Method for directional signal processing of a hearing device

Abstract

The invention relates to a method for directional signal processing of a hearing device (1), in which a first input signal (E1) is generated from ambient sound (2) by means of a first input transducer (M1) of the hearing device (1) and a second input signal (E2) is generated from ambient sound (2) by means of a second input transducer (M2) of the hearing device (1), an angular direction (alpha) of the sound source (16, 17) with respect to a first reference direction (R1), in particular with respect to a front direction (12) of a wearer (10) of the hearing device (1), is identified at least approximately on the basis of the first input signal (E1) and the second input signal (E2), and wherein the sound source (16, 17) is positioned in the first reference direction (R1), in particular with respect to the front direction (12) of the wearer (10) of the hearing device (1). According to the invention, an orientation direction (vw1-3, vwk) of the sound source (16, 17), in particular an orientation direction with respect to a second reference direction (R2), is identified at least approximately on the basis of the first input signal (E1) and the second input signal (E2).

Description

TECHNICAL FIELD

[0001] The present invention relates to a directional signal processing method for a hearing device, wherein an angular direction of a sound source relative to a frontal direction of a wearer of the hearing device is at least approximately identified from a first input signal and a second input signal of the hearing device. BACKGROUND

[0002] In a hearing device, in particular in a hearing aid in the narrower sense, an input signal is processed into an output signal by a signal processing, in particular frequency band-specific, and fed to the auditory organs of a wearer of the hearing device, for example by means of a loudspeaker. The signal processing can here additionally be specifically adapted to the wearer, in particular to his hearing requirements, for example in a hearing aid in the narrower sense, a hearing impairment of the wearer can be supplied for.

[0003] The signal processing is usually also realized directionally in the sense that a plurality of input signals is processed into an output signal in such a way that sounds from different spatial directions of the environment are differentially included in the output signal, i.e. sounds from some spatial directions are suppressed and sounds from other spatial directions are enhanced.

[0004] For such directional microphones, the signal processing is usually based on a specific environmental situation model, for example a sound source located in the frontal direction of the wearer is basically considered a relevant or useful signal source, since it is assumed that the wearer will direct his gaze to the sound source relevant to him. Another assumption is, for example, that noise from the rear half-space of the wearer is basically interpreted as disturbing noise, if necessary depending on its frequency spectrum.

[0005] However, such a signal processing is limited in complex conversation environments with a plurality of conversation partners, in particular in environments with many additional disturbing noises, for example in a restaurant or a similar so-called "cocktail party" hearing situation, since the differentiation between relevant and irrelevant sound signals for the wearer is usually too coarse or too imprecise. SUMMARY

[0006] The technical problem addressed by the present invention is therefore to provide a method with which a hearing device is provided with the additional possibility of evaluating the relevance of a sound source for a wearer of the hearing device.

[0007] According to the application, the above-mentioned technical problem is solved by a method for directional signal processing of a hearing device, wherein a first input signal is generated from an ambient sound by means of a first input transducer of the hearing device and a second input signal is generated from the ambient sound by means of a second input transducer of the hearing device, wherein an angular direction of a sound source relative to a first reference direction, in particular relative to a wearer frontal direction of the hearing device, is at least approximately identified from the first input signal and the second input signal, and wherein an orientation direction of the sound source, in particular an orientation direction of the sound source relative to a second reference direction, is at least approximately identified from the first input signal and the second input signal. Advantageous and partly individually inventive design solutions are the subject of the following description of the application.

[0008] A hearing device here generally comprises any apparatus which is designed to generate an electrical input signal from an ambient sound by means of at least one, in particular acoustoelectric, input transducer; to process the input signal into an output signal by means of amplification and / or compression, in particular band-specific amplification and / or compression; and to generate a sound signal from the output signal and to feed this sound signal, in particular by means of an electroacoustic output transducer (for example a loudspeaker, a so-called balanced armature receiver, also including a bone conduction earpiece), to the hearing organs of a wearer of the apparatus. A hearing device here includes, in particular, earphones (such as "earbuds"), headsets, data glasses with loudspeakers, etc., which are equipped with corresponding input transducers. However, a hearing device also includes a hearing aid in the narrower sense, i.e. a device for supplying a hearing-impaired wearer, in which, in the processing of the input signal into the output signal, the input signal is amplified and / or compressed, in particular band-specifically, in order to at least partially compensate for the hearing impairment of the wearer in a user-specific manner by means of the output sound signal generated from the output signal.

[0009] A hearing device can also be designed, in particular, as a binaural hearing system with a first partial device and a second partial device, wherein, in this case, the hearing device can preferably also have (at least) two further input transducers, wherein the first and second input transducers are arranged in the first partial device and the two further input transducers are arranged in the second partial device. The applicability of the method is generally independent of this.

[0010] The first or second input transducer here comprises, in particular, any apparatus which is designed to generate a corresponding electrical signal from a sound signal. A pre-processing, for example in the form of a linear pre-amplification and / or an A / D conversion, can also take place, in particular in the case of the generation of the first or second input signal by means of the corresponding input transducer. The corresponding generated input signal is here given, in particular, by means of an electrical signal whose current fluctuations and / or voltage fluctuations substantially embody the air's sound pressure fluctuations.

[0011] The angular direction of the sound source with respect to the front direction of the wearer is understood here in particular as establishing a clear relationship between the two input transducers and the front direction of the wearer by wearing the hearing device as prescribed on the head (also in the case of a binaural hearing system as hearing device), in particular on or in the ear (in the case of a monaural hearing device), by means of a corresponding directional processing of the two input signals, it is thus possible, for example, to recognize in which angular direction the sound source is arranged by means of a temporal difference between the respective signal components in the first and second input signals, wherein the angle here is related to the first reference direction, which itself is given by the described front direction or can be defined in accordance with the front direction.

[0012] The orientation direction of the sound source is understood here in particular as the direction in which the sound source emits its maximum sound energy or in which the emission maximum of the sound energy of the sound source lies. In the case of a speaker as sound source, this direction is usually identical to the front direction of the speaker. In the case of a loudspeaker, the orientation direction is usually given by the axis of symmetry of the arrangement of the diaphragms. The orientation direction can be related here to the first reference direction, i.e. preferably to the front direction of the wearer of the hearing device, as a vector direction (i.e. moving towards the sound source), so that the vectors of the first and second reference directions are parallel. The second reference direction is preferably given, however, by the described angular direction. However, due to the fixed relationship between the first reference direction and the second reference direction, the definition of the second reference direction can preferably be chosen in accordance with the subsequent application, since all alternative definitions are equivalent to one another (except for a corresponding angular transformation).

[0013] The approximate recognition of the angular direction and / or the orientation direction is understood here in particular as including the predefinition of a plurality of, at least three (preferably more) discrete angular values as possible value range for the respective direction and the determination of the angular value that is closest to the actual angular direction or orientation direction and the output of this angular value as recognized angular direction or orientation direction.

[0014] The at least approximate recognition is understood here as being able to recognize the respective direction in the described discrete angular values or also to recognize the respective direction continuously, in particular exactly (within the range of the respective possibilities caused in particular by discretization, sampling rate and limited computing and storage capabilities, etc.).

[0015] The recognition of the angular direction or orientation direction from the first and second input signals here in particular comprises a direct recognition of signal components of the first and second input signals, i.e. a direct application of respective angle-dependent filters for recognizing the relevant direction (e.g. notch filters) to the first and second input signals, and / or a direct determination of the run-time difference of the signal components in the first and second input signals. Such filters for recognition can preferably also take into account the shadowing effect of the head, in particular by means of one or more head-related transfer functions, which are in particular dependent on the orientation direction.

[0016] The recognition of the angular direction or orientation direction from the first and second input signals, however, also comprises an application of the filters for recognizing the relevant direction to the first and second intermediate signals, and / or a determination of the run-time difference of the signal shares in the first and second intermediate signals. The first intermediate signal here is preferably derived exclusively from the first input signal (i.e. no signal share of the second input signal directly enters the second intermediate signal), and the second intermediate signal is preferably derived directly from the second input signal (i.e. there is preferably no signal share of the first input signal in the second intermediate signal).

[0017] The intermediate signals here can be generated from the respective input signals by means of single-channel preprocessing. The intermediate signals, however, can also be generated from the first and second input signals by means of directional preprocessing, respectively, for example as forward and backward cardioid signals.

[0018] The recognition of the angular direction of the sound source here can be carried out before the recognition of the orientation direction, or also together with the recognition of the orientation direction.

[0019] By recognizing the orientation direction of the sound source, which in particular also comprises the maximum emission direction of the sound energy, and in relation to the angular direction, it is possible to more easily distinguish whether the sound source is an interfering noise source or a possible useful signal source for the wearer. If the orientation direction of the sound source (in the given angular direction) results in only a small proportion of the sound energy being emitted in the direction of the wearer of the hearing device, the sound source can in particular be interpreted as a noise source. This can be distinguished, for example, by a respective angular threshold from the sound source, etc.

[0020] It is preferred here to identify the sound source as a wearer-related sound source in accordance with its orientation direction. This can also be identified in accordance with an angle threshold from the sound source. It is thereby possible, for example, to check whether the orientation direction is greater than (±) 10°, for example, or greater than (±) 22.5° or greater than (±) 45° with respect to the angle direction, i.e. the angle direction of the sound source as a second reference direction, preferably the inverse angle direction, i.e. whether the deviation of the orientation direction from the angle direction is greater than 10°, 22.5° or 45°. If this is not the case for the set threshold, i.e. the deviation is smaller, it is assumed that the sound source is essentially directed towards the wearer of the hearing system and it is classified as a wearer-related sound source.

[0021] It is preferred here to identify the speaker as the sound source. In principle, the method can also be applied to other types of sound sources, however it presents particular advantages in the case of a speaker as a sound source, since a speaker moves, in particular changes its orientation with respect to the wearer, during a conversation. In particular, a speaker can be identified from the spectral features themselves, i.e. the sound containing its speech share is identified as speech in accordance with characteristic spectral features, such as formants.

[0022] It is particularly preferred here to identify the speaker as a conversation partner of the wearer in accordance with the orientation direction. This means, in particular, that the speaker identified as a relevant sound source is also identified as being in a conversation with the wearer, in particular as long as the orientation direction of the conversation partner is sufficiently aligned with the wearer.

[0023] The directionality of the sound of the sound source is advantageously determined, wherein the orientation direction of the sound source is determined only if the determined directionality is not lower than a lower limit value. It is thereby possible to prevent wasting computing power on identifying a relevant sound source, for example, when the sound source can only be perceived vaguely. The determination of the directionality can be carried out, in particular, in accordance with spectral features of the sound, for example in accordance with an identification of a diffuse share of the sound.

[0024] It is expedient to detect whether the sound source is located in a first region, which is located in the front half-space of the wearer, wherein the orientation of the sound source is identified only if the sound source is located in the first region. It is thereby possible to prevent wasting computing power on identifying a relevant sound source when the sound source is located in a spatial range which is assumed to be irrelevant by the wearer in advance. The first region is preferably selected here as the range of [-67.5°, 67.5°] with respect to the first reference direction or the frontal direction of the wearer, in particular preferably the range of [-60°, 60°].

[0025] The orientation direction is advantageously identified approximately by selecting from a plurality, preferably at least three, discrete core orientation directions. Here, too, the orientation direction can be identified by identifying one of three orientation ranges: the orientation direction toward the wearer, the orientation direction passing in front of the wearer, the orientation direction passing behind the wearer. For the determination of whether the sound source is associated with the wearer, a higher resolution is generally not required, i.e. a greater range of values of the orientation direction (relative to the angular direction or the inverse angular direction). More complex determinations, in particular more complex arithmetic operations, can thereby be dispensed with.

[0026] It has proven to be further advantageous to provide a first number of angle-dependent filters, in particular two-channel or multi-channel, which are based on angles covering at least a partial region of space, wherein the first number of angle-dependent filters are respectively applied to the first and second input signals and / or to first and second intermediate signals derived from the first and second input signals, and thereby determine a respective set of angle-dependent features, and wherein the angle direction of the sound source is identified from the angle-dependent features. The angle-dependent features are here in particular quantitative features, i.e. the features for different angle directions can be ordered according to magnitude. The angle-dependent filters here "scan" a partial region of space or the entire space, so that the angle corresponding to the relevant feature assigned to it can be determined as the angle direction from the maximum or minimum of the angle-dependent features.

[0027] As angle-dependent features, here preferably the sound level of the sound source arranged in the relevant angle direction or the degree of attenuation of the sound is determined. The angle-dependent filters are here in particular designed as notch filters which are applied directly to the first and second input signals or to the first and second intermediate signals derived from the first and second input signals, respectively, and perform a corresponding spatial filtering with the relevant angle range, so that the maximum prominence or the maximum attenuation of one of the angles can be determined.

[0028] In another advantageous design, a second number of orientation-dependent filters is provided for at least one angle direction, which respectively correspond to one orientation direction in the case of the angle direction, wherein the second number of orientation-dependent filters are respectively applied to the first and second input signals and / or to the first and second intermediate signals, and thereby determine a second set of respective orientation-dependent features, and wherein the orientation direction is identified from the orientation-dependent features. The orientation-dependent features are here in particular quantitative features, i.e. the features for different orientation directions can be ordered according to magnitude.

[0029] The filter depending on the orientation direction here "scans" the space around the sound source with respect to the given angular direction, so that the orientation direction to which the respective feature corresponds can be determined, in particular, from a maximum or a minimum of the feature depending on the orientation direction.

[0030] The degree of attenuation of the sound of the sound source oriented in the respective orientation direction is here preferably determined as the feature depending on the orientation direction. The filter depending on the orientation direction is here in particular designed as a notch-filter, which is applied directly to the first and second input signals or to the first and second intermediate signals derived from the first and second input signals, respectively.

[0031] It is particularly preferred here to compare the comparison signal, which is produced by applying the relevant filter depending on the angle and / or depending on the orientation direction to the first and second input signals and / or to the first and second intermediate signals, to a reference signal, which preferably has an omnidirectional directional characteristic, with respect to the degree of attenuation. This comparison can be realized particularly easily, and, in addition, the reference signal also provides a comparison quantity for the overall sound level as a comparison quantity for the attenuation by the respective filter.

[0032] The reference signal is here in particular derived from the first input signal only. Thereby, the omnidirectional directional characteristic can be ensured without further signal processing steps.

[0033] It is advantageous if a plurality of discrete configurations can be preset, which are each given by a corresponding assigned angular direction of a first number of discrete angular directions and a corresponding assigned orientation direction of a second number of discrete orientation directions for the sound source, wherein for each configuration a filter depending on the orientation direction is provided, which is applied to the first and second input signals and / or to the first and second intermediate signals, respectively, and wherein the orientation direction is identified from the determined features depending on the orientation direction for the respective configuration.

[0034] This means in particular that only a determined number of possible configurations considered to be relevant are examined, wherein each configuration is given by an angular direction of a plurality of possible angular directions and a corresponding assigned orientation direction (in particular with respect to the respective angular direction or the inverse angular direction), wherein for one angular direction there can be exactly one or also a plurality of orientation directions (i.e. one or more configurations for this angular direction). This makes it possible to keep the effort low in that only a lower number of configurations (compared to the much larger number of possible configurations) have to be considered and the respective filter has to be provided and the corresponding assigned feature has to be determined.

[0035] It is expedient if for each orientation direction with respect to a given angular direction a head direction related transfer function is provided, which is applied to the first and second input signals and / or to the first and second intermediate signals, and wherein the orientation direction is identified from the determined features depending on the orientation direction for the respective configuration.O rientation D ependent H ead R elated T ransfer F The invention relates to a method for filtering sound from a sound source (ODHRTF, orientation-dependent head-related transfer function), wherein each ODHRTF represents the propagation path of sound from a sound source arranged in a given angular direction with a given orientation direction, and wherein the orientation-dependent filters are each designed based on the respectively assigned ODHRTF. This means, in particular, that the respective filter, for example, designed as a notch filter, takes into account or preferably includes the respectively assigned ODHRTF from the sound source (at a given angular direction and a given orientation direction) to the first or second input transducer. This ensures a precise mapping of the spatial propagation of sound from the sound source to the respective input transducer, thereby minimizing spatial distortion effects of the filter that could lead to inaccurate determination of the angular direction or orientation direction. The ODHRTF is an extension of the concept of a head-related transfer function (HRTF), which represents the propagation path of sound from a sound source arranged in a specific angular direction. However, unlike an ODHRTF, a head-related transfer function is not dependent on the orientation direction of the sound source. The respective ODHRTF is preferably measured and / or simulated prior to calibration (i.e., prior to the actual method).

[0036] Advantageously, the orientation direction is identified by means of an artificial neural network based on a feature that is dependent on the orientation direction. This means that the feature that is dependent on the orientation direction is transmitted as an input variable to an artificial neural network, for example a recurrent neural network ( R ecurrent N eural N In one embodiment, a neural network (RNN) or similar network is used, and the neural network outputs an at least approximately determined orientation direction as an output variable. In particular, the input variables can be processed simultaneously, thereby improving the runtime.

[0037] It is advantageous here that for each orientation direction with respect to a given angular direction, a head direction-dependent transfer function or said head direction-dependent transfer function is provided, wherein input quantities are used for the artificial neural network, which input quantities are derived from the first and / or second input signal filtered with the correspondingly assigned head direction-dependent transfer function. This makes it possible to calculate the orientation direction by means of an artificial neural network which is particularly suitable for this, since even smaller artificial neural networks can perform this kind of task comparatively efficiently, and these artificial neural networks can also be set up for said task with relatively little training effort (because of the small number of parameters in small artificial neural networks). On the other hand, by using acoustic features, such as the input signals filtered as described, as input quantities, the process of the signal processing for identifying the orientation direction can be significantly shortened.

[0038] The application also discloses a hearing device comprising a first input transducer for generating a first input signal from environmental sound, a second input transducer for generating a second input signal from environmental sound and a signal processing unit, wherein the hearing device is designed to carry out the above-described method, and wherein, in particular, the signal processing unit is designed to carry out the signal processing steps of the above-described method (by equipping the processor power and the working memory addressable by the signal processing unit accordingly and by program instructions).

[0039] The hearing device according to the application shares the advantages of the method according to the application. The advantages of the method and its expanded design can here be transferred by analogy to the hearing device. BRIEF DESCRIPTION OF DRAWINGS

[0040] Embodiments of the application are explained in more detail below with the aid of the drawings. In the drawings, respectively schematically:

[0041] Figure 1 A block diagram of a hearing device is shown,

[0042] Figure 2 A top view of a hearing situation of a wearer of a hearing device according to Figure 1 is shown, with a speaker as a sound source,

[0043] Figure 3 A flowchart of a method for determining an orientation direction of a speaker according to Figure 2 is shown in a block diagram,

[0044] Figure 4 A plurality of configurations is shown in a top view, which are respectively given by a specifically oriented speaker around a wearer of a hearing device according to Figure 1 ,

[0045] Figure 5 A diagram is shown for a plurality of configurations of a hearing device according toFigure 3 The method, according to Figure 4 The configuration applies the effects of each filter,

[0046] Figure 6 Different possible configurations are shown in each plan view with respect to three different angular directions of the speaker.

[0047] Corresponding components and variables are provided with the same reference numerals in all figures. DETAILED DESCRIPTION

[0048] Figure 1 The block diagram schematically illustrates a hearing device 1 having a first input transducer M1 and a second input transducer M2. The first input transducer M1 and the second input transducer M2 are each provided by a corresponding microphone. The first input transducer M1 is designed to generate a first input signal E1 from ambient sound 2 during operation of the hearing device 1. Correspondingly, the second input transducer M2 is designed to generate a second input signal E2 from ambient sound 2 during operation of the hearing device 1. The first and second input signals E1 and E2 are fed to a signal processing unit 4, where the two input signals E1 and E2 are processed into an output signal A1 and amplified and / or compressed, in particular in a frequency-band-specific manner. The signal processing of the two input signals E1 and E2 into the output signal A1 is particularly direction-dependent, meaning that the contributions of individual sound sources from different spatial directions to the ambient sound can be amplified to varying degrees. Furthermore, the signal processing can be tailored to the hearing requirements of the wearer of the hearing device 1.

[0049] The hearing instrument 1 further comprises an output transducer L1 , which is designed to generate an output sound signal 6 from the output signal A1 . Figure 1 The schematic diagram of a hearing device 1 shown in FIG shows a so-called behind-the-ear hearing device (BTE) with an earplug 8 in which an output transducer L1 is arranged. However, the hearing device 1 can also be designed as an in-the-ear hearing device (ITE), an in-the-canal hearing device (ITC), a completely in-the-canal hearing device (CIC), a receiver-in-the-canal hearing device (RIC), or other configurations, and in particular can also be designed as an earphone not solely or primarily for accommodating hearing loss. The hearing device 1 can also be designed as a binaural hearing system with a first local device and a second local device (not shown). In this case, the hearing device 1 can also have two additional input transducers, wherein the two additional input transducers are arranged in the second local device. The applicability of the method is generally irrelevant to this.

[0050] Figure 2The hearing situation of the wearer 10 of the hearing device 1 is schematically shown in a top view. The hearing device 1 can also be provided here by a binaural hearing system, which has a first partial device 1a and a second partial device 1b, wherein two input transducers Ml, M2 can be assigned to the two partial devices 1a, 1b (one input transducer Ml, M2 each in each partial device 1a, 1b) or each partial device 1a, 1b can have two input transducers Ml, M2 (input transducers of the partial device 1b are not shown). By arranging two input transducers Ml and M2 in the hearing device 1, a first reference direction Rl can be determined for a prescribed wearing of the hearing device 1, which is preferably chosen as the frontal direction 12 of the wearer 10. A sound source 16 given by a speaker S 1 is arranged in the front half-space 14 in an angular direction a with respect to the first reference direction Rl. In order to be able to identify whether the sound source 16 is a sound source relevant to the wearer 10, and thus whether the speaker S 1 is a conversation partner of the wearer 10, the orientation direction vw1 of the sound source 16 in which the sound source 16 emits its greatest acoustic energy is determined by the hearing device 1 from the first and second input signals E1, E2 in a manner still to be described. For the speaker 1 as the sound source 16, this orientation direction vw1 is identical to the speaking direction or line-of-sight direction of the speaker S 1.

[0051] The orientation direction vw1 is preferably defined here with respect to a suitably chosen second reference direction R2. Here, the second reference direction R2 is chosen as the (opposite) angular direction a of the speaker S 1, but can also be chosen identically to the first reference direction Rl, among others. Figure 2 A further sound source 17 given by a further speaker S2 is shown. The line of sight of the speaker S2 is not directed at the wearer 10, which means that its orientation direction vw2 is directed away from the wearer.

[0052] Figure 3 The flow of a method for determining the orientation direction vw1 of the sound source 16 described above is schematically shown according to a block diagram. The individual signal processing steps or signal analysis steps are preferably carried out here in the signal processing unit 14 of the hearing device 1, in particular on a correspondingly designed, preferably programmable signal processor and / or a circuit designed specifically for this application (for example an ASIC). In the individual analysis steps, the first and second input signals E1, E2 can be used directly here, i.e. a spatial filter for identifying the angular direction and / or the orientation direction of the sound source 16 is applied directly to the first and second input signals. However, the first and second input signals E1, E2 can also be further processed into at least two intermediate signals (not shown) (such as cardioid and anti-cardioid signals) with at least partial preservation of their spatial information, and the filter is applied to these intermediate signals.

[0053] In a first method step VI, the angular direction a of the sound source 16 is determined from the first and second input signals E1, E2. This corresponds to determining the so-called "Direction of Arrival (DoA)" and can be carried out, for example, by determining the time difference and / or level difference of the two input signals E1, E2 and / or from a first spatial filter F1(aj) which is applied to the two input signals E1, E2 (or intermediate signals derived therefrom) and is designed as a notch filter such that the input signals have a maximum attenuation at one angle aj. The individual first filters F1(aj) can "scan" the space by a variation of the angle argument aj.

[0054] If the angular direction a is determined, it can optionally be checked in a method step V2 whether the sound source 16 is located in the first region 18 in the front half-space 14 from the angular direction a. The first region 18 can here be defined, inter alia, by an angular region with Preferably, it is selected from the (semi-open) interval [45°, 90°), in particular from the interval [60°, 80°]. The subsequent method steps can be carried out, inter alia, only if the sound source 16 is located in the first region 18, i.e. the angular direction a lies in the angular range .

[0055] Furthermore, in an also optional method step V3, the directivity d of the sound of the sound source 16 can be determined, for example, from spectral characteristics of the input signals E1, E2. The subsequent method steps can be carried out, inter alia, only if the distance d of the sound source 16 is not below a preset lower limit (not shown).

[0056] In a next method step V4, the orientation direction vw1 of the sound source 16 is determined. To this end, in a first intermediate step V4.1, a second filter F2(a, vwk) which depends on the orientation direction is provided for the determined angular direction a, i.e. there are different sets of second filters F2(ai, vwk), F2(aj, vwk) for different angular directions ai, aj. For the second filter F2, here inter alia, an orientation direction-dependent ODHRTF(a, vwk, M1 / M2) from the sound source 16 arranged at the angular direction a to the first or second input transducer M1, M2 can be used. In a next intermediate step V4.2, the second filter F2(a, vwk) is applied to the first and second input signals E1, E2 for each orientation direction vwk (the set {E1, E2} of input signals is applied to the second filter F2), and, if necessary, normalized with a reference signal N1. Here inter alia, the first or second input signal E1, E2 can be used directly as the reference signal N1.

[0057] Finally, in an intermediate step V4.3, a minimum of the input signals E1, E2, as filtered with the respective second filter F2(a, vwk) and normalized with the reference signal N1, is formed with respect to the individual orientation directions vwk, and the corresponding assigned argument arg min is determined as the orientation direction vwi of the sound source 16. Here, the determination can be made approximately, i.e. only a limited number of orientation directions vwk (e.g. k = 1... 3 or k = 1... 5) is examined. The orientation direction vwi can here be related to the (opposite) angular direction a, such that as a value range for the orientation directions for k = 1... 3 only the following results are defined: "aligned to the wearer" (vwi = a or 180° + a), "passes from the side in front of the wearer", "passes from the side behind the wearer".

[0058] The orientation direction vwi can also be determined by a correspondingly trained artificial neural network DNN (dashed signal path), which receives the input signals E1, E2 filtered with the orientation direction dependent ODHRTF(a, vwk, M1 / M2) as input variables and outputs the orientation direction vwi as a result.

[0059] Figure 4 A top view schematically shows a plurality of configurations of the wearer 10 in a conversation situation, wherein each of these configurations is given by a speaker S1, S2, S3 and its specific orientation direction vwi, vw2, vw3. The speaker S1 sits opposite to the wearer 10 at a table 20, but the line of sight of this speaker S1 is directed forward to its own frontal direction 22 (and thus the orientation direction vwi of this speaker is not aligned to the wearer 10, but is parallel to the frontal direction 12 of the wearer 10; first configuration), while the speaker S2 sits next to the wearer 10, the line of sight of this speaker S2 is directed towards the speaker S1, thus its orientation direction vw2 is also directed towards the speaker S1 (second configuration). The speaker S3 is located diagonally behind the wearer 10, its orientation direction vw3 is aligned to the wearer 10 (third configuration). In a real-life lively conversation situation, the speakers S1, S2, S3 here speak at different points in time, sometimes interrupting each other.

[0060] Figure 5 A top view schematically shows a plurality of configurations of the wearer 10 in a conversation situation, wherein each of these configurations is given by a speaker S1, S2, S3 and its specific orientation direction vwi, vw2, vw3. The speaker S1 sits opposite to the wearer 10 at a table 20, but the line of sight of this speaker S1 is directed forward to its own frontal direction 22 (and thus the orientation direction vwi of this speaker is not aligned to the wearer 10, but is parallel to the frontal direction 12 of the wearer 10; first configuration), while the speaker S2 sits next to the wearer 10, the line of sight of this speaker S2 is directed towards the speaker S1, thus its orientation direction vw2 is also directed towards the speaker S1 (second configuration). The speaker S3 is located diagonally behind the wearer 10, its orientation direction vw3 is aligned to the wearer 10 (third configuration). In a real-life lively conversation situation, the speakers S1, S2, S3 here speak at different points in time, sometimes interrupting each other. Figure 3 Figure 4 ​The respective configurations of the second filter F2 are applied. In the first graph on the left, the respective output quantities of the second filter applied to the first and second input signals for the respective configurations, i.e. with respect to a given angular direction of the talker with its orientation direction, are plotted against the time axis for the three configurations K1, K2, K3. It can be seen that for the configuration K1 the first filter provides an approximate but distinguishable minimum up to a time point of nearly 6 seconds, while after this time point the second filter provides an identifiable minimum for the configuration K2. This indicates that the talker S1 of the first configuration is active up to a time point of nearly 6 seconds, after which the talker S2 of the second configuration is active. The respective results are shown in the second graph (upper right) against the time axis.

[0061] In the third graph on the left, the respective output quantities of the second filter for the respective configurations are plotted against the time axis again for the three configurations K1, K2, K3. It can be seen that for the configuration K3 the second filter provides a clear minimum up to a time point of nearly 6 seconds, while after this time point the second filter provides a clear minimum for the configuration K2. This indicates that the talker S3 of the third configuration is active up to a time point of nearly 6 seconds, after which the talker S2 of the second configuration is active. The respective results are shown in the fourth graph (lower right) against the time axis.

[0062] Figure 6 Possible different configurations for the orientation directions vw1-3 are shown for three different angular directions a (with respect to the frontal direction 12 of the wearer 10) of the talker S1.

[0063] The above configurations show the talker S1 with an angular direction a = 90° with respect to the first reference direction R1 given by the frontal direction 12 of the wearer 10 of the hearing device 1. In the configuration shown on the left, the orientation direction vw1 of the talker S1 is 0° with respect to the second reference direction R2 given by the (opposite) angular direction a. In the configuration shown next to it, the talker S1 has an orientation direction vw2 of 45° or an orientation direction vw3 of 90° with respect to the second reference direction R2.

[0064] The middle configurations show the talker S1 with an angular direction a = 45° with respect to the first reference direction R1. In the configuration shown on the left, the talker S1 has an orientation direction vw1 of 0° with respect to the second reference direction R2. In the configuration shown next to it, the talker S1 has an orientation direction vw2 of 45° or an orientation direction vw3 of 90° with respect to the second reference direction R2.

[0065] The lower configuration shows that the speaker S1 has an angular direction of α = 0° with respect to the first reference direction R1. In the lowermost configuration, the orientation direction vwi of the speaker S1 is 0° with respect to the second reference direction R2 given by the (opposite) angular direction α. In the configuration above, the speaker S1 has an orientation direction vw2 of 45° or an orientation direction vw3 of -45° with respect to the second reference direction R2.

[0066] Although the details of the application are explained and described in detail by means of preferred embodiments, the application is not limited to the disclosed examples and other variants can be derived therefrom by a person skilled in the art without departing from the scope of protection of the application.

[0067] List of reference signs

[0068] 1 hearing device

[0069] 1a / b local device (of the hearing device)

[0070] 2 ambient sound

[0071] 4 signal processing unit

[0072] 6 output sound signal

[0073] 8 earplug

[0074] 10 wearer

[0075] 12 frontal direction

[0076] 14 front half-space

[0077] 16 sound source

[0078] 17 sound source

[0079] 18 first region

[0080] 20 table

[0081] 22 frontal direction

[0082] A1 output signal

[0083] DNN artificial neural network

[0084] E1 / 2 first / second input signal

[0085] F1 first filter (depending on the angle)

[0086] F2 second filter (depending on the orientation direction)

[0087] L1 output transducer

[0088] K1-3 output quantity of the second filter (for configurations 1-3)

[0089] M1 / 2 1st / 2nd input converter

[0090] N1 reference signal

[0091] R1 / 2 first / second reference direction

[0092] S1-3 Speaker

[0093] V1-4 Method Steps

[0094] V4.1-3 Intermediate Steps

[0095] vw1-3, vwk orientation direction

[0096] α, αj angle direction

Claims

1. A method for directional signal processing of a hearing device (1), - wherein, - generating a first input signal (El) from an ambient sound (2) by a first input transducer (Ml) of the hearing device (1), generating a second input signal (E2) from the ambient sound (2) by a second input transducer (M2) of the hearing device (1), - wherein an angular direction (a) of a sound source (16, 17) relative to a first reference direction (Rl), in particular relative to a frontal direction (12) of a wearer (10) of the hearing device (1), is at least approximately identified from the first input signal (El) and the second input signal (E2), and - wherein an orientation direction (vw1-3, vwk) of the sound source (16, 17), in particular relative to a second reference direction (R2), is at least approximately identified from the first input signal (El) and the second input signal (E2).

2. The method according to claim 1, wherein - identifying the sound source as a sound source related to the wearer (10) from the orientation direction (vw1-3, vwk) of the sound source (16, 17).

3. The method according to claim 1 or 2, wherein - identifying a speaker (S1-3) as the sound source (16, 17).

4. The method according to claim 3 when dependent on claim 2, wherein - the speaker (S1-3) is identified as a conversation partner of the wearer (10) from the orientation direction (vw1-3).

5. The method according to any of the preceding claims, wherein - identifying a directivity (d) of the sound of the sound source (16, 17), and - wherein the orientation direction (vw1-3, vwk) of the sound source (16, 17) is determined only if the identified directivity (d) is not lower than a lower limit value.

6. The method according to any of the preceding claims, wherein - detecting whether the sound source (16, 17) is located in a first region (18) of a front half-space (14) of the wearer (10), and - wherein the orientation direction (vw1-3, vwk) of the sound source (16, 17) is identified only if the sound source (16, 17) is located in the first region (14).

7. The method according to any of the preceding claims, wherein - the orientation direction (vw1-3, vwk) is approximately identified by selecting from a plurality, preferably at least three, discrete core orientation directions.

8. The method according to any of the preceding claims, wherein - providing a first number of angle-dependent filters (Fl) depending on the angular direction (a), the angles (aj) on which the filters are based covering at least a partial area of the space, - wherein the first number of angle-dependent filters (Fl) are applied to the first and second input signals (El, E2) and / or to first and second intermediate signals derived from the first and second input signals (El, E2), respectively, and from this a respective set of angle-dependent features is determined, and - wherein the angular direction (a) of the sound source (16, 17) is identified from the angle-dependent features.

9. The method according to any of the preceding claims, wherein a second number of orientation direction-dependent filters (F2) is provided for at least one angle direction (a), which orientation direction-dependent filters respectively correspond to one orientation direction (vw1-3, vwk) in the case of the angle direction (a), wherein the second number of orientation direction-dependent filters (F2) is respectively applied to the first and second input signals (E1, E2) and / or the first and second intermediate signals, and from this a second set of orientation direction-dependent features is determined, and wherein the orientation direction (vw1-3, vwk) is identified from the orientation direction-dependent features.

10. The method according to any one of the preceding claims, wherein a plurality of discrete configurations is predefined, which configurations are respectively given by a corresponding assigned angle direction (a) of a first number of discrete angle directions (aj) for a sound source (16, 17) and a corresponding assigned orientation direction (vw1-3) of a second number of discrete orientation directions (vwk), wherein for each configuration an orientation direction-dependent filter (F2) is provided, which is respectively applied to the first and second input signals (E1, E2) and / or the first and second intermediate signals, and from this a respective orientation direction-dependent feature is determined for the respective configuration, and wherein the orientation direction (vw1) is identified from the determined orientation direction-dependent features.

11. The method according to claim 9 or 10, wherein for each orientation direction (vw1-3, vwk) with respect to a given angle direction (a) a head direction-related transfer function (ODHRTF) is provided, wherein each head direction-related transfer function (ODHRTF) represents a propagation path of a sound of a sound source (16, 17), which sound source is arranged in the relevant angle direction (a) and in the orientation direction (vw1-3, vwk), and wherein the orientation direction-dependent filter (F2) is respectively formed from the respectively corresponding assigned head direction-related transfer function (ODHRTF).

12. The method according to any one of claims 8 to 11, wherein a degree of attenuation of a sound of a sound source (16, 17) is determined as an angle- and / or orientation direction-dependent feature, which sound source is arranged in the relevant angle direction (a) and / or oriented in the corresponding assigned orientation direction (vw1-3, vwk).

13. The method according to claim 12, wherein a comparison signal (N1), which is preferably of all-around directional characteristic, is compared with a reference signal (N1) having a preferably omnidirectional directional characteristic for the degree of attenuation, which comparison signal is produced by applying the relevant angle- and / or orientation direction-dependent filter (F1, F2) to the first and second input (E1, E2) signals and / or the first and second intermediate signals.

14. The method according to claim 13, wherein the reference signal (N1) is derived exclusively from the first input signal (E1).

15. The method according to any one of claims 8 to 14, wherein The orientation direction (vw1-3, vwk) is recognized by means of an artificial neural network (DNN) depending on features which depend on the orientation direction.

16. The method according to claim 15, wherein a head direction dependent transfer function or the head direction dependent transfer function (ODH RTF) is provided for each orientation direction (vw1-3, vwk) with respect to a given angular direction (a), and wherein an input quantity is used for the artificial neural network (DNN), which input quantity is derived from the first and / or second input signal (E1, E2) filtered with the corresponding assigned head direction dependent transfer function (ODH RTF).

17. A hearing device (1) comprising - a first input transducer (M1) for generating a first input signal (E1) from an ambient sound (2) - a second input transducer (M2) for generating a second input signal (E2) from the ambient sound (2), and - a signal processing unit (4), wherein the hearing device (1) being designed to perform the method according to any one of the preceding claims.

Images