention relates to a microphone array, in particular an arrangement of a plurality of microphone capsules that operate together to pick up sound.
Microphone arrays are frequently used e.g. for beamforming, for noise suppression or for searching an acoustic source. They comprise several microphone capsules, the output signals of which are electronically interconnected in order to work together for the directional recording of sound. The type of interconnection can produce a preferred direction in which the sensitivity of the microphone array for audio recording is particularly high. Due to the electronic combination of the individual microphone signals, this preferred direction can be adjusted electronically, which enables the preferred direction to be changed with a very short response time. However, a microphone array does not necessarily have equally good directional effects for all directions, but often has one or more fixed preferred directions that depend on the arrangement of the microphone capsules. In addition, microphone arrays do not work equally for all frequencies, but rather show a frequency dependency. This depends, among other things, on the distance between the microphone capsules. A very important aspect of a microphone array is therefore the geometric arrangement of the microphone capsules on the microphone surface: With a given number of microphone capsules, these should cover as many inter-element distances as possible, i.e. distances between individual microphone capsules of an array, in as many different directions as possible.
There are various strategies with regard to the type, number and positioning of the microphone capsules. Often, for example for microphone arrays that can be mounted on room ceilings, a large number of microphone capsules are combined with one another in order to be able to capture as many different preferred directions as possible and a specific frequency range. This is often the range of speech frequencies, e.g. 100 Hz-10 kHz. Typical approaches in this field are heuristic and therefore very time-consuming searches by “trying out” all conceivable analytically describable manifolds, such as lines, circles, spirals etc., and numerical simulation.
For example, in U.S. Pat. No. 6,205,224 B1 at least 63 sensor elements such as antennas or microphone capsules are arranged on concentric circles and at the same time on spirals in order to enable broadband detection that is largely independent of direction and that has a high degree of directivity. The directional and frequency properties of a sensor arrangement are indicated by means of a so-called coarray, which shows inter-element distances and the direction of these distances. In US2013/0101141 A1, which also aims at direction-independent and broadband detection, thirty microphone capsules are evenly distributed over the surface of a hexagonal circuit board, several of which can then be interconnected. In US2016/0323668 A1, too, numerous microphone capsules are interconnected to form a microphone array and are largely evenly distributed over several circuit boards. A central board contains 64 microphone capsules, while each of 7 boards arranged in a circle around it contains a further 8 microphone capsules, which leads to a total of 120 microphone capsules. In all these cases, the large number of microphone signals leads to a high computational effort and an overall large microphone array results.
Another strategy than in the documents mentioned is therefore to use as few microphone capsules as possible for an array. In order to reduce the noise in this case, or to obtain a high signal-to-noise ratio (SNR) respectively, the microphone capsules must have as little noise as possible, i.e. be of high quality. Furthermore, the electroacoustic properties of all microphone capsules in an array must be largely identical within tight tolerances. A number-theoretical approach to minimizing the number of microphone capsules and their positioning is pursued in DE10 2010 012388 A1, by positioning them at the intersections of Golomb rulers. Although this leads to a reduction in the number of microphone capsules, it also leads to a directional characteristic that is not uniform in all directions, due to the asymmetric distribution. Moreover, the microphone capsules are almost evenly distributed over the entire surface. In U.S. Pat. No. 9,894,434 B2, a microphone array with 17 microphone capsules is described which are arranged on the diagonals of a relatively large square area of approximately 60×60 cm. This size is typical for most of the arrays mentioned. Furthermore, most of the arrays mentioned have the problem that they are susceptible to even small incorrect positioning of the microphone capsules and cannot be scaled in size without disruptive non-linear effects occurring.
In the field of seismology, investigations into sensor arrays were carried out a long time ago. In the article “Array Design” by R. Haubrich in the “Bulletin of the Seismological Society of America”, Vol. 58, June 1968, it is described how arrays for the detection of seismic waves, ocean waves or electromagnetic radio waves can be constructed with as few sensors as possible. The directional and frequency properties of various sensor arrangements are assessed using coarrays. Various sensor arrangements that are considered to be “perfect” or “optimal” according to this criterion are proposed, including isometric arrays, the sensors of which are located at the intersections of an isometric coordinate system.
The invention is based on the object of providing a microphone array which is as small as possible and has as few microphone capsules as possible, but is more robust against small incorrect positioning of the capsules, has a high and direction-independent directivity and an essentially uniform frequency dependence over a speech frequency range, and which can be used as a ceiling microphone. This object is solved by a microphone array according to claim 1.
As it turned out, the structures of some of the arrays of seismic sensors proposed by R. Haubrich many years ago are also suitable for arrays of microphone capsules. In particular, isometric arrays with fifteen or twenty-one microphone capsules have particularly good acoustic properties, such as e.g. good localization of sound sources and a high level of directivity, as well as other advantages, e.g. low manufacturing costs. This applies even if the size of the arrays is scaled down according to the speech frequencies to be recorded, so that smaller arrays than before are possible.
According to the invention, a microphone array has a small number of microphone capsules, in particular fifteen or twenty-one microphone capsules, and a circuit arrangement which is connected to the microphone capsules and which is suitable for receiving the microphone signals and for processing them together. The microphone capsules are arranged in a plane at certain positions on a carrier board, namely on three similar (i.e. in principle identical) branches each with the same number of microphone capsules, the branches being rotated by 120° from one another around a common center, and wherein, in a flat isometric coordinate system with three axes rotated by 120° against each other which form a so-called L2-grid (L2-lattice) of equilateral triangles, each of the microphone capsules lies on a corner of a triangle of the L2-grid. The use of electret capsules is particularly advantageous, since they typically have less inherent noise and lower self-resonance and can cover a higher sound pressure level range. However, other microphone capsules can also be used, e.g. MEMS.
One advantage of the array according to the invention is the good and uniform directivity over the entire relevant speech frequency range and in all directions, as can be calculated using coarrays. However, further advantages of the array according to the invention also include the relatively high robustness with respect to small incorrect positioning of microphone capsules, the small size of the array and thus lower costs, as well as the relatively free possibility of size scaling.
Further advantageous embodiments are disclosed in the claims 2-11.
Further details and advantageous embodiments are depicted in the drawings, showing in
In a first embodiment of the invention,
Positions in the isometric coordinate system are specified as multiples of the side lengths of the equilateral triangles. For example, the upper right corner of the center triangle DM is shifted from the reference point R by only one side length in the direction of the L1 axis, which is specified in isometric coordinates as position (L0,L1,L2)=(0,1,0). Correspondingly, the upper left corner of the center triangle DM is shifted from the reference point R by only one side length in the direction of the L2 axis, i.e. at the position (L0,L1,L2)=(0,0,1) in isometric coordinates. Starting from the reference point R, the microphone capsules are at the following positions:
In Cartesian coordinates (X,Y), this results in approximately the following values, depending on the scale (for example for a side length of the triangles, or isometric length unit respectively, of 0.05 m, as shown in
The scale is to be chosen such that the smallest distance between two microphone capsules corresponds to a side length of a triangle of the isometric coordinate system. Thus, the microphone array depicted in
The positions apply to the coordinate systems indicated in
The coarray of the microphone arrangement according to the invention has the advantageous property that (at least in the inner region of the coarray) each coarray point has six neighboring points arranged evenly around it, each at the same distance. This allows the size of the microphone arrangement to be scaled to the wavelengths of interest. The coarray points with the smallest distance to the origin (smallest inter-element distances) indicate the highest frequency that is spatially clearly resolvable, before undersampling begins, i.e. below the so-called spatial aliasing. The coarray points with the greatest distance to the origin correspondingly determine the beamformer's performance for low frequencies. As a result, the smallest inter-element spacing of the microphone arrangement can be scaled to the smallest wavelength or highest frequency of interest, while the closest possible coverage of all wavelengths is maintained for all larger inter-element spacings or larger wavelengths, respectively. For example, scaling the microphone arrangement of the first or second embodiment to a diameter of 35 cm (L=5 cm) results in a highest frequency (below spatial aliasing) of approximately 6.9 kHz.
One advantage of the invention is that the microphone capsules are not evenly distributed over the entire area of the array, but rather from groups. This means that relatively large parts of the surface do not have to be covered by circuit boards or printed circuit boards for contacting the capsules. In particular, it is not necessary to provide a circuit board or group of circuit boards in the size of the entire arrangement. This further reduces the manufacturing costs for the array, which are relatively low due to the small number of microphone capsules, and its weight. In addition, since the microphone capsules are distributed over three congruent branches, equal circuit boards can also be used for each branch.
The microphone capsules can be distributed very compactly, for example on two boards per branch. One option for one of the circuit boards P21,1 with five capsules K21,11-K21,15 of the first branch is depicted in
Note that only a relative scale is indicated in
Corresponding relationships with regard to scalability also apply to the other embodiments. For example, rmax=L*3.512, D=L*7.024 and dmax=L*6,557 (rounded) applies to the first and second embodiment.
The microphone capsules can in this case be distributed to one circuit board per branch or, because of the small overall size, they can all be mounted on a single circuit board P6. Resulting values are (rounded) rmax=L*1.527, D=L*3.054 and dmax=L*2.646.
Because all microphone capsules of a branch are attached together on a circuit board or group of circuit boards and the positioning of the circuit boards on the carrier T can also take place with very little deviation, the relative position of the capsules to one another is very accurate. The carrier may comprise, e.g., one or more solid or sound-reflecting plates made of metal, plastic or the like. In an embodiment, the carrier is a metal or plastic plate with holes through which the sound can reach the microphone capsules (in the ceiling microphone from the bottom when installed). The plate in that case is sound reflecting, so that the sound pressure level at the microphone capsules is increased by up to 6 dB and the array works as a boundary microphone. On the other hand, the arrangement of the microphone capsules according to the invention allows small deviations from the predefined position of up to 0.5 mm, for example, which makes assembly easier and therefore cheaper. Conventionally, a higher degree of accuracy is necessary here in order to achieve a certain audio quality. The microphone capsules can also be mounted on at least two groups of three identical (sub-) boards PCB1,1-PCB3,2 each, with one board of each group belonging to each branch. Each (sub-) board may comprise at least two microphone capsules. A middle region of the array between the three rotated boards or groups of boards may comprise no board, or a board without a microphone capsule. Alternatively, there may also be an additional microphone capsule in the middle, which increases the total number of capsules. The other positions remain unchanged. Thus, the modified first and second embodiments have sixteen microphone capsules, the modified third embodiment has twenty-two capsules and the modified fourth embodiment has seven capsules. Such center capsule has the advantage that it acquires a sound signal at the position of the highest sound pressure (dynamic pressure) and thus improves the directivity and the SNR for the entire array. However, such additional central capsule is not located on a point of the L2-lattice and therefore leads to an unsymmetric coarray with holes, so that the array gets an uneven directivity and is not easily scalable in size anymore.
Electret capsules are particularly suitable as microphone capsules. Each microphone signal may be corrected or normalized individually, e.g. by means of filtering in the individual digital processing blocks DP1-DP5. The corresponding filtering parameters depend on characteristics of the respective microphone capsule, for example its phase response and frequency response. Therefore, in particular such electret capsules are well suited that have an internal memory element with corresponding correction data from which filter parameters may be determined. In addition, the filter parameters can be influenced by the examined or detected direction of the sound source (i.e. the localization of the sound sources or the beamforming). The localization of sound sources and the actual recording of sound from the main sound source can be two separate processes. It is possible to use only some of the microphone capsules for the localization in order to keep the processing effort low while using all capsules for the actual sound recording.
An advantage of the microphone arrays according to the invention is the good directivity and the high SNR, i.e. a good noise suppression. Noise suppression is the more difficult the less microphone signals are available. However, this relationship is non-linear, depending, among other things, on the positions of the microphone capsules, and therefore difficult to predict. In particular the microphone arrays according to the invention that have fifteen or twenty-one microphone capsules show a good and uniform directivity over all relevant frequency components and directions of incidence of the sound, or a very good noise suppression given the small number of microphone capsules, and are particularly well-suited for ceiling mounted microphones.
Number | Name | Date | Kind |
---|---|---|---|
5377166 | Kuhn | Dec 1994 | A |
6205224 | Underbrink | Mar 2001 | B1 |
8213634 | Daniel | Jul 2012 | B1 |
8787114 | Plotke | Jul 2014 | B1 |
9143879 | McElveen | Sep 2015 | B2 |
9894434 | Rollow, IV et al. | Feb 2018 | B2 |
20130101136 | McElveen | Apr 2013 | A1 |
20130101141 | McElveen | Apr 2013 | A1 |
20150117672 | Christoph | Apr 2015 | A1 |
20160323668 | Abraham | Nov 2016 | A1 |
20180139535 | Cengarle | May 2018 | A1 |
Number | Date | Country |
---|---|---|
10 2010 012388 | Nov 2010 | DE |
2006308409 | Nov 2006 | JP |
Entry |
---|
Li, Computational Acoustic Beamforming of Noise Source on Wind Turbine Foil (Year: 2014). |
International Search Report and Written Opinion dated Aug. 12, 2022 in the corresponding International application No. PCT/EP2022/060420. |
Number | Date | Country | |
---|---|---|---|
20220353608 A1 | Nov 2022 | US |