Method for directional signal processing for a hearing instrument

Abstract

The invention describes a method for directional signal processing for a hearing instrument (1), wherein a first input signal (E1) and a second input signal (E2) of the hearing instrument (1) are generated from an ambient sound signal (4) by a first input converter (M1) and a second input converter (M2), respectively, wherein a first front intermediate signal (Z1v) and a first rear intermediate signal (Z1h) are each generated from the first input signal (E1) and the second input signal (E2), and wherein, in particular frequency band by frequency band, a first superposition (U1) of the first front intermediate signal (Z1v) and the first rear intermediate signal (Z1h) is formed by means of a complex-valued first superposition parameter (a1), and is adapted based on the first superposition parameter (a1). It is provided that a complex value (a1) resulting from said adaptation of the first superposition (U1) is used.0) of the first superposition parameter (a1) is converted into a first alternative parameter (ap1) and a second alternative parameter (ap2), wherein at least the second alternative parameter (ap2) has at least a semicircular monotonic relationship to an angle (θ) of minimum sensitivity of the first superposition (U1), wherein the angle (θ) of minimum sensitivity is limited to a predetermined angular range (Δθ) by means of a corresponding limitation of the second alternative parameter (ap2), and thereby a limited second alternative parameter (ap2') is formed, and wherein an output signal (out) is generated on the basis of the first alternative parameter (ap1) and the limited second alternative parameter (ap2') as well as on the basis of a superposition of the first input signal (E1) and the second input signal (E2).

Description

[0001] The invention relates to a method for directional signal processing for a hearing instrument, wherein a first and second input signal are generated from an ambient sound signal by a first and second input converter of the hearing instrument, respectively, wherein a first front intermediate signal and a first rear intermediate signal are formed on the basis of the first and second input signals, respectively, wherein, in particular frequency band-wise, a first superposition of the first front intermediate signal and the first rear intermediate signal is formed by means of a first superposition parameter, and is adapted on the basis of the first superposition parameter, and wherein an output signal is generated on the basis of a value of the first superposition parameter and on the basis of a superposition of the first input signal and the second input signal, in particular with a time delay.

[0002] In hearing instruments, such as hearing aids for the treatment of a wearer's hearing loss, a number of input signals are generated from ambient sound by a number of input transducers, such as microphones. These input signals represent the air pressure fluctuations of the ambient sound at the respective input transducer. Based on these input signals, an output signal is generated through signal processing. This output signal is then converted into an output sound signal by an output transducer of the hearing instrument (e.g., a loudspeaker). The signal processing can be specifically tailored to the wearer's audiological requirements (e.g., hearing impairment) and may include frequency-band-specific amplification and / or compression.

[0003] In the case of two (or more) input transducers in a hearing aid, directional processing of the generated input signals is also possible. This allows, for example as an intermediate signal during the generation of the output signal, a directional signal to be directed towards a presumed source of the desired signal (usually a conversation partner or similar), and / or which suppresses sources of interference through spatial "attenuation".

[0004] Such attenuation can be achieved by a time-delayed superposition of the two input signals, or by two different such superpositions, for example, a so-called cardioid and an anticardioid signal, which are themselves adaptively superimposed. A potential problem here, however, is that the most complete possible attenuation of an interference source requires a signal level that is as identical as possible in the two (or more) input transducers of the hearing aid. Due to shadowing effects caused by both the wearer's head (or parts of the outer ear) and the hearing aid housing, this is often not the case. Therefore, an input signal for complete attenuation of a directional interference source must be adjusted accordingly by an angle-dependent gain factor. However, such a gain factor is often difficult to determine.Moreover, such an amplification factor can lead to strong fluctuations in the desired signal, which is undesirable. A further challenge is the additional requirement that the aforementioned suppression of the interference source often needs to be limited to a specific angular range relative to the user's field of vision (e.g., the rear hemisphere).

[0005] For example, DE102020210805 B3 discloses a method for directional signal processing for an acoustic system, by means of which a highly directional interference signal source is not completely suppressed, but in particular remains audible in an output signal of the acoustic system.

[0006] The invention is therefore based on the objective of providing a method for directional signal processing for a hearing instrument which is as robust as possible against different signal levels of the individual input signals involved, and which allows an efficient limitation of an angle of the minimum sensitivity of a resulting directional signal.

[0007] The aforementioned problem is solved according to the invention by a method according to claim 1 for directional signal processing for a hearing instrument, wherein a first input signal is generated from an ambient sound signal by a first input transducer of the hearing instrument, wherein a second input signal is generated from the ambient sound signal by a second input transducer of the hearing instrument, and wherein a first front intermediate signal and a first rear intermediate signal are formed based on the first input signal and the second input signal respectively, and preferably by a time-delayed superposition.

[0008] It is provided that, particularly frequency band-wise, a first superposition of the first front intermediate signal and the first rear intermediate signal is formed using a complex-valued first superposition parameter, and is adapted based on the first superposition parameter, wherein a complex value of the first superposition parameter resulting from said adaptation of the first superposition is converted into a corresponding pair of real-valued alternative parameters, consisting of a first alternative parameter and a second alternative parameter, wherein at least the second alternative parameter has at least a semicircular monotonic relationship to an angle of minimum sensitivity of the first superposition, wherein the angle of minimum sensitivity is modified by a corresponding modification of the second alternative parameter, and thereby a modified second alternative parameter is formed.and wherein an output signal is generated based on the first alternative parameter and the modified second alternative parameter, as well as on a superposition of the first input signal and the second input signal. Advantageous and partly inventive embodiments are the subject of the dependent claims and the following description.

[0009] In this context, a hearing instrument generally includes any device designed to generate at least two corresponding input signals using at least two input converters, and to process these signals to generate an output signal. This output signal is then converted into an output sound signal by an output converter and delivered to the ear of a wearer of the device. Specifically, a hearing instrument can include headphones equipped with the appropriate input converters (e.g., earplugs), a headset, smart glasses with a speaker, etc.However, a hearing instrument also includes a hearing aid in the narrower sense, i.e., a device for treating a hearing impairment of the wearer, in which the input signals generated from ambient sound by means of the input transducers are processed into the said output signal depending on the audiological requirements of the wearer and are amplified and / or compressed in particular depending on the frequency band, so that the output sound signal is suitable to at least partially compensate for the hearing impairment of the wearer, especially in a user-specific manner.

[0010] An input transducer (especially an electroacoustic one) is defined here as any device designed and configured to generate a corresponding electrical signal (the associated input signal) from the ambient sound signal, the voltage or current fluctuations of which preferably represent the fluctuations in the air pressure of the sound signal and reproduce them within the respective resolution. In particular, a microphone is included here as an input transducer.

[0011] In particular, the angle of minimum sensitivity is limited as a modification to a predetermined angular range by means of a corresponding limitation of the second alternative parameter, and a limited second alternative parameter is formed as the modified second alternative parameter, whereby an output signal is generated on the basis of the first alternative parameter and the limited second alternative parameter as well as on the basis of a superposition of the first input signal and the second input signal.

[0012] By modifying, in particular limiting, the first alternative parameter, a minimum sensitivity (i.e., in particular a depth of a so-called "notch") at the corresponding angle can advantageously be modified.

[0013] The formation or generation of a resulting signal from one or more incoming signals means in particular that the respective signal components of the incoming signals, especially in frequency bands, are incorporated into the resulting signal according to a mapping rule, such that preferably a monotonic, particularly preferably linear relationship exists between the amplitudes and / or envelopes and / or signal levels of the incoming signals and the respective magnitude of the resulting signal.

[0014] The first front intermediate signal and the first rear intermediate signal are preferably generated from the first and second input signals using mutually symmetrical mapping rules, particularly as time-delayed superpositions, such that the directional characteristics of the two first intermediate signals are symmetrical to each other with respect to free space. However, the first front intermediate signal and the first rear intermediate signal can also be generated using mutually different (in particular, asymmetrical) mapping rules, with the two intermediate signals preferably being linearly independent of each other. In particular, it is conceivable that one of the intermediate signals has an omnidirectional directional characteristic.

[0015] The first superposition U1 is formed in particular in the form U 1 ω , t = Z 1 v ω , t + a 1 ¯ ω , t ⋅ Z 1 h ω , t , where Z1v and Z1h respectively denote the first front and rear intermediate signals, a 1 ¯ ∈ ℂ the first superposition parameter, as well as ω and t being a frequency and a discrete time index respectively.

[0016] The adaptation of the first superposition based on the first superposition parameter includes, in particular, that the first superposition (the actual superposition, e.g. according to equation (i), is used synonymously with the signal resulting from said superposition) is optimized with respect to a characteristic such as the total energy, the total level or a deviation from a reference signal, etc., via the first superposition parameter, whereby the optimization can also be carried out numerically in several steps, so that the first superposition parameter (even for a given time index) converges via the adaptation to a value (which can be determined, for example, on the basis of a limit value for a step size between two adaptation steps).

[0017] The value of the first superposition parameter, which generally has a real and imaginary part, is now converted into a pair of real-valued alternative parameters, i.e., a first alternative parameter and a second alternative parameter, the latter having a monotonic relationship to an angle of minimum sensitivity of the first superposition.

[0018] This can be motivated in particular by the following consideration: For a preferably stationary sound signal, which strikes the hearing instrument with respect to its frontal direction (defined in particular by the direction from the second to the first input transducer) at an angle of θ, the relative transfer function from the first to the second input transducer (i.e. the amplitude and phase difference due to the propagation of the sound signal from the sound source at the angle θ to the second instead of to the first input transducer) is A θ · e -iωτ cos θ< , where A θ is an angle-dependent amplitude factor (which takes into account, among other things, shadowing effects by the wearer's head or by the housing of the hearing instrument).With a suitable choice of the two alternative parameters, in particular by a relation between the value of the first superposition parameter and the said relative transfer function, at least a semicircular monotonic relationship can be formed between the second alternative parameter and the angle of minimum sensitivity of the first superposition.

[0019] A relationship that is at least semicircularly monotonic includes, in particular, that the relationship between the second alternative parameter and the angle of minimum sensitivity holds at least for an angular range of said angle which covers at least one semicircle, i.e., that a ψ ∈ ℝ exists such that the said monotone relationship holds at least for an angular range of [ψ, ψ + π].

[0020] Using the second alternative parameter, this angle can now be limited to a desired, predetermined angular range, e.g., to the rear half-space (θ ∈ [90°, 270°]), or a narrower "wedge" in the rear half-space (e.g., θ ∈ [120°, 240°]), via the aforementioned monotonic relationship (and possibly via a sign of the angle and / or a transformation of the angle of minimum sensitivity by π), by restricting the range of values ​​of the alternative second parameter to a corresponding interval (and possibly taking into account a sign of the said angle θ with respect to the frontal or 180° direction).

[0021] By limiting the angle of minimum sensitivity to the specified angular range, the alternative second parameter is adjusted to the interval of its value range corresponding to this angular limitation, thereby generating the limited second alternative parameter. This limited second alternative parameter can preferably be identical to the second alternative parameter if the corresponding angle θ of minimum sensitivity already lies within the specified angular range, or otherwise be defined by a limit value of such an interval.

[0022] An output signal is generated using this limited second alternative parameter and a superposition of the first and second input signals. This can be achieved, in particular, by reversing the calculation of the two alternative parameters from the first superposition parameter, such that an adapted first superposition parameter is formed based on the first alternative parameter and the limited second alternative parameter. Accordingly, the superposition of the two input signals to generate the output signal is given by the first superposition (which, due to its generation from the first front and first rear intermediate signals, also represents a superposition of the two input signals), but now with the first adapted superposition parameter.

[0023] The output signal can be directly converted into an output sound signal by an output converter of the hearing aid (such as a loudspeaker), which is then delivered to the hearing aid user's ear. Alternatively, the output signal can undergo further signal processing steps (such as additional noise reduction and / or frequency-band-wise amplification or compression) before the output sound signal is generated. In particular, another signal can be mixed into the output signal before it is converted into the output sound signal.

[0024] Such a pair of alternative parameters can be formed in particular by representing the first superposition U1 according to equation (i) by a corresponding conversion in the basis of the two input signals as U 1 = E 1 ⋅ w 1 + E 2 ⋅ w 2 = E T ⋅ w with the vector of input signals ET< = (E1, E2) and the coefficient vector w = (w1, w2) T< , where the coefficients w1 and w2 depend on the specific form of the generation of the first front and rear intermediate signal Z1v, Z1h in equation (i).

[0025] To gain a better understanding of the coefficient vector w, the underlying adaptation of the first superposition U1 is assigned to a first-order finite impulse response (FIR) filter, which then performs a kind of "spatial sampling" of the sound signal. The corresponding filter polynomial is P z = w 1 + w 2 ⋅ z − 1 .

[0026] The zeros of the polynomial in equation (iii) are z0 = - w2 / w1, and are, up to a complex prefactor, c _ ∈ ℂ unambiguously determined. Accordingly, signals that differ from the first superposition U1 by such a scalar prefactor exhibit c _ ∈ ℂ They differ in that they exhibit the same properties as these with regard to their directional effect.

[0027] Given the aforementioned ambiguity of the zero of the FIR filter according to equation (iii), which is associated with the adaptation of the first superposition, the coefficient vector is now set to w 0 = c · [1, - r · e iφ< ], where the relative phase φ and the quotient r of the magnitudes of the two coefficients w2 / w1, as already mentioned, depend on the specific configuration of the first front and rear intermediate signals Z1v, Z1h. The first alternative parameter can now be formed based on the aforementioned quotient r (which thus indicates the ratio of the magnitudes of the coefficients), and in particular as this, the second alternative parameter based on the relative phase φ of the coefficients to each other, and in particular as this.

[0028] This can be seen by applying a stationary sound signal from an angle θ, as mentioned above, which should be attenuated as completely as possible. Using the relative transfer function between the two input transducers mentioned above, the vector is obtained. E of the two input signals as E = E1 · h with h = [1, A θ · e -iωτcos θ< ] T< . In the present representation, cancellation of the sound signal then requires h T ⋅ w 0 = 0 , bzw . 1 , A θ ⋅ e − iωτ cos θ ⋅ 1 , − r ⋅ e iφ T = 0 .

[0029] One solution derived from equation (vi') is r = 1 / A θ , φ = ωt · cos θ. This shows that for θ ∈ [0, π] there is a monotonic relationship between the relative phase φ of the two coefficients w1, w2 of the first superposition (in the representation of the input signals) and the angle of minimum sensitivity. This angle can then be limited using the relative phase φ as a second alternative parameter.

[0030] It proves to be further advantageous if the value of the first superposition parameter is converted into a corresponding real-valued second superposition parameter and an associated value of a real-valued gain factor, wherein the real-valued gain factor corresponds to a gain of the second input signal when forming the first front oris assigned to the rear intermediate signal, and the second superposition parameter is adjusted such that for a second superposition, which is formed from the first front intermediate signal and the first and rear intermediate signals using the second superposition parameter and amplifying the second input signal with the said amplification factor, the angle of minimum sensitivity is limited to the specified angular range, and thereby an adapted second superposition parameter is generated, and the output signal is generated using the adapted second superposition parameter and the amplification factor, as well as using a particularly time-delayed superposition of the first input signal and the second input signal.

[0031] The aforementioned real amplification factor m ∈ ℝ This corresponds to an amplification of the second input signal during the generation of the first front or first rear intermediate signal. In other words, the first superposition parameter is used to determine the... a 1 ¯ ∈ ℂ (for the first superposition of the first front and the first rear intermediate signal) i.e. the gain factor m and the second superposition parameter a 2 ∈ ℝ determined in such a way that the first superposition transitions into a second superposition of the first front and the first rear intermediate signal, wherein in said first intermediate signals the second input signal was previously amplified, i.e. scaled, by the gain factor m, and wherein the second superposition of these intermediate signals is formed based on the second superposition parameter a 2 ∈ ℝ In general, this conversion between the real and imaginary parts of a 1 ¯ ∈ ℂ after a 2 , m ∈ ℝ 2 well-defined.

[0032] It is not absolutely necessary that the second superposition (i.e., the resulting signal) is actually generated (similar to equation (i)); rather, it is sufficient to simply perform the conversion a1 → (a2, m) according to the restrictions arising in particular from the intermediate signals (e.g., for the gain factor).

[0033] Ideally, the gain factor is determined such that a superposition of the first intermediate signals (after appropriate prior application of the gain factor to the second input signal when generating the intermediate signals) allows for the complete elimination of an interference source, thus fulfilling the function of level matching between the two input converters of the hearing instrument. In this case, a general monotonic relationship can be established between the second superposition parameter a2 and the angle for which the aforementioned second superposition U 2 ω , t = Z 2 v ω , t + a 2 ω , t ⋅ Z 2 h ω , t , exhibits a minimum sensitivity or maximum attenuation. In equation (i'), Z2v and Z2h denote a second front and second rear intermediate signal, respectively, which are derived from the first front and first rear intermediate signal by a prior amplification of the second input signal with the aforementioned gain factor.

[0034] As already described, a monotonic relationship can now be established between the angle of maximum attenuation of the second superposition U2 and the second superposition parameter a2 (however, the monotonicity is only defined over angular ranges of half a circular rotation, i.e., for Winkel θ ∈ γ , γ + π mit γ ∈ ℝ .

[0035] The second superposition parameter can form the second alternative parameter, and the gain factor the first alternative parameter. However, the first alternative parameter can also be formed, as described by equation (iv'), from the quotient r of the magnitudes of the two coefficients w1, w2 of the first superposition with respect to the two input signals, and the second alternative parameter from the relative phase φ of the two coefficients to each other, whereby the second superposition parameter is fitted according to the aforementioned alternative parameters r and φ, and thus the fitted second superposition parameter is formed.

[0036] An output signal is generated using this adjusted second superposition parameter and a superposition of the first and second input signals. This superposition of the first and second input signals can be achieved, in particular, through the second superposition (of the second front and second rear intermediate signals) according to equation (i'), where the adjusted second superposition parameter a2' (instead of the "original" second superposition parameter a2) is used. When generating the output signal, a frequency response correction filter can be added to ensure a flat frequency response, for example, in the frontal direction (defined, for instance, by the direction from the second to the first input transducer of the hearing instrument).In the case where the time delay in the relevant superpositions is implemented using a frequency factor, the correction filter can also be given by a frequency-dependent correction factor.

[0037] In particular, a second front intermediate signal and a second rear intermediate signal are each formed from the first input signal and the second input signal scaled by the real-valued gain factor, preferably by a time-delayed superposition, wherein the output signal is generated from the second superposition using the adapted second superposition parameter.

[0038] The aforementioned superposition of the two input signals to generate the output signal can also be achieved through the first superposition according to equation (i), where, however, the gain factor m and the adjusted second superposition factor a2' are again set back to the then "adjusted" value for the first superposition parameter. a1 ' is mapped, in particular by means of the reverse mapping rule (a2', m) → a1' .

[0039] Preferably, the first rear intermediate signal is generated such that it exhibits a relative attenuation in a frontal direction, which is defined in particular by a direction from the second input converter to the first input converter, and the first front intermediate signal is generated such that it exhibits a relative attenuation in a direction opposite to the frontal direction. In particular, the first front and the first rear intermediate signals are symmetrical to each other. The same applies in particular to the second front and second rear intermediate signals. A relative attenuation is understood to mean, in particular, a local and preferably a global minimum of sensitivity across all angles. This minimum need not necessarily represent a maximum attenuation in the sense of total attenuation, but can, in particular for the first intermediate signals, also assume finite values ​​for the respective sensitivity.

[0040] Advantageously, the first front intermediate signal and the first rear intermediate signal are each generated by a time-delayed superposition of the two input signals, whereby for the first front intermediate signal the second input signal and for the first rear intermediate signal the first input signal, the first input signal is delayed, preferably by the acoustic propagation delay between the two input converters. This results in directional signals as the first intermediate signals, which exhibit a cardioid or anticardioid directional characteristic in free space and are particularly suitable for the present method due to their simple yet stable generation.

[0041] Advantageously, at least in one frequency band, preferably up to a band cutoff frequency of 500 Hz, a delay between the input signals, particularly in the time-frequency domain, is implemented by means of an additional all-pass filter. In the time-frequency domain, a delay can be implemented via a phase factor that depends on the center frequency of the relevant frequency band. Depending on the implementation, however, this center frequency for the first frequency band may be 0 Hz, meaning no delay would be possible. In this case, an alternative implementation of the delay via an all-pass filter is advantageous. This can also be beneficial for other, lower frequency bands if the phase within a frequency band exhibits large variations that are inadequately represented by a phase factor that remains constant across the frequency band.

[0042] It is further advantageous if, in a first adaptation step, an initial value of the complex first superposition parameter is determined, this initial value is converted into the corresponding first and second alternative parameters, and the limited second alternative parameter is determined from these. Based on the first alternative parameter and the limited second alternative parameter, a second value of the first superposition parameter is then determined, and this second value is used for a second adaptation step. In other words, it is not necessary for the limitation of the angular range for the angle of minimum sensitivity of the second superposition to occur only after the complete adaptation of the first superposition parameter.Rather, such a restriction can also occur in a single adaptation step, and the limited second alternative parameter can form the basis for the next adaptation step.

[0043] Advantageously, the first superposition parameter is determined using a least-mean-squares algorithm and / or a gradient descent method. These methods are particularly suitable for adapting the complex-valued first superposition parameter with its real and imaginary parts, and thus, in particular, for optimizing the corresponding first superposition with respect to a characteristic value based on the first superposition parameter. The gradient descent method can, in particular, involve applying a gradient of the real and imaginary parts with respect to such a characteristic value (such as a signal level or a deviation from an error or reference signal).

[0044] The invention further describes a hearing instrument according to claim 17, comprising a first input converter for generating a first input signal from an ambient sound signal, a second input converter for generating a second input signal from the ambient sound signal, and a control unit, wherein the hearing instrument is configured to perform the aforementioned method. The hearing instrument is, in particular, configured by means of the control unit to perform the method steps in which one of the input signals or signals derived therefrom is processed. For this purpose, the control unit is, in particular, equipped with at least one signal processor.

[0045] The hearing instrument according to the invention shares the advantages of the method according to the invention. The advantages stated for the method and for its further development can be transferred analogously to the hearing instrument.

[0046] An embodiment of the invention is explained in more detail below with reference to the drawings. The drawings schematically depict: Fig. 1 shows a top view of the directional characteristics of intermediate signals from a hearing instrument; Fig. 2 shows a top view of the directional characteristics of the intermediate signals. Fig. 1 In the case of unequal signal levels of the input converters, Fig. 3 shows in a block diagram the sequence of a method for directional signal processing in a hearing instrument, and Fig. 4 shows in a block diagram a method according to Fig. 3 alternative design.

[0047] Corresponding parts and sizes are marked with the same reference symbols in all figures.

[0048] In Figure 1The schematic top view shows the directional characteristics of a hearing instrument 1. The hearing instrument 1 is designed as a hearing aid 2, intended and configured to treat hearing loss. The hearing instrument 1 has a first input transducer M1 and a second input transducer M2, which are arranged at a distance d from each other and are each represented by a corresponding microphone. From an ambient sound signal 4, the first input transducer M1 generates a first input signal E1, and the second input transducer M2 generates a second input signal E2. The hearing instrument 1 also has a control unit 5, which is configured to process the aforementioned input signals E1 and E2, and for this purpose includes, in particular, a signal processor (not shown in detail).

[0049] A first front intermediate signal Z1v is generated by a time-delayed superposition of the first input signal E1 and the second input signal E2, where the time delay corresponds exactly to the acoustic travel time of the distance d: Z 1 v ω , t = E 1 ω , t − E 2 ω , t − τ , bzw . Z 1 v ω , t = E 1 ω , t − e − iωτ E 2 ω , t

[0050] Ideally, if the signal levels of the first and second input signals E1 and E2 are identical (and, in particular, if there are no shadowing effects and no attenuation over the distance d), the first front intermediate signal exhibits a cardioid directional characteristic (dashed line). In a manner comparable to equations (v) and (v'), but with a delay of the first input signal E1, a first rear intermediate signal Z1h = e - iωτ < E1 - E2 is generated. In the aforementioned ideal case, the first rear intermediate signal Z1v exhibits an anticardioid directional characteristic (dotted line), which has its maximum attenuation in a frontal direction 6. The direction of maximum attenuation of the first front intermediate signal Z1v is opposite to the frontal direction 6.

[0051] From the first front and the first rear intermediate signals, a complex-valued first superposition parameter is now used. a 1 ¯ ∈ ℂ a first superposition U1 is formed according to equation (i), where the value of the first superposition parameter a1 (i.e., its real and imaginary parts) can be determined by adapting the first superposition U1, for example, by minimizing the signal energy or level using a gradient descent method. A disturbance source 8, which contributes a directional disturbance 10 to the ambient sound signal 4, can now be "shut out" by means of the first superposition U1, as shown by the directional characteristic of the first superposition U1 (solid line). This directional characteristic exhibits maximum attenuation at the angle θ, at which the disturbance source 8 is located.

[0052] However, if the signal levels for the two input signals E1 and E2 are not equal, for example due to shadowing effects (e.g., by the head and / or the pinna of the wearer of the hearing instrument 1, but also by the housing of the hearing instrument 1), then, depending on the type of these shadowing effects, the attenuation of the first rear intermediate signal Z1h in the frontal direction 6 may no longer be complete, but rather have a finite value. A similar situation can occur for the first front intermediate signal Z1v, depending on the specific level differences between the input signals E1 and E2. As a result, complete attenuation and thus complete suppression of the noise 10 may no longer be achieved based on the first superposition U1 in the direction of the noise source 8.

[0053] This situation is schematically represented in a top view in Figure 2The first front intermediate signal Z1v (dashed line) and the first rear intermediate signal Z1h each exhibit a directional characteristic which, in some directions, no longer allows for complete attenuation. For this reason, the first superposition U1 (not shown), formed according to equation (i) based on the first front and rear intermediate signals Z1v and Z1h, is also Figure 2 , in the present case not suitable to completely eliminate the source of interference 8.

[0054] To solve this problem, a method is proposed which is illustrated using a block diagram in Figure 3 is depicted. Figure 3 The sound signal 4 is transmitted to the environment. Figure 1The interference noise 10 from the directional interference source 8 (not shown) is converted by the first and second input converters M1 and M2 into the first and second input signals E1 and E2, respectively. The first front and first rear intermediate signals Z1v and Z1h, respectively, are then generated from the two input signals E1 and E2 by time-delayed superposition (see description of...). Figure 1 , in particular equation (ii')): Z 1 v = E 1 − e − iωτ E 2 , Z 1 h = e − iωτ E 1 − E 2 .

[0055] In general, the signal levels of the first and second input signals E1, E2 are not equal, so the first front and first rear intermediate signals Z1v, Z1h have directional characteristics comparable to those in Figure 2 shown.

[0056] Based on a complex first superposition parameter a 1 ¯ ∈ ℂ A first superposition U1 is now formed from the first front and the first rear intermediate signals Z1v, Z1h according to equation (i). This first superposition U1 is subjected to an adaptation 12 in which a specific value a1.0 for the first superposition parameter a1 is determined. The adaptation 12 can, for example, be carried out by minimizing the signal energy of the first superposition U1 using a gradient descent method with respect to the real and imaginary parts of the first superposition parameter a1, or similarly.

[0057] According to equations (i), (v") and (vi), the first superposition U1 is: U 1 = E 1 ⋅ 1 + a 1 ¯ e − iωτ − E 2 ⋅ e − iωτ + a 1 ¯ = E 1 ⋅ w 1 + E 2 ⋅ w 2 , U 1 = E T ⋅ w with the vector of input signals E T< = (E1, E2) and the coefficient vector w = (w1, w2) T< , where the concrete form of the coefficients w1 and w2 is now given by equation (vii).

[0058] For the further procedure, the coefficient vector w according to equation (vii) is therefore transformed into the form w0 brought to equation (iv'), where r e i ϕ = e − i ωτ + a 1 ¯ 1 + a 1 ¯ e − i ωτ The relative phase φ is simply obtained from the argument of the right-hand side of equation (viii), and the factor r is given by the quotient r of the magnitudes of the coefficients w2 / w1 according to equation (vii). The latter is now used as a first alternative parameter ap1, and the relative phase φ as a second alternative parameter ap2. This can now be used, according to the relationship e iφ - iωτ cosθ< resulting from equation (iv'), to limit an angular range Δθ for the angle θ, resulting in a limited relative phase φ' or a limited alternative second parameter ap2'. In particular, this limited relative phase φ' can be identical to the relative phase φ if the angle θ of the minimum sensitivity of the first superposition U1 already lies within the desired angular range Δθ (e.g., the rear hemisphere with respect to the frontal direction 6).

[0059] Due to the adjusted relative phase φ', the first superposition parameter a1 in equation (viii) is also adjusted accordingly. Equation (viii) can then be solved for this adjusted first superposition parameter. a1' to be dissolved a 1 ′ ¯ = e − iωτ − r e iϕ ′ r e iϕ ′ − iωτ − 1 .

[0060] Based on the adjusted first superposition U1' (dashed line) with said adjusted first superposition parameter a1 An output signal Out can now be generated, whereby the adapted first superposition U1' is multiplied, in particular, by a correction factor c cor to correct the frequency response, so that the frequency response of the output signal Out is flat in the frontal direction 6. In addition, further signal processing steps 20, such as noise or feedback suppression, etc., but also frequency band-dependent boosting depending on the audiological requirements of the wearer, etc., can be interposed.

[0061] In Figure 4An alternative design of the procedure is shown using a block diagram. Figure 3 As shown in the previous example, the first superposition U1 is formed based on the first superposition parameter a1, and in adaptation 12 the value a1 .0 of the first superposition parameter a1 determined. In the next step, the value will be calculated. a1 .0 of the first superposition parameter a1 on a real-valued second superposition parameter a 2 ∈ ℝ and a real-valued gain factor m ∈ ℝ depicted, with the latter being assigned to the second input signal E2.

[0062] To determine the relationship between the first superposition parameter a1 (or from its value) a1.0) and to be able to determine the specific values ​​of the second superposition parameter a2 and the gain factor m, a second front intermediate signal Z2v and a second rear intermediate signal Z2h are defined (dashed signal path), in which, however, the second input signal E2 is each subjected to the gain factor m, i.e. Z 2 v = E 1 − m ⋅ e − iωτ E 2 , Z 2 h = e − iωτ E 1 − m ⋅ E 2 .

[0063] This gain factor m allows for compensation of different signal levels between the first and second input signals E1, E2. Therefore, even in the case of different signal levels, the second front and second rear intermediate signals Z2v, Z2h exhibit the characteristics shown in Figure 1The directional characteristics shown no longer apply to the first front and first rear intermediate signals Z1v, Z1h in the general case (i.e., not in free space, but with shadowing effects, etc.). (For this general case, these directional signals exhibit directional characteristics as described.) Figure 2 on).

[0064] If a second superposition U2 (dashed signal path) is formed from the aforementioned second front and second rear intermediate signals Z2v, Z2h (which differ from the corresponding first front and first rear intermediate signals Z1v, Z1h respectively by the aforementioned gain factor m in the component of the second input signal E2) using the second superposition parameter a2, analogous to equation (i), then the following applies to this: U 2 = E 1 ⋅ 1 + a 2 e − iωτ − m ⋅ E 2 ⋅ e − iωτ − a 2 = E 1 ⋅ w 1 ′ + E 2 ⋅ w 2 ′ , und mithin U 2 = E ⊤ ⋅ w ′ .

[0065] The amplification factor m and the second superposition parameter a2 are to be determined in such a way that the representation limits the angle θ of the maximum attenuation (see Figure 1 ) should be possible within a desired angular range.

[0066] In a manner analogous to equations (vii') and (viii), the relationship between the relative phase φ and the coefficient quotient r = |w2' / w1'| on the one hand, and the gain factor m as the first alternative parameter ap1 and the second superposition parameter a2 as the second alternative parameter ap2 on the other hand, is now established from equation (x) with the consideration motivated from equations (iii) and (iv'): r e i ϕ = m e − i ωτ + a 2 1 + a 2 e − i ωτ

[0067] This exploited the fact that the zeros of the polynomial in equation (iii) only vary up to a factor of c _ ∈ ℂ are defined, from which it follows that w2 / w1 = w2' / w1'. Since the fraction on the right-hand side has a magnitude of 1, the amplification factor m = r, where r is determined by the value a1 .0 of the first superposition parameter a1 is given by the magnitude of equation (viii).

[0068] For the second superposition parameter a2 as the second alternative parameter ap2 of the method according to Figure 4 This results from equation (xi): a 2 = e − iωτ − e iϕ e iϕ − iωτ − 1 = cos ϕ − cos ωτ 1 − cos ϕ − ωτ

[0069] Based on corresponding tabulated values, the relationship between φ (the relative phase of the coefficients w1' and w2' in equation (x)) and the angle θ of the minimum sensitivity of the first superposition U1 (e iφ - iωτ cosθ<= 1, see equation (iv')) and thus also of the second superposition U2 (which initially only represents a conversion of the first superposition U1) can now be established.

[0070] From this, a corresponding adapted value for the second superposition parameter a2, i.e., an adapted second superposition parameter a2' or a limited second alternative parameter ap2', can be determined.

[0071] The output signal out can now be generated (possibly after further signal processing steps 20 and correction factors for the frequency response not shown) from the second superposition U2 according to equation (i') with the second front and second rear intermediate signals Z2v, Z2h according to equation (ix), but based on the adjusted second superposition parameter a2' (instead of, as in equation (i'), based on the second superposition parameter a2). The gain factor m in the second front and second rear intermediate signals Z2v, Z2h according to equation (ix) is derived as r = m according to equation (xi) with r according to equation (viii) from the first superposition parameter a1 .

[0072] However, the amplification factor m and the adjusted second superposition parameter a2' can also be calculated back into the domain of the first superposition parameter. a1 (not shown), so that the output signal out is then formed in that case from a first superposition based on the adjusted first superposition parameter determined in this way. a1 '. This approach has the advantage that a prefactor in the output signal out, which corrects a high-pass behavior in the frequency response of the first superposition U1, is independent of the angle θ of the minimum sensitivity.

[0073] Although the invention has been illustrated and described in detail by the preferred embodiment, the invention is not limited by the disclosed examples and other variations can be derived by the person skilled in the art without leaving the scope of protection of the invention. Reference symbol list

[0074] 1 Hearing instrument 2 Hearing aid 4 Sound signal (of the environment) 5 Control unit 6 Frontal direction 8 Source of interference 10 Background noise 12 Adaptation 20 Signal processing steps a1 ( ' ) (Adapted) first superposition parameter a2(') (Adapted) second superposition parameter ap1, ap2 first or second alternative parameter ap2' limited second alternative parameter C cor correction factor E1, E2 first or second input signal M1, M2 first or second input converter out output signal m gain factor r quotient (of the magnitudes of the coefficients) U1, U2 first or second superposition w1('), w2(') coefficients Z1v, Z1 first front or first rear intermediate signal Z2v, Z2h second front or second rear intermediate signal ΔθAngle range θAngle (minimal sensitivity) τTime delay φRelative phase (of the coefficients)

Claims

1. Method for directional signal processing for a hearing instrument (1), - wherein a first input signal (E1) is generated by a first input transducer (M1) of the hearing instrument (1) from a sound signal (4) of the surroundings, - wherein a second input signal (E2) is generated by a second input transducer (M2) of the hearing instrument (1) from the sound signal (4) of the surroundings, - wherein a first front intermediate signal (21v) and a first rear intermediate signal (Z1h) are each formed on the basis of the first input signal (E1) and the second input signal (E2), - wherein, in particular by frequency band, a first superposition (U1) of the first front intermediate signal (21v) and the first rear intermediate signal (Z1h) is formed by means of a complex-value first superposition parameter (a1), and is adapted on the basis of the first superposition parameter (a1), - wherein a complex value (a1.0) of the first superposition parameter (a1) resulting from the adaptation of the first superposition (U1) is converted into a corresponding pair of real-value alternative parameters, consisting of a first alternative parameter (ap1) and a second alternative parameter (ap2), wherein at least the second alternative parameter (ap2) has an at least semicircular monotonous relationship to an angle (θ) of minimal sensitivity of the first superposition (U1), - wherein the angle (θ) of minimal sensitivity is modified via a corresponding modification of the second alternative parameter (ap2), and a modified second alternative parameter is formed here, and - wherein an output signal (out) is generated on the basis of the first alternative parameter (ap1) and the modified second alternative parameter and on the basis of a superposition (U1, U2) of the first input signal (E1) and the second input signal (E2).

2. Method according to Claim 1, wherein the angle (θ) of minimal sensitivity is limited as a modification to a specified angle range (Δθ) via a corresponding delimitation of the second alternative parameter (ap2), and a limited second alternative parameter (ap2') is formed here as the modified second alternative parameter, and - wherein an output signal (out) is generated on the basis of the first alternative parameter (ap1) and the limited second alternative parameter (ap2') and on the basis of a superposition (U1, U2) of the first input signal (E1) and the second input signal (E2).

3. Method according to Claim 1 or Claim 2, wherein a minimal sensitivity at the corresponding angle (θ) is modified on the basis of a modification, in particular a delimitation, of the first alternative parameter (ap1).

4. Method according to any one of the preceding claims, wherein a coefficient vector (w) of the coefficients (w1, w2) of the first input signal (E1) and the second input signal (E2) is formed in the first superposition (U1), wherein the first alternative parameter (ap1) is formed on the basis of a quotient (r) of the absolute values of the two coefficients (w1, w2), and wherein the second alternative parameter (ap2) is formed on the basis of a relative phase (φ) of the two coefficients (w1, w2) in relation to one another.

5. Method according to Claim 4 in conjunction with Claim 2, wherein an adapted first superposition parameter (a1') is formed on the basis of the first alternative parameter (ap1) and the limited second alternative parameter (ap2'), and wherein the superposition for generating the output signal (out) is formed by the first superposition (U1) on the basis of the adapted first superposition parameter (a1').

6. Method according to any one of Claims 1 to 4, - wherein the value of the first superposition parameter (a1) is converted into a corresponding real-value second superposition parameter (a2) and an associated value of a real-value amplification factor (m), wherein the real-value amplification factor (m) is assigned to a corresponding amplification of the second input signal (E2) in the formation of the first front or rear intermediate signal (Z1v, Z1h), - wherein the second superposition parameter (a2) is adapted in such a way that for a second superposition (U2), which is formed on the basis of the second superposition parameter (a2) from the first front intermediate signal (21v) and the first rear intermediate signal (Z1h) with amplification of the second input signal (E2) by the amplification factor (m), the angle (θ) of minimal sensitivity is limited to the specified angle range (Δθ) and an adapted second superposition parameter (a2') is generated in this way, and - wherein the output signal (out) is generated on the basis of the adapted second superposition parameter (a2') and the amplification factor (m) and on the basis of a superposition (U1, U2) of the first input signal (E1) and the second input signal (E2).

7. Method according to Claim 6, wherein the second superposition parameter (a2) is used as the second alternative parameter (ap2) and the amplification factor (m) is used as the first alternative parameter (ap1).

8. Method according to Claim 6 in conjunction with Claim 4, wherein the first alternative parameter (ap1) is formed on the basis of the quotient (r) of the absolute values of the two coefficients (w1, w2) of the first input signal (E1) and the second input signal (E2) in the first superposition (U1), wherein the second alternative parameter (ap2) is formed on the basis of the relative phase (φ) of the two coefficients (w1, w2) in relation to one another, and wherein the adaptation of the second superposition parameter (a2) takes place on the basis of the first alternative parameter (ap1) and the limited second alternative parameter (ap2'), and the adapted second superposition parameter (a2') is thus formed.

9. Method according to any one of Claims 6 to 8, wherein the output signal (out) is generated on the basis of the first superposition (U1), wherein for this purpose the value (a1.0) of the adapted first superposition parameter (a1) is ascertained on the basis of the adapted second superposition parameter (a2') and on the basis of amplification factor (m).

10. Method according to Claim 8, - wherein on the basis of the first input signal (E1) and the second input signal (E2) scaled by means of the real-value amplification factor (m), a second front intermediate signal (Z2v) and a second rear intermediate signal (Z2h) are each formed, and wherein the output signal (out) is generated on the basis of the second superposition (U2) using the adapted second superposition parameter (a2').

11. Method according to any one of the preceding claims, - wherein the first rear intermediate signal (Z1h) has a relative attenuation in a frontal direction (6), which is defined in particular on the basis of a direction from the second input transducer (M2) to the first input transducer (M1), and - wherein the first front intermediate signal (21v) has a relative attenuation in a direction opposite to the frontal direction (6).

12. Method according to Claim 11, wherein the first front intermediate signal (21v) and the first rear intermediate signal (Z1h) are each generated on the basis of a time-delayed superposition of the two input signals (E1, E2), and wherein the second input signal (E2) is delayed for the first front intermediate signal (21v) and the first input signal (E1) is delayed for the first rear intermediate signal (Z1h) here.

13. Method according to Claim 12, wherein a delay is implemented by means of an allpass filter at least in one frequency band.

14. Method according to any one of the preceding claims, wherein, in a first adaptation step, a first value of the complex first superposition parameter (a1) is ascertained, wherein the first value of the first superposition parameter (a1) is converted into the corresponding first and second alternative parameters (ap1, ap2), and the limited second alternative parameter (ap2') is ascertained therefrom, wherein a second value of the first superposition parameter (a1) is ascertained on the basis of the first alternative parameter (ap1) and the limited second alternative parameter (ap2'), and wherein the second value of the first superposition parameter (a1) is used for a second adaptation step.

15. Method according to any one of the preceding claims, wherein the first superposition parameter (a1) is ascertained by means of a least mean squares algorithm and / or by means of a gradient method.

16. Method according to any one of the preceding claims, wherein the output signal (out) is generated on the basis of a correction filter (ccor) for the frequency response in addition to the superposition of the first and the second input signal (E1, E2), wherein the correction filter (ccor) for the frequency response is selected such that the frequency response is flat for the frontal direction (6).

17. Hearing instrument (1), comprising - a first input transducer (M1) for generating a first input signal (E1) from a sound signal (4) of the surroundings, - a second input transducer (M2) for generating a second input signal (E2) from the sound signal (4) of the surroundings, and - a control unit (5), wherein the hearing instrument (1) is configured to carry out the method according to any one of the preceding claims.

Images