RU2185710C2 - Method and acoustic transducer for electronic generation of directivity pattern for acoustic signals - Google Patents

Abstract

FIELD: acoustics; beam generation by acoustic signals. SUBSTANCE: predetermined gain characteristic between received electric signal S_r and acoustic signals depending on direction (θ) wherefrom mentioned acoustic signals arrive at two acoustical- electrical transducers 1, 2 spaced apart are shaped so that repeated relatively delayed signal (A₁₀) is found from output signals of transducers and in compliance with assumed transducer delay. In the process one (S₁) out of output signals is filtered off so that transient characteristic of filtration is controlled depending on relatively delayed signal (A₁₂). Output filtration signal 14 is used as received electric signal (S_r). EFFECT: improved frequency response, reduced size, enhanced signal-to-noise ratio. 30 cl, 26 dwg, 1 tbl

Description

 The present invention generally relates to the so-called "beamforming" method by acoustic signals.

 The use of directional acoustoelectric transformations and especially such microphones is one of the most effective ways to improve the signal-to-noise ratio in audio systems. A known method for the implementation of directional microphones using a matrix of microphone cells and a time delay, as well as superimposing the output signals of such cells, which is called the method of "delay and summation".

Таким образом, время τ задержки обычно выбирают равным отношению расстояния р и скорости звука с. Такое расположение приводит в итоге к кардиодной характеристике направленности первого порядка (фиг.2). Можно показать, что амплитуда результирующего сигнала А_r, пропорциональна синусу частоты

сигнала и расстоянию р. Максимальное усиление в направлении 180^o цели происходит на частоте f_r=с/(4р). При расстоянии р, равном 12 нм, частота f_r становится приблизительно равной 7 кГц.This known principle with two omnidirectional microphone cells is shown in FIG. Omni-directional microphones 1 and 2 are located at a distance p. The output signal of the microphones corresponding to the signal A ₁ is delayed in time by the amount of time τ, while the time-delayed signal corresponding to A ₁ is superimposed in the overlay unit 3 on the unavailable output signal A _{2 of the} microphone 2. At the output of the overlay unit 3, the resulting output signal A _r with gain depending on the characteristic of the angle of incidence θ, as shown in FIG. 2, for one frequency under consideration

Thus, the delay time τ is usually chosen equal to the ratio of the distance p and the speed of sound c. This arrangement ultimately leads to a cardioid characteristic of the directivity of the first order (figure 2). It can be shown that the amplitude of the resulting signal A _r is proportional to the sine of the frequency

signal and distance p. The maximum gain in the direction of 180 ^o target occurs at a frequency f _r = s / (4P). At a distance p equal to 12 nm, the frequency f _r becomes approximately equal to 7 kHz.

With the location of more than one device with a double cell (Fig. 1) in a checkerboard pattern and superposition of the resulting signals A _r from more than one device with a double cell, one can obtain a high-order cardiodic directivity characteristic.

In FIG. 3 shows a known device diagram in which second-order cardioid directivity characteristics are realized (FIG. 4). Thus, a narrower beam ("radiation pattern") can be obtained. The higher the order of the directional microphone device, the higher the directivity and gain index at the frequency f _r , and the steeper the bend for low and high frequencies and the higher the number of unwanted side lobes. Regarding the determination of the directivity index, you can refer to voice communication 20 (1996), 229-240, “Microphone array systems for hands-free telecommunications” Garry W. Eiko .

In FIG. 5 shows the dependence of the gain on the frequency response of the first and second cardioid directivity characteristics for the angle of incidence θ = 180 ^° . High-frequency and low-frequency bends are clearly visible from it.

 Such beamforming methods are well known and implemented using the processing of analog signals, for example, in patents US-A-2237298, US-A-4544927, US-A-4703506, US-A-5506908, or using digital signal processing in time or frequency domain, for example, in patents EP-A-0381498 (time domain) or in US-A-5581620 (frequency domain).

Formation methods implemented using any of these approaches have the following disadvantages:
a) The resulting signal attenuates at low frequencies, resulting in a low signal-to-noise ratio.

 b) The directivity index is very sensitive to the matching of individual microphone cells, especially at low frequencies.

 c) The distance p between the microphone cells should be large (> 12 mm) for the sound frequency range.

 d) The frequency range with a high gain in the direction of the target is too small, as can be clearly seen from FIG.

 e) Directivity greatly depends on the number of microphone cells and thus on the complexity of the entire installation.

 f) As a result of high directivity, more unwanted side lobes appear due to an increase in the number of cells.

 Several methods have been proposed to address some of these shortcomings.

 WO 95/20305 (E. Lindemann) proposes an adaptive system for noise reduction and use in a stereo hearing aid. It detects the strength of the received signals to separate desired signals from unwanted ones.

 A grid of microphone cells with a “wide side” is proposed, that is, a direction to the target perpendicular to an axis passing from one microphone to another, as opposed to being “in line”, corresponding, for example, to FIG. 1 and the principles of the present invention.

 The disclosed device is bulky (> 5 cm), so it is not suitable for a hearing aid designed for one ear.

 Two equal lobes of the radiation pattern are generated in the direction of the target and in the opposite direction.

 In such a hearing aid, a connection must be provided between the left and right hearing systems, which makes the hearing aid inconvenient. In addition, as described by the same author in Two Microphone Non-linear Frequency Domain Beam former for hearing aid noise reduction 1995, IEEE ASSP Workshop on applications of Signal Processing to Audio and Acoustics, October 15-18, Mohonk, New Paltz, New York), such beamforming ("radiation patterns") is effective only up to about 2 kHz and leads to distortion of the necessary signals.

 US-A-4,653,102 proposes the use of two directional microphones aimed at a target, and a third microphone aimed in the opposite direction. The signal of the third microphone, presumably containing only noise, is used to formulate the characteristics of the two main microphones. This method, obviously, has a disadvantage inside reverberating rooms where the necessary signal is reflected from walls, floors, ceilings and furniture, and therefore is considered as noise. In addition, this method is inconvenient, since it requires the use of at least three microphones.

 Additional attention is given to patents US-A-5400409 and 5539859.

 As an example of known methods that generate radiation patterns in US Pat. No. 5,539,859, a method is proposed in which the reception characteristic is recorded in the direction from which the sound wave with the highest energy falls onto a pair of microphones and is considered in a sound medium. The fundamental point is that all sound falling from directions other than the directions of sound with the highest energy is considered as noise, and its reception is canceled.

 Thus, the analog-to-digital conversion and the subsequent conversion of the time domain into the frequency domain is performed on the output signals of two microphones. Using the knowledge of a fixed mutual distance between two microphones, on which the phase difference of the spectrum of the incident signal depends, one can determine the mutual phasing and thus the direction of incidence of sound signals with the highest energy, that is, the direction to the sound source with the highest energy within the acoustic environment . Signals coming from this direction are amplified by phase shift and summation, similar to the autocorrelation method, whereby signals incident at other incidence angles are suppressed as noise.

 Thanks to this method, the distribution of energy in a sound environment captures the selectivity of reception, and it is not possible to freely select or pre-select the maximum response characteristic, for example, in the direction from which the sound should selectively come, regardless of its relative energy. One area where such selectivity, regardless of the distribution of energy within the sound environment will have an obvious advantage, is the auxiliary listening method.

 An object of the present invention is to perform a method of electronically generating a predetermined gain characteristic depending on the direction from which acoustic signals are supplied to at least two spatially separated acoustoelectric transducers and a corresponding acoustic sensor device with which only a small number of microphones or microphone cells are to be used, and which is thus an implementation, which is a small compact directional m transducer or microphone. Thus, a preferred device according to the present invention is a hearing aid, and in particular a hearing aid intended for one ear.

 Another task is to perform such a method and device with a good frequency response in the range of sound frequencies, for example in the range of 0.1-10 kHz.

 Another objective of the present invention is to implement such a method and device that allows to obtain an implementation with a high signal to noise ratio without undesirable side lobes and with an easily changeable radiation pattern, for example, for acoustic magnification.

 These and other tasks are implemented using the method of the invention, which includes the steps of: periodically determining from the signals depending on the acoustic signals, the corresponding mutual delay signal corresponding to the delay when receiving at least two transducers, they subject the signal depending on the output signal at least two converters for filtering using the filtering transmission characteristic, and controlling the filtering transmission characteristic depending on the mutual delay signal, further and polzujut signal which depends on the output signal filtering as electrical reception signal.

 To accomplish the above tasks, the acoustic sensor device of the invention comprises at least two acoustoelectric transducers located at a predetermined mutual distance in the direction of the target, a time delay detection unit that has at least two inputs and an output, and its inputs are respectively connected in working condition to the outputs of two converters, whereby the time delay detection unit generates an output signal depending on the time delay acoustic signals incident on at least two spatially separated transducers, preferably a time-domain to frequency-domain converting unit generating an output signal from a time-delay detection unit in a private area, a weighing unit with a predetermined weighing characteristic and with input and output, while the input is connected in working condition to the output of the time delay detection unit and receives a signal from the specified output of the time delay detection unit in having a frequency domain using a filtering unit with a controlled transmission characteristic that has at least one input, a control input and an output, and then the input is connected in working condition to at least one of the outputs of at least two converters, preferably through at least at least one converter of the time domain into the frequency domain, while the control input is connected in working condition to the output of the weighing unit, a filtering unit generating an output signal depending on its input one signal and its transmission characteristic, which is controlled by a signal (preferably a spectral signal) that is supplied to the control input of the filtering unit, and this weighting (preferably spectral weighting) of the resulting signal depends on the output signal of the time delay detection unit and the block weighing characteristic weighing.

Brief Description of the Drawings
Other objectives, advantages and specific embodiments of the present invention is illustrated by reference to the accompanying drawings, in which:
FIG. 1 depicts a functional block diagram of a two-cell directional microphone in accordance with a principle known in the art of “delay and sum”;
FIG. 2 shows a first-order cardioid gain characteristic of the prior art of FIG. 1;
FIG. 3 shows the difference between a device circuit corresponding to the “delay and sum” method for realizing a second-order characteristic from a device circuit known from the prior art in FIG. 1;
FIG. 4 depicts a second-order cardioid gain characteristic that is implemented using a circuit of a device known in the art of FIG. 3;
5 depicts a gain characteristic of a device as a function of frequency, according to FIG. 1 or 3, at an angle of incidence for maximum amplification of acoustic signals;
6 depicts a simplified functional block diagram of a device in accordance with the invention, operating according to the method in accordance with the invention, and further shows a signal processing sequence;
FIG. 7 shows, in accordance with FIG. 6, a first preferred embodiment of a device in accordance with the invention, which operates according to the method in accordance with the invention;
FIG. 8 depicts an additional preferred embodiment of a time delay detection unit of a device in accordance with the invention that operates in accordance with the method of the invention of FIG. 6;
Fig.9 depicts a graph in polar coordinates of the signals in accordance with the embodiment of Fig.8 to explain the operation of the comparison unit, which is provided in the embodiment of Fig.8;
FIG. 10 is a graph of comparison results versus angle of incidence of an acoustic signal according to the embodiment of FIG. 8;
11 depicts a preferred form of implementation of the signal obtained by overlay, which depends on the angle of incidence of the acoustic signal in the embodiment of FIG.
FIG. 12 depicts in a view corresponding to FIG. 10 a diagram of the comparison results that are realized according to the preferred embodiment and are the final dependence of FIG. 11;
Fig.13 schematically depicts in polar coordinates the dependence of the signals obtained as a result of the superposition on the angle of incidence of the acoustic signals and on the frequency that is realized according to the embodiment of Fig.8;
FIG. 14 depicts a preferred embodiment of the embodiment of FIG. 8 with further suppression of the frequency dependence, which is shown in FIG. 13;
FIG. 15 shows in a view corresponding to FIG. 13 the dependence of the normalized signal obtained as a result of the overlay according to the embodiment (FIG. 14) as a function of the frequency function of the first preferred normalization;
Fig.16 depicts a view corresponding to Fig.15, implemented using the frequency function of the second preferred normalization, according to the embodiment of Fig.14;
FIG. 17 depicts a first (solid line) and a second (dashed line) form of a preferred implementation of the amplitude characteristic of a filter according to an embodiment (FIG. 6 or 7);
FIG. 18a shows the effect of the amplitude characteristic of the filter depending on the amplitude characteristic of the transmission, according to FIG. 17 (solid line), on the output of the delay detection unit, which is made according to the embodiment (FIG. 6 or 7);
Fig. 18b depicts the output signal of the time delay detection unit, which passes through an amplitude filter with a transition characteristic corresponding to Fig. 17 (solid line) and implemented according to an embodiment (FIG. 6 or 7);
FIG. 19 depicts a spectrum of an acoustic signal that is converted into an electric signal and introduced into a filter with a controlled frequency, which is carried out using the present invention (FIG. 6);
FIG. 20 depicts a received electrical signal, which is realized by means of an amplitude filter characteristic, in accordance with FIG. 17 (solid line), and a reception signal, which is shown in FIG. 19 in the embodiment of FIG. 6;
FIG. 21 depicts the resulting dependence of the gain on the angle of incidence of the acoustic signal, which is realized by the characteristic of the amplitude filter of FIG. 17 (solid and dashed lines);
FIG. 22 depicts the gain depending on the characteristic of the angle of incidence, as presented in the embodiments of FIGS. 6 or 8, 14 of the invention, using the characteristics of the amplitude filter with the nature of the change in spectral amplitude transmission from maximum to minimum;
FIG. 23 is a simplified signal / functional block diagram of a further preferred embodiment of the invention;
Fig.24 depicts a functional diagram with the passage of the signal to work in an additional mode of implementation of the detection unit time delay (Fig.6) and
Fig.25 depicts a functional diagram with the passage of the signal to work in an additional implementation mode of the detection unit time delay, according to the method (Fig.8 or Fig.14).

Detailed Description of a Preferred Embodiment
In FIG. 6 are shown in the form of a functional block diagram together with the signal processing circuit diagrams of the method and apparatus in accordance with the invention.

At least two acoustoelectric transducers 1 and 2, in particular microphones or microphone cells, are located at a predetermined distance p from each other along the axis a. The acoustic signals IN are received using transducers 1 and 2, since they fall from different spatial directions θ. Acoustic signals IN have a frequency spectrum that varies over time. The output signals of the transducer 1, S ₁ (t, ω) and the transducer 2, S ₂ (t, ω) are formed in the form of electrical signals at the output of the transducers 1 and 2. Due to the mutual distance p of the two transmitters 1 and 2, which is preferably less than 5 cm , preferably between 0.5 and 1.5 cm, in particular for a sensor in accordance with the invention, which is a hearing aid for one ear, and as shown by two corresponding indicating circuits located below the functional block diagram (Fig.6 ), the acoustic signals IN fall on the transform ers 1 and 2 with a delay time dt which can be expressed by the phase difference Δφ _ω at each specified frequency ω as
Δφ _ω = ω • dt _ω , where (1)
dt _ω = p / c cosθ _ω . (2)
If the source of the acoustic signal IN is a point source, then the time delay dt _ω becomes equal for all spectral components at the frequency ω. The output signals S ₁ and S _{2 of the} converters 1 and 2 are connected in working condition to the corresponding inputs of the time delay detection unit 10, which generates an output signal A ₁₀ corresponding to the spectral distribution of the delays dt _ω in time, which are, as explained above, a function of angle of incidence θ, at which the corresponding frequency components fall on the converters 1 and 2 and thus in fact at an angle θ _ω . 6 only by way of example also shows a possible spectrum of the output signal A ₁₀ . This spectrum changes in time in accordance with the change in time of the incident acoustic signal IN. The output signal A ₁₀ of the time delay detection unit 10 is input to the weighing unit 12. Since the spectrum dt _ω with the corresponding spectral amplitudes A ₁₀ is introduced into the weighing unit 12 with the preselected weighing transition characteristic W, at a certain point in time in the form of the output signal A ₁₂ this leads to the spectral signal W (ω), which is also shown in FIG. .6. A _{12 is} obtained from the corresponding weighting of the spectral amplitudes A ₁₀ in accordance with the characteristic W. As A ₁₀ shows in accordance with dt _ω , from which direction θ _ω each frequency component of the acoustic signal IN falls, its specific weighting by the function W is zero in comparison with a predefined in which of the directions θ _{ω of} incidence this signal will amplify or weaken. Thus, the weighing unit 12 determines with its characteristic W the type of radiation pattern (“beam shape”).

The output signal A ₁₂ enters the filtering unit 14 with a controlled transient response of the filtering. In it, each spectral line of the spectrum, which depends on time, of the output signal S ₁ (t, ω) is amplified or attenuated in accordance with the control spectrum W _ω • A _10ω . Thus, block 14 is a filtering block for the input signal S ₁ , at which the transient response is changed, which is controlled by A ₁₂ . Depending on the type of filter unit 14, the weighing unit 12 generally calculates the adjustment of the filter characteristics, which determine the coefficients depending on A ₁₀ .

Thus, along the channels 10 and 12, it is determined, using the transfer functions W of the weighting, which spatial directions θ will be “targeted”. In the filtering unit 14, this beamforming information is supplied to the electrical analog S _{1 of the} acoustic signal IN, thereby obtaining an output signal S _r (t, ω) that represents the generated received signal.

By adjusting the transfer function W of the weighing by supplying a control signal C _{w to the} control input C ₁₂ , it is possible to adjust the beam shape and thus realize acoustic magnification.

 As shown by the dashed line, the controlled filtering performed in block 14 can be useful for the two output signals of the converter.

In FIG. 7 shows a first preferred embodiment of the principle of the invention (FIG. 6). The output signals S ₁ and S _{2 are} converted first from analog to digital in the corresponding analog-to-digital converters 16 and 17. The digital output signals of the respective converters 16 and 17 are input into the corresponding complex time domain converters 18 and 19 into the frequency domain.

The output spectra S ₁ (t, ω) and S ₂ (t, ω) of the transducers 18, 19 are input to the spectral time delay detection unit 10 '. Block 10 'calculates in accordance with formula (1) the phase difference spectrum Δφ _ω divided by the corresponding frequency ω ,, resulting in the output signal spectrum A ₁₀ ' in accordance with the time delay dt _ω , as explained with reference to FIG. .6. The output signal (A ₁₀ ′) of the time delay detection unit 10 ′ is further processed, as explained with reference to FIG. 6, by means of a weighted filtering unit 12 and a controlled filtering unit 14. The table shows examples of how unit 10 'works. From the spectral-phase distribution φ _{1n of the} signal S ₁ and φ _{2n of the} signal S _{2, the} time delay dt _{ω is} calculated for each spectral line within the spectral range of interest (see table at the end of the description).

In order to extract phase information φ from two signals S ₁ and S _{2 of} blocks 18 and 19 converting the time domain into the frequency domain, a complex (real and imaginary) operation is performed.

 A second preferred embodiment of the present invention, and especially one that relates to the implementation of the time delay detection unit 10, will be explained using FIGS. 8 and 9.

The output signal of one of the converters, for example, converter 1, S ₁ (t, ω), is supplied to the time delay unit 20, in which, in the first form of this implementation, the signal S _{1 is} delayed in time using a predetermined time frequency τ independent of .
Thus, returning again to FIG. 1, signal S ₁ corresponds to signal A ₁ . The output signal of the time delay unit 20 thus corresponds to the signal A ₁ ^′ (FIG. 1).

The time delay signal corresponding to A ₁ ′ is superimposed on the output signal S ₂ (t, ω), which is supplied from the converter 2 to the overlay unit 23, which corresponds to the block 3 (FIG. 1), thereby obtaining an output signal corresponding to _Ar (t, ω) (Fig. 1). As shown above and as explained with reference to FIG. 1, the output signal _Ar (t, ω) depends on the direction θ of the incidence of the acoustic signal corresponding to the beam of the first-order cardiodes (FIG. 2), the cardio function of which never changes with frequency ω. The output signal A _r of the overlay unit 23 and, for example, the output signal S ₂ (t, ω), which comes from the transducer 2, is input to the ratio generating unit 25, which is a comparison unit.

To understand the functioning of block 25 for generating a relationship, reference is made to FIG. 9. Figure 9 shows the attenuation characteristic of the cardiodes of the output signal A _r at a specific spectral frequency ω ₁ . At a specifically considered angle of incidence θ _o, the output signal A _r of the overlay unit 23 is A _ro (ω ₁ ) with an amplitude value (shown in FIG. 9). At the same time, at this frequency of interest ω ₁ and this angle of incidence θ _o considered, the amplitude of the signal S ₂ is equal to A _2o (ω ₁ ), which is also shown in FIG. 9. It should be emphasized that since the amplitude A _2o changes, the amplitude A _{ro also} changes proportionally. Thus, the ratio of A _ro to A _2o (Fig. 9) shows the angle θ _{o of} incidence. In the division block 25 (Fig. 8), for each amplitude of the spectral component, a ratio A _r to A _ro is formed, from where the signal spectrum is obtained at the output of the division block 25 with the ratio spectrum. Thus, the spectrum A ₁₀ (Fig.6) becomes the spectrum of the amplitude ratio, which nevertheless shows the angle of incidence θ, at which each frequency component of the spectrum of the acoustic signal falls relative to the axis of the two transducers (see Fig.6). In Fig. 8, a block indicated by a dashed line shows a delay detection unit 10 (Fig. 6). Additional signal processing is performed, as explained above, by means of FIG. 6, that is, through the weighing unit 12 and the controlled filtering unit 14.

 In this embodiment, it is possible to perform the conversion of the time domain to the frequency domain on the output side of the comparison unit 12.

Thus, the output signal of the ratio of block 25 is a measurement of the time delay dt _ω and is input to the weighing block 12.

In FIG. 10 shows a graph of the ratio of A _r to A ₀ as a function of θ at a specific frequency ω ₁ .
This ratio of amplitudes is shown for τ in block 20 (Fig. 8), which is chosen equal to
τ = p / c,
where p is the distance between transmitters 1 and 2 and c is the speed of sound.

When choosing τ equal to p / s, and as can be seen from the cardiodes beam function (FIG. 2), the signal attenuation or attenuation for θ equal to approximately 0 ^o becomes very high.

Thus, in this region of the angle θ of incidence, any choice of noise in A ₂ according to S ₂ (Fig. 8) will distort the comparison result obtained in block 25. This problem can be eliminated by choosing the delay τ, which is different and preferably more than r / s.

In FIG. 11 shows a graph of the resulting cardiodes for τ = 1.2 p / s, while FIG. 12 shows, similarly to FIG. 10, a graph of amplitude A _r divided by amplitude A ₂ .

In addition, it should be noted that the cardio function, which is shown in FIGS. 2, 9 and 11, is valid for only one specific frequency in question. Indeed, considering the different frequencies, the cardiode changes as shown in Fig. 13, in which the amplitude A of the output signal of the overlay unit 23 (Fig. 8) is shown for p = 12 mm, delay τ = 42 ms, and for frequencies 0.5, 1, 2, 4 and 7.2 kHz. From this graph in polar coordinates, the functional dependence of the frequency on the gain of the cardiodes is clearly visible. Although this dependence can be neglected to a first approximation, in a preferred form of implementing the method of the invention it is necessary, as shown in Fig. 8, to take into account such a dependence. Thus, a preferred embodiment of the method (Fig. 8) is shown in Fig. 14. In this case, the same positions are used as in Figs. 7 and 8. The output signals of the converters 1 and 2 are digitized using the corresponding analog-to-digital converters 16, 17, and the resulting digital signal of the converter 1 is delayed by a time delay τ, which is much larger than the p / s ratio. The output signal S _{2 of the} converter 2 is further converted to the frequency domain using the linear (non-complex) time conversion unit 18 ′ to the frequency domain, while the output signal A _r of the overlay unit 23 is converted to the frequency domain in the linear time to frequency domain conversion unit 19 ′. The frequency dependence on the graph in polar coordinates (Fig. 13) is taken into account using block 30 of the normalizer, which is actually a filter. In the first embodiment, the transient response of the filter is selected proportional to 1 / ω. This leads to the frequency dependence of the graph in polar coordinates, as shown in Fig. 15, for the same distance and frequency values, as shown in Fig. 13.

It can be seen that good agreement is achieved for small angles θ and frequencies up to about 4 kHz. At a frequency of 4 kHz, the deviation (deviation) is approximately 10% at θ = 0 ^° .
In addition, even an improved normalization function or filtering characteristic in block 30 (Fig. 14) is achieved when the filter characteristic is selected as a function 1 / sin (ω). The result is shown in FIG. The characteristics are in good agreement with the frequencies of 0.5-4 kHz. Another advantage of this normalization method is improved sensitivity in the opposite direction. This improved sensitivity can be used to form an adapted beam, that is, to selectively suppress noise sources from the rear.

 For specialists, it is obvious that such normalization can also be performed with a signal path of 1-23 and / or 2-23.

 In this embodiment (FIG. 14), it is highly preferred that only one-dimensional complex time domain converters 18 ′, 19 ′ to the frequency domain and non-complex converters are used, as in the embodiment (FIG. 7).

In FIG. 24 shows in block diagram form that the signal A ₁₀ (dt _ω ) can also be generated as an output signal of a comparison unit 60, in which, on the one hand, an output signal is supplied from an omnidirectional transducer 61, which has the same gain of its acoustoelectric receiver the characteristics are essentially independent of the angle of incidence θ, and the output signal from the directional converter 62 with the selected and formed radiation pattern receiving characteristic. According to Fig.25, the time delay τ can be performed directly using one of the converters.

 Thus, in the embodiment (FIG. 25), as in FIG. 8, τ can be selected equal to zero.

Using FIG. 23, a further preferred embodiment is explained, especially implementing the time delay detection unit 10 (FIG. 6). The output signals of the converters 1 and 2 are first converted using the corresponding analog-to-digital converters 16 and 17 and then, using the corresponding converters 18, 19 of the time domain, finally to the frequency domain. One signal as an example of S ₂ converted output signals of the converters, which after converting the time domain into the frequency domain can be represented as a spectrum of points S _2ω , is converted to its complex conjugate points in the conversion unit 50. At the output of this block 50, complex conjugate points S are generated $\genfrac{}{}{0ex}{}{\text{*}}{\text{2ω}}$ . This spectrum S * ₂ and the point spectrum S ₁ are multiplied to form the spectrum S _{3 of the} scalar product in the multiplication unit 52. It can be shown that the points S _{3ω of the} spectrum of S ₃ have a phase angle with respect to the real axis equal to Δφ _ω .
Thus, the imaginary part of the pointer S _3ω from S ₃ becomes equal.

Ιμ (Σ _3ω ) = | S _3ω | sin (Δφ _ω | (3)
Δφ _ω = ω • (p / c) • cos (θ) (4)
In accordance with Fig. 23, the conversion unit 53 forms the imaginary part of the pointers S _3ω , and in addition, the block 54 generates the amplitudes | S _3ω | these pointers.

For small values of Δφ _{ω, the} sine in (3) can be approximated directly using Δφ _ω , so that as a result, from (3) we obtain
Im (S _3ω ) = | S _3ω • S $\genfrac{}{}{0ex}{}{\text{*}}{\text{2ω}}$ | ω • (p / c) • cos (θ) (3)
Thus, and since it is performed using block 55, dividing the imaginary parts Im (S _3ω ) of the pointers S _{3ω of the} spectrum S ₃ using the corresponding values of the scalar product corresponding to | S _2ω | results in an output signal that corresponds to Δφ _ω . As already explained with reference to FIG. 7, Δφ _{ω is} further divided in block 56 by the corresponding pointer frequency ω. The resulting signal is equal to A ₁₀ (Fig.6) or A ₁₀ '(Fig. 7).

 All blocks 50, 52, 53, 54, 55 and 56 are preferably implemented in one block of the calculator.

 Let us return to the generalized block diagram of FIG. 6, describing various implementation possibilities of the delay detector unit 10.

 By means of FIG. 17-22, the effect of the amplitude filtering unit 12 and the controlled filtering unit 14 is further explained by a specific example.

17 shows examples of two characteristics of the weighing signals of block 12. According to characteristic I, each dt _{ω the} amplitude of the spectral line of signal A ₁₀ (see FIG. 6) is attenuated to zero if this amplitude falls below or above the predetermined dt _min values _{, ω} , dt _{mas, ω} and is set equal to "unity" if such an amplitude of the spectral component is within these two values.

Such a selection of the weighting function W results in a spectrum A _{12 of the} output signal, as shown in FIGS. 18a and 18b.

 Figa and 18b understandable to specialists without further explanation.

Fig. 19 shows an example of a spectrum of a signal S ₁ . In the controlled filtering unit 14, all spectral lines S ₁ (Fig. B) are amplified by a value 1 corresponding to A ₁₂ , or zeroed to zero value A ₁₂ , respectively. As a result, this leads to a spectrum S _r (FIG. 20) as the spectrum of the output signal of the controlled filtering unit 14 (FIG. 6). If the weighting function I (Fig. 17) is used in the method according to Fig. 7, then the beam shape, which is shown in Fig. 21, looks like highlighted lines. If an amplitude response is used, as shown II in FIG. 17, this results in a characteristic that is shown in FIG. 21 with a dashed line.

In FIG. 22 shows the resulting beam, if, by analogy with FIG. 17 and from the point of view of FIGS. 8 and 9, all values of ratios that exceed (A ₁ / A ₂ ) max are discarded. This is implemented using the amplitude response of the filter, which is also shown in FIG.

22, the ratio A ₁ / A _{2 is} designated r (ω).
Those skilled in the art will appreciate that only examples of the invention have been described using the drawings. For example, you can use more than two transducers or microphones located linearly, on a plane or in the form of a spatial matrix. In addition, unidirectional microphones can be used instead of omnidirectional. The beam forming in accordance with the invention can also be performed using combinations of functions from two or more microphones. It will also be clear to those skilled in the art that a delayed detector can be implemented in many other ways. In addition, the normalization, which was explained using the normalizer block 30 (Fig. 14), can be easily done by converting the time domain to the frequency domain only after analog-to-digital converters 16 and 17 and performing frequency-specific matrices or time delay tables τ _ω .about

Claims (30)

1. A method of electronically generating a predetermined gain characteristic between an electric receiving signal S _r and acoustic signals depending on the direction θ from which said acoustic signals IN come in at least two spatially separated acoustoelectric transducers 1, 2, containing at least within a predetermined frequency range steps: periodically determined from the output signals S ₁ , S _{2 of} these acoustoelectric transducers 1, 2 respectively the dithering signal dt _{ω of the} mutual spectral delay in accordance with the delays of the spectral reception in at least two converters is subjected to a signal depending on the output signal S ₁ of at least one 1 of the at least two converters 1, 2, spectral filtering using characteristic 14 spectral filtering, control the specified spectral filtering characteristic depending on the specified signal dt _ω , mutual spectral delay, use a signal that depends on the output signal filtering signal, as the electrical receiving signal S _r . 2. The method according to p. 1, characterized in that it further comprises the step of re-determining from these output signals S ₁ , S ₂ converted to the frequency domain. 3. The method according to p. 1 or 2, characterized in that it comprises the step of subjecting the signal, depending on the output signal S ₁ , to filtering with conversion to the frequency domain.  4. The method according to p. 3, characterized in that it further comprises the step of re-converting the signal that is used in the time domain.  5. The method according to one of paragraphs. 1-4, characterized in that it further comprises the step of periodically determining, by monitoring the phase difference of the spectral components of said output signals and dividing the phase difference of said spectral components as being controlled by the corresponding frequency of said spectral components.  6. The method according to one of paragraphs. 1-4, characterized in that the indicated determination is carried out using the steps of: providing one of at least two transducers with at least approximately omnidirectional acoustoelectric receiving characteristic, providing one of at least two transducers with a directional, profiled acoustoelectric characteristic, comparing signals depending on the output signals of at least two converters, and use the resulting comparison signal as a signal with mutual by holding.  7. The method according to one of paragraphs. 1-4, characterized in that it further carry out the determination using the following steps: superimpose signals depending on the specified output signals of at least two converters, whereby a resulting signal is generated and said resultant signal is compared with at least one of said signals depend on the specified output signals.  8. The method according to p. 7, characterized in that it further comprises the step of delaying one of the dependent signals before applying it for a predetermined time, which is independent of frequency or a function of frequency.  9. The method according to one of paragraphs. 1-4, characterized in that they further determine using the following steps: convert the signals depending on the output signals of at least two converters into the frequency domain, form the complex conjugate vectors of one of the converted signals, multiply the vectors of the other converted signals by complex conjugate vectors to obtain the vectors obtained as a result of multiplication, form the amplitudes of the vectors obtained as a result of multiplication, form the imaginary components of the vectors from the vector s obtained as a result of multiplication, form the ratio of the imaginary components of the vectors and the amplitudes multiplied by the corresponding frequency, and the resulting signal forming the ratio is the corresponding delayed control signal in the spectral representation.  10. The method according to p. 8, characterized in that it further comprises the step of delaying with a time delay, which differs from and preferably is greater than the ratio of the mutual distance p of at least two transducers 1, 2 and the speed of sound.  11. The method according to one of paragraphs. 7-8 or 10, characterized in that it further comprises the step of normalizing at least one of the signals obtained by comparing the signal obtained as a result of the overlay and at least one of the dependent signals using the function 30 normalization, depending on the frequency.  12. The method according to one of paragraphs. 7-8, 10 or 11, characterized in that it further comprises the step of comparing using the spectral formation of the ratio of the amplitudes of the signal obtained by superimposing and at least one of the dependent signals.  13. The method according to one of paragraphs. 1-12, characterized in that it further comprises the step of controlling the response due to the fact that the signal with spectral delay is subjected to spectral weighting and controlling the response by means of the result obtained by spectral weighting.  14. The method according to p. 13, characterized in that it further comprises the step of adjusting a predetermined gain characteristic by adjusting the weighting characteristic for the specified weighing. 15. An acoustic sensor device comprising at least two acoustoelectric transducers 1, 2 located at a predetermined mutual distance p, a time delay detection unit 10 with at least two inputs and an output, and its inputs are respectively connected in working condition to the outputs through at least two transducers 1, 2, while the time delay detection unit 10 generates an output signal A ₁₀ depending on the time delay of the acoustic signals IN, which fall at least two transducers 1, 2, a weighing unit 12 with a predetermined weighing characteristic and with an input and an output, and its input is connected in working condition to the output of the time delay detection unit 10, a filtering unit 14 with a controlled transition characteristic and at least one input, input with characteristic control and output, and its input in working condition is connected to the output of at least one of at least two converters, while its control input is connected in working Toyan to the output 12 of weighing unit block 14 filtering produces S _r output signal as a function of its input signal and the characteristics of which is controlled by a signal A _12, which is fed to the control input, which is dependent on the output signal of delay detection unit converted via weighing characteristics of the specified weighing unit.  16. The device according to p. 15, characterized in that the control output of the filtering unit 14 receives the output signal of the weighing unit in the form of a frequency mode region.  17. The device according to p. 15 or 16, characterized in that it further comprises time-to-frequency-domain converters connected to each other between at least two converters and a time delay detection unit 10, the detection unit being a spectral time delay detection unit.  18. The device according to one of paragraphs. 15-17, characterized in that one of the at least two transducers is a transducer with at least approximately omnidirectional acoustoelectric receiving characteristic in the time delay detection unit, comprises a comparison unit, its inputs being connected in operation to the outputs of at least two transducers while the output of the comparison unit is connected in working condition to the input of the time delay detection unit.  19. The device according to one of paragraphs. 15-17, characterized in that the time delay detection unit comprises an overlay unit, wherein its inputs are connected in working condition to the outputs of at least two converters, while its output is connected in working condition to the output of the time delay detection unit.  20. The device according to p. 19, characterized in that the time delay detection unit comprises a time delay unit 20 with an input that is connected in working condition to the output of one of the converters 1, and with an output that is in working condition connected to one input of the block 23 overlay, and the second input of the overlay unit is connected in working condition to the output of the second of at least two converters 2.  21. The device according to p. 20, characterized in that the output of the overlay unit is connected in working condition to one output of the comparison unit 25, while its second input is connected in working condition to the output of the second converter 2, and the output of the comparison unit is the output of the spectral detector unit delays.  22. The device according to p. 20 or 21, characterized in that the time delay unit 20 carries out a signal delay by an amount that differs from the set by the mutual distance p of the converters 1, 2, divided by the speed of sound c.  23. The device according to one of paragraphs. 15-22, characterized in that the time delay detection unit comprises a normalizing filter unit 30.  24. The device according to one of paragraphs. 19-23, characterized in that the comparison unit is a unit that forms the ratio of the amplitudes of the corresponding frequency components that are supplied to its inputs. 25. The device according to p. 15, wherein the time delay detection unit 10 is a spectral time delay detection unit and performs a measurement of the spectral phase difference Δφ _ω and dividing the spectral phase difference by the corresponding frequency ω.  26. The device according to p. 25, characterized in that the time delay detection unit comprises a calculation unit with two inputs that are connected in working condition to the outputs of at least two converters that form the complex conjugate signal vectors at one of its inputs, multiplied complex conjugate vectors by the corresponding signal vectors, which are fed to its second input, divide the imaginary part of the vectors obtained as a result of multiplication by the amplitude of the vectors obtained as a result of multiplied In addition, the vectors obtained as a result of the division are additionally divided by their respective frequency and the vectors obtained as a result of this additional division are output as the output signal at its output.  27. The device according to p. 15, characterized in that it further comprises a block 18 'converting the time domain into the frequency domain, which is connected between at least one of these outputs of the at least two acoustoelectric transducers and the filtering unit 14.  28. The device according to p. 20, characterized in that it further comprises a block 19 'converting the time domain into the frequency domain, which is connected between the output of the overlay unit 23 and the control input of the filtering unit 14.  29. The device according to one of paragraphs. 15-28, characterized in that it is a hearing aid, and the mutual distance of at least two transducers is preferably 0.5 to 1.5 cm and at most 4 cm.  30. The device according to one of paragraphs. 15-29, characterized in that the weighing unit contains a control input for adjusting the weighing characteristics of the specified weighing unit.

Images