Method for operating a hearing system

The invention is related to a method for operating a hearing system comprising at least a first hearing device and a second hearing device, the first hearing device comprising at least a first reference microphone and a first auxiliary microphone, and the second hearing device comprising at least a number of microphones, wherein for the first hearing device, a first reference signal and a first auxiliary sig nal are generated from an environment sound by the first reference microphone and the first auxiliary microphone, respectively, and a first pre-processed signal is generated by applying a direction-sensitive pre-processing to the first reference and auxiliary signals, wherein for the second hearing device, a second pre-pro cessed signal is generated, said second pre-processed signal being representa tive of said environment sound, by means of said number of microphones, and wherein a direction-sensitive signal processing task is performed on the first pre- processed signal and the second pre-processed signal.

In many applications of binaural hearing systems with two hearing devices, a di rectional signal processing task is implemented by some type of directional pre processing for each hearing device, and using the pre-processed signals for finally performing the desired direction-dependent signal processing task. For example, blocking matrices may be generated from the microphone signals of the micro phones in the hearing devices, using different combinations of the microphones of the full microphone array consisting of all of the hearing system’s microphones, and the information of the different blocking matrices may be used for direction-de- pendent noise reduction or source localization. This in particular holds for those binaural hearing systems in which each of the hearing devices comprises at least two or even more microphones. In such a case, very often, local pre-processing is applied to the several microphone signals obtained from an environment sound for each hearing device. For example, a sin gle hearing device of the binaural hearing system may comprise two microphones, and the resulting to microphone signals are being locally pre-processed by some direction-dependent algorithm, to generate a local signal which already may show some noise reduction or other kind of enhancement (e.g., by attenuating signals from the back hemisphere of the user of the system). A direction-dependent signal processing task, such as source localization or beamforming, may then be per formed by using the corresponding local pre-processed signals from each side.

For a direction-dependent pre-processing of these microphone signals, the relative positions and the resulting level differences and sound delays for the involved mi crophones have to be taken into account, as well as the position of the micro phones with respect to the user’s head. This can be done via a head related trans fer function (FIRTF) for each microphone, which represents the propagation of a generic sound signal from a certain spatial direction towards the corresponding mi crophone and also takes into account shadowing effects coming from the head and/or the pinna of the user. However, in case that an overall direction-dependent signal processing task shall also be implemented by use of one or more HRTFs, the local pre-processing may introduce certain inaccuracy with respect to the transfer function is to be used for the global directional processing.

It is therefore an object of the invention to provide a method for operating a hear ing system, which allows for a direction-dependent local pre-processing of the sig nals of the hearing system’s individual devices without distorting the performance of a global direction-dependent signal processing using the hearing device’s out put signals for said global processing. It is furthermore the object of the invention to provide a hearing system comprising certain hearing devices, which allows for a local pre-processing in said hearing devices prior to a global, direction-dependent signal processing based on signals generated from the local pre-processing in each hearing device with as little spatial distortion as possible. According to the invention, the first object is achieved by a method for operating a hearing system, said hearing system comprising a first hearing device and a sec ond hearing device, the first hearing device comprising at least a first reference mi crophone and a first auxiliary microphone, and the second hearing device compris ing at least a number of microphones, wherein for the first hearing device, a first reference signal and a first auxiliary signal are generated from an environment sound by the first reference microphone and the first auxiliary microphone, respec tively, and a first pre-processed signal is generated by applying a direction-sensi tive pre-processing to the first reference and auxiliary signals by means of corre sponding first reference and first auxiliary pre-processing coefficients, respectively, and wherein for the second hearing device, a second pre-processed signal is gen erated by means of said number of microphones, said second pre-processed sig nal being representative of said environment sound, and a second position related transfer function is provided, representative of the propagation of a generic sound signal from said given angle towards the second hearing device when the second hearing device is mounted at a specific location, in particular, on said users body.

According to the method, a first frontal direction is defined as the direction from the first auxiliary microphone towards the first reference microphone, and said first ref erence and auxiliary coefficients are chosen and accordingly applied to the first reference signal and the first auxiliary signal in such a way that said first pre-pro cessed signal shows a maximal attenuation for a generic sound signal originating from an angular range of [+90°, +270°], preferably of [+105°, +255°] and most preferably of [+125°, +235°], with respect to the first frontal direction as a result of said direction-sensitive pre-processing, and a first head related transfer function is provided, said first head related transfer function being representative of the prop agation of a generic sound signal from said given angle towards the first hearing device when the first hearing device is mounted on the head of said user, wherein a direction-sensitive signal processing task is performed on the first pre-processed signal and the second pre-processed signal, using the first head related transfer function and the second position related transfer functions for said task. Embodi ments of particular advantage, which may be inventive in their own right, are out lined in the depending claims and in the following description.

According to the invention, the second object is achieved by a hearing system, comprising a first hearing device with at least a first reference microphone and a first auxiliary microphone, and a second hearing device with at least a number of microphones, the hearing system further comprising a control unit with at least one signal processor, wherein the hearing system is configured to perform the method for operating as given above.

The hearing system according to the invention shares the advantages of the method for operating a hearing system according to the invention. Particular as sets of the method and of its embodiments may be transferred, in an analogous way, to the hearing system and its embodiments, and vice versa. In an embodi ment, the hearing system may be configured as a binaural hearing system, wherein the first hearing device and said second hearing device are configured to be worn by a user on and/or at different ears during operation of the binaural hear ing system.

Generally, a hearing system is understood as meaning any system which provides an output signal that can be perceived as an auditory signal by a user or contrib utes to providing such an output signal. In particular, the hearing system may have means adapted to compensate for an individual hearing loss of the user or contrib ute to compensating for the hearing loss of the user. The hearing devices in partic ular may be given as hearing aids that can be worn on the body or on the head, in particular on or in the ear, or that can be fully or partially implanted. The hearing system may comprise other types of hearing devices, such as ear-buds. In particu lar, a device whose main aim is not to compensate for a hearing loss, for example a consumer electronic device (mobile phones, MP3 players, so-called “hearables” etc.), may also be considered a hearing system. Within the present context, a hearing device can be understood as a small, bat tery-powered, microelectronic device designed to be worn behind or in or else where at the human ear or at or on another body part by a user. A hearing device in the sense of the invention comprises a battery, a microelectronic circuit compris ing a signal processor, and the specified number of microphones, wherein a micro phone shall be understood as any form of acousto-electric input transducer config ured to generate an electric signal from an environment sound. The signal proces sor is preferably a digital signal processor.

In particular, the first hearing device is a hearing device to be worn by the user on and/or at one of his ears during operation of the hearing system and in particular providing an output sound signal to the respective hearing of the ear. According to variations, the first hearing device need not comprise a traditional loudspeaker as output transducer. Examples that do not comprise a traditional loudspeaker are typically found in the field of hearing aids in the stricter sense, i.e. , hearing devices designed and configured to correct for a hearing impairment of the user, and out put transducers may be also be given by cochlear implants, implantable middle ear hearing devices (IMEHD), bone-anchored hearing aids (BAHA) and various other electro-mechanical transducer-based solutions including, e.g., systems based on using a laser diode for directly inducing vibration of the eardrum. How ever, a hearing aid may also comprise a traditional loudspeaker as output trans ducer.

The second hearing device may be configured as a hearing device to be worn by the user at or in the other ear (than the first hearing device), and may comprise an acoustic output transducer as described for the case of the first hearing device. Thus, the hearing system, in particular, may be given by a binaural hearing system with two hearing devices, configured to be worn by the user on and/or at different ears during operation.

The first hearing device and the second hearing device, however, may also be given by different types of devices, wherein the second hearing device may be given as an additional or auxiliary device of the hearing system not necessarily lo cated at the other ear, but, e.g., worn around the neck, or on a wrist. The second hearing device, thus, need not be a hearing device with an output transducer of its own, but may be a device that, using its microphone(s), provides one or more in put signals for signal processing, such that a resulting signal from said signal pro cessing using also the signals generated from the second hearing device, is repro duced to the hearing of the user by the output transducer of the first hearing aid

Apart from the first reference microphone and the first auxiliary microphone, the first hearing device may also comprise one or even more further microphones, each of which configured to generate a respective signal from the environment sound. Preferably, the second hearing device comprises an equal number of mi crophones as the first hearing device, however, this is not a necessary condition for operation of the hearing system according to the method. Preferably, during operation, the first and second hearing device are located noticeably apart from each other. In particular, each microphone of the hearing system may have an omni-directional characteristic.

The first reference microphone may in particular be given by a front microphone and the first auxiliary microphone by a back microphone of the first hearing device, i.e. , due to the positioning of the first hearing device for operation of the hearing system, the first reference microphone is located before the first auxiliary micro phone with respect to a frontal direction of the first hearing device.

Preferably, the first pre-processed signal is generated from the first reference sig nal and the first auxiliary signal by applying the first reference pre-processing coef ficient to the first reference signal, and the first auxiliary pre-processing coefficient to the first auxiliary signal, preferably as multiplications in each case. Thus, the first reference signal in particular may be generated as a weighted sum of the first reference and auxiliary signal, weighted by the first reference and auxiliary pre processing coefficients. In particular cases, one of the first reference or auxiliary pre-processing coefficient may be trivial in the sense that it may be set to unity up to a global gain and/or phase factor shared with the respective other pre-processing coefficient.

As a part of the direction-sensitive pre-processing for generating the first pre-pro cessing signal, the first reference and auxiliary pre-processing coefficients are de termined by imposing the spatial condition onto the resulting first pre-processed signal that the angle at which the first pre-processed signal shows a maximal at tenuation, i.e. , the angle at which any impinging probe sound signal would get at tenuated the most when varying the angle of the probe sound source, falls into the angular range of [+90°, +270°], and preferably [+105°, +255°], most preferably [+125°, +235°], with respect to the first frontal direction.

In particular, this means that the first frontal direction is defined as a direction of preference for the first hearing device, giving a spatial reference for the surround ing of the first hearing device. The angular range is then preferably understood in terms of a vector with an origin in the first hearing device and an angle from the mentioned range of [+90°, +270°], preferably of [+105°, +255°] and most prefera bly [+125°, +235°], with respect to the first frontal direction, i.e., an angular range of ± 90° (preferably of ± 75° and most preferably of ± 55°), around the 180° or first backward direction (opposite to the first frontal direction) here, the assumption is made that the size of the first hearing device, and thus, possible differences in the choice of the origin of said vector, are negligible in comparison to the distance of the sound source.

In this respect, the first pre-processed signal may in particular be a beamformer signal, wherein a frequency-dependent phase factor may be implemented in at least one of the first reference or auxiliary pre-processing coefficients in time-fre quency domain. In particular, either of the first reference and auxiliary pre-pro cessing coefficients in time-frequency domain may be given by a spectral ampli tude and said phase factor in time-frequency domain. The direction-sensitive pre processing in general may be implemented in any way, and in particular comprises any linear combination of the first reference and auxiliary pre-processing signal - with possibly frequency-dependent linear coefficients (the first and second pre-pro cessing coefficients) - that may result in a non-trivial directional characteristic of the first pre-processed signal, and thus, leads to a maximum for the attenuation at some angle. This maximum attenuation angle, according to the method, shall be restricted to the indicated angular range. The direction-sensitive pre-processing implementation may be given by sum-and-delay beamforming, differential micro phones arrays (differential beamforming), delay-and-subtract beamforming, linear constraints minimum variance beamforming, minimum variance distortionless re sponse beamforming, among others.

Particularly, the first pre-processed signal may be generated such that its direc tional characteristic shows a cardioid shape or a figure-of-eight shape or any smooth transitional shape between these cases, such as a hyper-cardioid shape or a super-cardioid shape, which preferably may be described as a convex combi nation of a cardioid shape and a figure-of-eight shape (or in an equivalent formula tion, for instance, as a linear combination of a cardioid shape and an anti-cardioid shape, as long as the spatial constraints on the angle of maximal attenuation are fulfilled, which in this case may translate to a constraint for the respective linear factor). However, the direction-sensitive pre-processing shall not be limited to these cases, but may also contain other shapes for directional characteristics, as long as the angle of maximal attenuation for the directional characteristic, accord ing to the method, falls into the angular range of [+90°, +270°] (i.e. , the back hemi sphere), and preferably [+105°, +255°], most preferably [+125°, +235°], with re spect to the first frontal direction.

The second pre-processed signal is generated by means of the number of micro phones of the second hearing device in the sense that the second hearing device may comprise only one microphone, and the respective microphone signal, gener ated from the environment sound by said microphone of the second hearing de vice is then also used as the second pre-processed signal, or may receive single channel pre-processing, such as frequency dependent amplification for generating the second pre-processed signal. However, the second hearing device may also comprise more than one micro phone. In particular, the second pre-processed signal may be generated in a simi lar way as the first pre-processed signal, i.e. , the second hearing device may com prise a second reference microphone and a second auxiliary microphone, each of which generating a respective signals from the environment sound which are be ing applied to a direction-sensitive pre-processing by means of corresponding pre processing coefficients, just as in the case for the first pre-processed signal and its generation from the first reference and auxiliary signal. In particular, the second pre-processed signal is being representative of the environment sound, in the sense that it contains signal contributions from one or more signals directly gener ated by a microphone from the environment sound.

By means of the first head related transfer function, in particular, propagation time differences (that may cause phase differences in time-frequency domain) between the hearing devices may be taken into account (by a respective phase factor in the first head related transfer function with respect to a global phase frame or to the second position related transfer function), as well as other possible differences in the propagation from the generic sound source located at said given angle to wards one or another hearing device, in particular, the shadowing by the head (and possibly the pinna) of the user affecting the first reference and auxiliary mi crophone when the first hearing device is mounted properly for operation on the user’s head, possibly causing also level differences.

The second position related transfer function may also be given by a head related transfer function, in case the second hearing device is configured to be worn by the user at or on his head. In case that the second hearing device is configured for a different position on the user’s body, e.g., worn at the chest using a strap around the neck, or worn at the wrist, the second position related transfer function has to be adapted accordingly, in particular with respect to the shadowing effects (and possible phase and level differences in case of two or more microphones in the second device) that may occur at this position. The direction-sensitive signal processing task may be any possible task using at least two input signals generated at different locations, and preferably also respec tive transfer functions for each location, which processes and/or extracts any kind of spatial acoustic information encoded in these at least two input signals. In par ticular, said task may be given by the generation of the output signal using signal contributions of the first and second pre-processed signal, in particular by a weighted sum of said pre-processed signals, where in the weighting coefficients are given by the first head related transfer function and second position related transfer function, respectively. The direction-sensitive signal processing task may, however, also be given by a control operation in the sense that a control signal or, more generally, a control information is obtained, such as the location of a domi nant sound source, or similar control operations.

The restriction of the angular range for a maximal attenuation, i.e. , for a “minimal” direction or even a null direction, of the first pre-processed signal, possible spatial inaccuracies due to the direction-sensitive pre-processing in the first hearing de vice which might lead to the distortion of spatial information, can be reduced. This is particularly true for the case that the direction-sensitive signal processing task, which uses the first and second pre-processed signal, operates in the frontal hemi sphere (with respect to the first frontal vector), i.e., in an angular range of ± 90° around the first frontal direction, e.g., by localizing a sound source in the frontal hemisphere, or by generating a beamformer signal directed towards a sound source in the frontal hemisphere. Local pre-processing in the first hearing device is then essentially restricted to the complementary space, i.e., to the back hemi sphere.

Preferably, said number of microphones of the second hearing device comprises at least a second reference microphone and a second auxiliary microphone, wherein for the second hearing device, a second reference signal and a second auxiliary signal are generated from said environment sound by the second refer ence microphone and the second auxiliary microphone, respectively, a second frontal direction is defined as the direction from the second auxiliary microphone towards the second reference microphone, and said second pre-processed signal is generated by applying a direction-sensitive pre-processing to the second refer ence and second auxiliary signal by means of corresponding second reference and second auxiliary pre-processing coefficients, respectively, to be accordingly chosen and applied to the second reference signal and the second auxiliary signal in such a way that said second pre-processed signal shows a maximal attenuation for a generic sound signal originating from an angular range of [+90°, +270°], pref erably of [+125°, +235°], with respect to the second frontal direction. One of the two hearing devices is to be worn by the user on or at his left ear during operation of the hearing system, while the other hearing device is to be worn on or at his right ear.

In this respect, the local pre-processing in the first and second hearing device can be performed by similar or even the same algorithms. However, the second pre- processed signal may differ from the first pre-processed signal even in case of equal pre-processing algorithms due to the mentioned head shadowing effects. These differences are then also reflected by the corresponding first and second head related transfer functions.

In an embodiment, as said second position related transfer function, a second head related transfer function is provided, said second head related transfer func tion being representative of a generic sound signal from said given angle towards the second hearing device when second hearing device is mounted on the head of said user, the first and second hearing device being mounted on different sides of the head. This means that for the case that the two hearing devices are to be mounted at the right and left side of the user’s head (this shall not establish any correspondence which device to be mounted on which side), as the second posi tion related transfer function, a second head related transfer function with similar properties as the first head related transfer function is used.

In an embodiment, as said direction-sensitive signal processing task, an angle of a sound source is determined and/or a beamformer signal is generated, said beam- former signal containing signal contributions from the first and second pre-pro cessed signal. For these tasks, the method shows particular advantages in that the spatial distortion is minimized by matching the first head related transfer func tion to the corresponding first pre-processed signal. Advantageously, for determin ing said angle of a sound source, a set of spatial filters is generated by means of said first and second head related transfer functions, each of said spatial filters forming an attenuation notch in space towards a different angle. For a source lo calization with said filters, using the first - and possibly the second - head related transfer function generated according to the method from the respective local pre processing coefficients, yields a particularly high accuracy.

In an embodiment, the first pre-processed signal is generated by means of an adaptive beamforming process employing said first reference and first auxiliary pre-processing coefficients. In particular, the first reference signal and the first auxiliary signal may be used to derive two respective intermediate basis signals, such as a forward cardioid and a backward cardioid signal (sometimes referred to as an anti-cardioid), and the adaptive beamforming may be performed by using said intermediate basis signals. The first reference and first auxiliary pre-pro cessing coefficients may then be derived from the respective coefficients for the in termediate basis signals obtained via the adaptive beamforming, and by the re spective relations for the first reference and auxiliary signal in the intermediate ba sis signals.

In an embodiment, as said first head related transfer function, a first reference head related transfer function or a first auxiliary head related transfer function is provided, being representative of the propagation of a generic sound signal from a given angle towards the first reference microphone or towards the first auxiliary mi crophone, respectively, when located at a respective position on the head of said user. The first reference microphone and the first auxiliary microphone, e.g., may be given as the respective front and rear microphone of the first hearing device, wherein the front/rear label is assigned according to the position of each micro phone when the hearing device is worn as intended and provisioned for normal operation. For a given hearing device, a relative transfer function of a specific microphone of the hearing device is particularly easy to measure, as the measurement may be done by using the corresponding microphone signal without any further input from other microphones.

The first pre-processed signal generated via a beamformer (i.e. , resulting from the direction sensitive pre-processing) has an angular restriction in the back hemi sphere in order not to affect (i.e., not to distort “too much”) the first head related transfer function. As a consequence, the direction-sensitive signal processing task can still be successfully performed. Furthermore, the first head related transfer function can be then approximated by the first reference head related transfer function.

Similar considerations may hold for the second hearing device. In particular, as said second head related transfer function, a second reference head related trans fer function or a second auxiliary head related transfer function is provided, being representative of the propagation of a generic sound signal from a given angle to wards the second reference microphone or towards the second auxiliary micro phone, respectively, when located at a respective position on the head of said user.

The attributes and properties as well as the advantages of the invention which have been described above are now illustrated with help of drawings of embodi ment examples. In detail, figure 1 shows a schematic block diagram of a binaural hearing system, figure 2 shows a schematic top view of a user of the binaural hearing system of figure 1 in an environment with different sound sources, figure 3 shows polar plot diagrams of directional characteristics for a locally pre-processed signal of the binaural hearing system of figure 1 , figure 4 shows a polar plot diagram for restricting the angular range of the null direction of the directional characteristics shown in figure 3, and figure 5 shows a schematic block diagram of a method for operating the bin aural hearing system to figure 1 in the environment shown in figure 2 by means of the restrictions shown in figure 3.

Parts and variables corresponding to one another are provided with the same ref erence numerals in each case of occurrence for all figures.

In figure 1 , a schematic block diagram for the signal flow in a hearing system 1 is shown. The hearing system 1 is given by a binaural hearing system 2 which com prises a first hearing device 6 and a second hearing device 8. However, in differ ent embodiments, the second hearing device 8 might also be given by some other type of external device. The binaural hearing system 2 in an embodiment may also comprise an external control device (not shown), though such an external control device is optional. The first hearing device 6 comprises a first reference micro phone 14 and a first auxiliary microphone 16, the second hearing device 8 com prises a second reference microphone 18 and a second auxiliary microphone 20.

The first reference microphone 14 may be given by a front microphone and the first auxiliary microphone 16 by a back microphone of the first hearing device 6, i.e. , during normal operation of the hearing system 1, due to the positioning of the first hearing device 6 for operation, the first reference microphone 14 is located be fore the first auxiliary microphone 16 with respect to a frontal direction (not shown). A similar arrangement may hold for the second reference and auxiliary micro phone 18. 20 in the second hearing device 8.

Each of the mentioned microphones has an a priori omni-directional characteristic in the sense that the microphones are configured and designed to have an equal sensitivity for all directions. In a way not shown in detail, the first hearing device 6 further comprises a control unit with at least one signal processor, and an output transducer for converting an output signal into an output sound that it presented to the hearing of a user 21 of the binaural hearing system 12. Likewise, the second hearing device 8 may also comprise a similar control unit and an output trans ducer.

An environment sound 22 is converted into a first reference signal sir by the first reference microphone 14, into a first auxiliary signal sia by the first auxiliary micro phone 16, into a second reference signal S2r by the second reference microphone 18, and into a second auxiliary signal as to a by the second auxiliary microphone 20. In a way yet to be described, a direction-sensitive pre-processing 24 is applied to the first reference signal sir and the first auxiliary signal sia, and as a result, a first pre-processed signal spi is generated. The direction-sensitive pre-processing in the present case is given by a first local beamformer 26. In a similar way, a di rection-sensitive pre-processing 28, given by a second local beamformer 30, is ap plied to the second reference signal S2r and the second auxiliary signal S2a, and as a result, a second pre-processed signal sp2 is generated. The second pre-pro cessed signal sp2 is transmitted to the first hearing device 6 in order to perform said direction-sensitive signal processing task.

For operation of the binaural hearing system 2, the user 21 is wearing the binaural hearing system 12 on his head 31 , i.e. , he is wearing the first hearing device 6 on the left side 32 of his head 31 , on or at his left ear, and the second hearing device 8 on the right side 34 of his head 31 , on or at his right ear. Obviously, the assign ment of first and second hearing device to left and right ear may be interchanged.

In figure 2, a schematic top view shows the location of the user 21 wearing the bin aural hearing system 2 of figure 1 , and other sound sources in an environment 35. The first hearing device 6 has a first frontal direction 36, as a direction of prefer ence for its microphones, i.e., for the first reference microphone 14 and the first auxiliar microphone 16. The first frontal direction 36 is defined as the direction from the first auxiliary microphone 16 to the first reference microphone 14. The second hearing device 8 has a second frontal direction 40 as a direction of prefer ence for its microphones, i.e., for the second reference microphone 18 and the second auxiliary microphone 20. The second frontal direction 40 is defined as the direction from the second auxiliary microphone 20 to the second reference micro phone 18.

Depending on the specific design of the first and second hearing device 6, 8 and on the resulting positions on the head 31 of the user 21 , the first and second frontal directions 36, 40 may coincide (i.e., the respective vectors if the first and second frontal direction 36, 40 may be parallel); however, it is also possible that due to design and construction of the binaural hearing system 2, the first and sec ond direction 36, 40 are different.

The direction-sensitive pre-processing 24 on the first reference signal sir and the first auxiliary signal sia, as shown in figure 1 , is done in an adaptive way, i.e., the first null direction 44 tracks the source 46 in the back hemisphere 47 (delimited by the dotted line). A corresponding directional characteristic 45 for the resulting first pre-processed signal spi is shown (dashed lines). In an analogous way, the direc tion-sensitive pre-processing 28 of figure 1 on the second reference signal S2r and the second auxiliary signal S2a for generating the second pre-processed signal sp2 may either be fixed, in particular giving a fixed second null direction (not shown), or adaptive with respect to an interferer. Preferably, the first and second pre-pro cessed signal sp1 , sp2 are both generated by adaptive direction-sensitive pre-pro cessing 24, 28. In the latter case, due to shadowing effects of the head 31 and also of the ears, the direction-sensitive pre-processing 28 of the second hearing device 8 may adapt to a different interferer than the direction-sensitive pre-pro cessing 24 of the first hearing device 6.

The direction-sensitive signal processing task to be performed by the binaural hearing system 2 according to figure 1 may be given by the localization of a domi nant sound source 50 in the environment 35 of the binaural hearing system 12, i.e., by finding an angular source direction 52 for the sound source 50 with respect to a global direction of preference 54 for the binaural hearing system 2, said global direction of preference being derived from the first and second frontal directions 36, 40 (e.g., as the angular mean direction). Said task may also be given by gen erating a first beamformer output signal, preferably pointing towards the dominant sound source 50, to be converted into an output sound by an output transducer of the first hearing device 6. In figure 2, the first beamformer output signal is repre sented by the main lobe of its respective directional characteristic 55 (solid line).

In an analogous way, a direction-sensitive signal processing task may be per formed in the second hearing device 8, based on the (local) second pre-processed signal sp2, and on the (remote) first pre-processed signal sp1 that has been trans mitted from the firs hearing device 6 to the second hearing device 8 for performing said task.

In figure 3, polar plot diagrams of directional characteristics 60, 62, 64 for the first (and possibly second) pre-processed signal spi (respectively, sp2) of figure 1 are shown. The directional characteristics 60, 62, 64 represent the sensitivity of the first pre-processed signal spi, as generated from the first reference signal sir and the first auxiliary signal sia by the respective direction-sensitive pre-processing, with respect to a probe sound signal when varying the angular direction of a corre sponding probe sound source (not shown) with respect to the origin 66. Note the indicated directions of +90°, +180° and +270° = -90° in the left and the middle dia gram for reference of conventions. Said origin 66 of the polar plots may, e.g., be given by the center of the first reference microphone 14 and the first auxiliary mi crophone. For practical applications, the dimensions of the first hearing device 6 may be assumed small compared to the distance to said probe sound source (or other sound sources in a realistic scenario), such that the origin 66 may be ap proximated by a point position of the first hearing device 6 with respect to the posi tion of another sound source.

Note that the directional characteristics 60, 62, 64 represent the sensitivity of the first local beamformer 26 without the hearing device 6 being mounted on the head 31 of the user 21 , i.e., without any head shadowing effects or the like, but only the spatial sensitivity of the microphone array consisting of the two microphones 14,

16. The first pre-processed signal spi then can be obtained by means of a first ref erence pre-processing coefficient wir and a first auxiliary pre-processing coefficient wia for the respective first reference and auxiliary signals si _ra , sia, as s _x =

W _lr S _lr + W a ^s la.

In the left diagram, the first pre-processed signal spi has a cardioid-shaped direc tional characteristic 60 with a null direction 44 at 180° with respect to the first frontal direction 36. The cardioid-shaped directional characteristic 60 in a free field for the first pre-processed signal spi then can be obtained by setting (up to a global phase and a global gain, possibly accounting for a high-pass behavior of the cardioid) wir = 1 and wia = -e ^iwT in time-frequency domain, being T the acous tic runtime difference between the first reference microphone 14 and the first auxil iary microphone 16 (a suitable global gain and phase factor for wir and wia may be given by l/(l - _e ^~2iu>T ) for cardioid-shaped directional characteristics). In the right diagram, the first pre-processed signal spi has a figure-of-eight-shaped directional characteristic 64 with two null directions 44 at ± 90° with respect to the first frontal direction 36. The figure-of-eight-shaped directional characteristic 64 can be ob tained by setting (up to a global phase and a global gain, which in this case may be given wir = 1 , wia = -1 , as an example.

In the middle diagram, the first pre-processed signal spi has a hypercardioid- shaped directional characteristic 62 with two null directions 44 at approx. ± 110° with respect to the first frontal direction 36. The hypercardioid-shaped directional characteristic 62 can be obtained by a linear combination of the respective first ref erence and auxiliary pre-processing coefficients wir, wia for the cases of the cardi oid-shaped directional characteristic 60 and the figure-of-eight-shaped directional characteristic 64.

Note that figure 3 shows only particular cases for the first pre-processed signal spi, and the null directions 44 a priori may enclose any angle with the first frontal direction 36, even angles less than ± 90° (i.e. , a null direction may also point to wards the frontal hemisphere). Note that when the binaural hearing system 2 is mounted on the head 31 of the user 21, the second pre-processed signal sp2 which preferably has similar proper ties as the first pre-processed signal, in particular may have different null direc tions resulting from an adaptation to a different sound source due to possible head shadowing effects. Thus, for a direction-sensitive signal processing task, such as binaural beamforming or source localization, which uses both the first and the sec ond pre-processed signal spi, sp2, two signals with possibly sharp differences in their directional characteristics are to be combined.

Therefore, in order to use the first pre-processed signal spi for said binaural signal processing task in combination with the second pre-processed signal sp2, the an gular range for the null direction 44, i.e. , for the direction of maximal attenuation 70 may be restricted, as shown in figure 4, where the shaded areas indicate forbidden angles for the directions of maximal attenuation 70.

In particular, the direction of maximal attenuation 70 may be restricted to a range of [+90°, +270°] with respect to the first frontal direction 36, as it is shown in the right polar plot diagram, depicting the figure-of-eight-shaped directional character istic 64 of figure 3. As a result, the direction of maximal attenuation 70 can still point towards ± 90° (sufficient for generating the figure-of-eight-shaped directional characteristic 64), but not much more forward.

A further restriction is displayed in the middle polar plot diagram of figure 4; here, the direction of maximal attenuation 70 may be restricted to a range of [+105°, +255°] with respect to the first frontal direction 36. Note that with this restriction, a lateral interferer at +90° or -90° (=+270°) would not be fully attenuated anymore, as it is the case for the figure-of-eight-shaped directional characteristic 64. In the middle polar plot diagram, the hypercardioid-shaped directional characteristic 62 of figure 3 is depicted, which would still be achievable within the restrictions of the prohibited zone 72. A yet further restriction is displayed in the left polar plot diagram of figure 4, which depicts the cardioid-shaped directional characteristic 60 of figure 3; here, the di rection of maximal attenuation 70 may be restricted to a range of [+125°, +235°] with respect to the first frontal direction 36. In this case, only an interferer in this angular range, i.e. , in a cone of ± 55° around the backward direction 74 would be fully attenuated by the first pre-processed signal spi. in case the interferer is out side the forbidden zone 76, the direction of maximal attenuation 70 would coincide with one of the two borders of the forbidden zone 76. In the present case, as the direction of maximal attenuation 70 is a null direction 44 in the backward direction 74, this implies a situation with an interferer at 180° with respect to the first frontal direction 36.

In figure 5, a block diagram of the signal flow of a method for operating the hearing system 1 according to figure 1 in the environment 35 according to figure 2 is shown. For the direction-sensitive pre-processing 24, the first reference pre-pro cessing coefficient wir and the first auxiliary pre-processing coefficient wia are pro vided by the adaptive first local beamformer 26, and for the direction-sensitive pre processing 28, a second reference pre-processing coefficient W2r and a second auxiliary pre-processing coefficient W2a are provided by the adaptive second local beamformer 30. Said first and second reference and auxiliary pre-processing coef ficients wir, wia, W2r, W2a are adaptive, as mentioned above, and thus depend on the respective first or second reference and auxiliary signals si _ra , sia or S2r, S2a, re spectively. The first pre-processed signal spi is then generated as a linear combi nation of the first reference and auxiliary signal sir, sia, weighted by the adaptive first reference and auxiliary pre-processing coefficients wir, wia, i.e., sp _± = w _lr s _lr + w _la s _la , while the second pre-processed signal sp2 is given by sp ₂ = ^w 2r ^s 2r + ^w 2a ^s 2a, in ^an analogous way. All involved signals and coefficients are fre quency dependent (the frequency dependence has been suppressed here for the sake of simplicity). The first and the second pre-processed signal spi, sp2, are generated by the first and second local beamformers 26, 30, according to re strictions on their directions of maximal attenuation 70, as described in figure 4. Now, in order to perform the direction-sensitive signal processing task by means of the first and second pre-processed signal spi, sp2 in the first hearing device 6, said task being, e.g., a source localization or the generation of a global beamformer signal, a first head related transfer function Hi and a second head related transfer function H2 are provided in a way yet to be described. The first and second head related transfer function H _x (w,q), H ₂ (w, Q) are intrinsically frequency-dependent (hence, the variable w), and represent the propagation of a sound signal from a given angle Q towards the first and second hearing device 6, 8, respectively, taking into account head shadowing effects and the positions of the microphones of the respective hearing device 6, 8 with respect to the head 31 and the ear (in particu lar, the ipsilateral pinna) of the user 21 . Due to this information on the propagation of sound in the direct vicinity of the head 31 of the user 21 , the first and second head related transfer function H _x (w, Q), H ₂ (w, Q) will be used for the direction-sen sitive signal processing task, as well as the locally pre-processed signals spi, sp2.

The first head related transfer function Hi may be given by either of the respective frequency- and angle-dependent first reference and auxiliary head related transfer functions hir, hia, which may be provided for the first reference microphone 14 or the first auxiliary microphone 16, wherein said first reference or auxiliary head re lated transfer functions hir, hia take into account the head (and possibly pinna) shadow effects for sound that propagates from the angle Q with respect to the global direction of preference 54 towards the corresponding microphone position on or at the left side 32 of the head 31 of the user 21. In a similar way, the second head related transfer function H2 may be given by either of the respective second reference and auxiliary head related transfer functions h2r, h2a, which may be pro vided for the second reference microphone 18 or the first auxiliary microphone 20.

Now, a direction-sensitive signal processing task 80 is performed on the first pre- processed signal spi and the second pre-processed signal sp2, wherein for per forming said task 80 locally in the first device 6, the second pre-processed signal sp2 is transmitted to the first device 6 (indicated in figure 3 by the domain enclosed by the dashed line). The second head related transfer function H2 may be stored beforehand in an appropriate memory of the first hearing device 6. The task 80 may be given by any directional processing that involves the first and second pre-process signal spi, sp2, as well as the first and second head related transfer function Hi, H2. In particular, during as a result of the task 60, and/or dur ing an intermediate step (dashed feedback loop), a globally-processed signal s _gi may be generated as

Sgi(o , q ₀ ) = o ₁ (w, q ₀ , H ₁ ,// ₂ )8r ₁ (w) + c ₂ ( ,0 _o , H ₁ ,// ₂ )sp ₂ ( )

(i) wherein ci and C2 represent frequency-dependent coefficients for the generation of the globally-processed signal Sbf which, in general, both also depend on the first and second head related transfer function Hi, H2, as well as on a spatial direction qo with respect to which a specific signal processing task is performed.

Among other examples, the globally-processed signal s _gi may be given by a binau ral beamformer signal (pointing towards the direction of preference 0o) or by a so- called notch-filtered signal s _n which shows a maximal (and ideally total) attenuation towards the direction 0o. A suitable set of such notch-filtered signals s _n may be used for determining the location of a sound source, by scanning the total space with the notch-filtered signals s _n (and varying the notch angle 0o for said scan).

Generally, the globally-processed signal s _gi can be represented as a scalar prod uct of a signal vector sv = [spi, sp2] ^T containing the two pre-processed signals spi, sp2 and a coefficient vector cv = [ci, C2] ^T containing the coefficients oi(w, qo, Hi,

H2) and 02(00, 0O, HI, H2), i.e.

Sgi = cv ^H sv = c ₁ sp ₁ + c ₂ sp ₂

(ii) in the case that the task 80 is using only the two pre-processed signals spi, sp2 as the only input signals. However, the task 80 may also involve one or more further signals, e.g., the first reference signal sir and/or the first auxiliary signal sia (c.f. dotted arrow from the first auxiliary signal sia towards the signal vector sv), and/or also another locally pre-processed signal, preferably generated in an analogous way as the first and second pre-processed signal spi and sp2. In such a case of more than two signals for the task 80, the signal vector sv has three (or four or more) components, and the coefficient vector cv is to be constructed accordingly to match the dimension of sv. For the first auxiliary signal sia, e.g., a correspond ing dependence on hia (not shown) can be implemented in the coefficients ci, C2,

C3 (and possibly further coefficients). For another locally pre-processed signal sp3, a corresponding head related transfer function Fh is to be implemented into the co efficients ci , C2 and C3.

The task 80, e.g., may be given by a generation of the binaural beamformer signal Sbf pointing towards a specific direction qo. In this case, the respective signal contri bution of the first and second pre-processed signal spi, sp2 also has to be filtered with respective filter coefficients ci and C2 (as given above in equation ii) involving the corresponding first or second head related transfer function Hi, H2, in order to properly account for the head shadowing effects of sound originating from the di rection qo towards which the beamformer signal Sbf shall be directed.

However, the direction-sensitive signal processing task 60 may also be given by the localization of an a priori unknown angle 0o of a sound source (taken with re spect to a global direction of preference such as a frontal direction of the hearing system 1 ).

To this end, a set of angle-dependent spatial filters F (Q) is formed by coefficient vectors cv(0) as given above from the first and second head related transfer func tion Hi, H2. Each of said spatial filters F (Q) effectively forms a notch in the direc tion Q corresponding to the argument, and scanning the entire space surrounding the user 21 of the hearing device 1 by incrementing the angle argument Q of the filters F (Q) (e.g., by 10° or 15° or 20° in each incremental step). Then, each of the spatial filters F (Q) is applied as its respective coefficient vector cv (c.f. above) to the signal vector sv = [spi, sp2] ^T , i.e. , to the first and second pre-processed signal. The angle 0o of the sound source of interest then corresponds to the spatial filter F (0o) with the minimum signal energy of the filtered signal vector, i.e., to the spatial filter which blocks most of the signal energy out of the first and second pre-pro cessed signal spi, sp2.

Then, the spatial filters F (Q) may be derived by imposing additional constraints on the gain, e.g., in frontal direction (0°). The spatial filter F (Q) can then be described by where the gain constraint vector g and the normalized constraint coefficient matrix M may be given by with the normalized gain constraints go, go representing the gain at 0° and at the angle Q, respectively (e.g., go = 1 , go = 0), and FI21 (0°) being the quotient H2 (0°)/Fli (0°) (and likewise for Q, wherein the frequency dependence of Hi, H2 has been omitted). In case that three or more signals are used for the task 80, the gain constraint vector g is a three or more component vector, wherein for each spatial filter F (Q), the total number of constraints shall match the total number of local and/or locally pre-processed signals used for the implementation of the task 80.

The designed spatial filter F(0) is applied to the signal vector sv = [spi, sp2] ^T as the scalar product F ^H (Q) sv. In this example, the spatial filter F(0) is designed to have maximum attenuation at a source angle qo and distortionless response at the frontal source direction (0°) based on the gain constraints go and go, respectively.

The angle qo of a dominant sound source can then be determined, at least as an approximation, by the angle Q for which the corresponding spatial filter F(0) ap plied to the signal vector sv = [spi, sp2] ^T as the scalar product F ^H (Q) sv, i.e. , the globally-processed signal s _gi for each of the angles Q, minimizes the total energy. The restrictions on the direction of maximal attenuation 70 of the first and second pre-processed signal spi, sp2, as shown in figure 4, help to reduce spatial distor tion that might stem from the use of the first reference or auxiliary head related transfer function hir, hia as the first head related transfer function Hi (see corre sponding dashed arrows), and the second reference or auxiliary head related transfer function h2r, h2a as the second head related transfer function hte (see cor responding dashed arrows), in combination with the local pre-processing by the first and second local beamformers 26, 30. Note that this spatial distortion, which could compromise the final result of the task 80, cannot simply be avoided by an other global choice for the first and second head related transfer functions Hi, H2, as the inaccuracies stem from the interplay of these functions with the adaptively generated pre-processed signals spi, sp2. Thus, restricting the beamforming for the local pre-processing might compromise the local pre-processing (and may lead to a sub-optimal improvement of the sound quality, e.g., in terms of signal-to-noise ratio, SNR), but can then improve the overall performance - also in terms of possi ble SNR - due to the increased spatial accuracy.

Even though the invention has been illustrated and described in detail with help of a preferred embodiment example, the invention is not restricted by this example. Other variations can be derived by a person skilled in the art without leaving the extent of protection of this invention.

Reference numeral

1 hearing system

2 binaural hearing system

6 first hearing device

8 second hearing device

14 first reference microphone

16 first auxiliary microphone

18 second reference microphone

20 second auxiliary microphone

22 environment sound

24 direction-sensitive pre-processing 26 first local beamformer

28 direction-sensitive pre-processing 30 first local beamformer

32 left side

34 right side

36 first frontal direction

40 second frontal direction

44 first null direction

45 directional characteristic

46 interferer

47 directional characteristic

50 (dominant) sound source

52 angular source direction

54 global direction of preference

55 directional characteristic

60 cardioid-shaped directional characteristic 62 hypercardioid-shaped directional characteristic

64 figure-of-eight-shaped directional characteristic

66 origin

70 direction of maximal attenuation

72 forbidden zone 74 backward direction

76 forbidden zone

80 direction-sensitive signal processing task cv coefficient vector

F spatial filter g gain constraints hir first reference head related transfer function hia first auxiliary head related transfer function h2r second reference head related transfer function h2a second auxiliary head related transfer function

Hi first head related transfer function

H2 second head related transfer function sir first reference signal sir first auxiliary signal

S2r second reference signal

S2a second auxiliary signal spi first pre-processed signal sp2 second pre-processed signal sv signal vector wir first reference pre-processing coefficient wia first auxiliary pre-processing coefficient

W2r second reference pre-processing coefficient

W2a second auxiliary pre-processing coefficient