Method for focusing a hearing aid beam former

Abstract

The invention relates to a method for focusing a beamformer of a hearing instrument. The object of the invention is to enable automatic adaptation of the beam width and / or the beam direction, which can be used comfortably and intuitively.A fundamental concept of the invention consists of a method for focusing a beamformer of a hearing instrument comprising the following steps: - Detecting the spatial orientation and / or position of the head of the hearing instrument user, - Upon detection of a lack of head movement, direction-dependent detection of acoustic signals, - Subsequently increasing the amplification of acoustic signals originating from a focus angle in front of the head of the hearing instrument user relative to acoustic signals from other angles, thereby activating or increasing the directivity, - Subsequently gradually focusing by decreasing the focus angle, thereby increasing the directivity, until the level of acoustic signals from the focus angle, actually the presence of the desired signals in the focus angle (purely theoretically the probability that the desired signal is present in the focus angle), decreases due to the reduction of the focus angle.This advantageously starts the direction-dependent, directional detection of acoustic signals automatically as soon as the user looks in the direction of an acoustic source, for example a speaker, and then looks at the source without looking.

Description

[0001] The invention relates to a method for focusing a beamformer of a hearing instrument.

[0002] Hearing instruments can be designed, for example, as hearing aids that are worn on or in the ear. A hearing aid is used to supply a hearing-impaired person with acoustic ambient signals that are processed and amplified to compensate for or treat the specific hearing impairment. It basically consists of one or more input transducers, a signal processing device, an amplification device, and an output transducer. The input transducer is usually a sound receiver, e.g. a microphone, and / or an electromagnetic receiver, e.g. an induction coil. The output transducer is usually an electroacoustic transducer, e.g. a miniature loudspeaker, an electromechanical transducer, e.g. a bone conduction receiver, or a stimulation electrode for cochlear stimulation. It is also referred to as a receiver.The output transducer generates output signals that are routed to the patient's ear and intended to produce an auditory perception. The amplifier is usually integrated into the signal processing unit. The hearing aid is powered by a battery integrated into the hearing aid housing. The essential components of a hearing aid are usually arranged on or connected to a printed circuit board (PCB).

[0003] It is extremely difficult for hearing instrument users to understand a single speaker or to hear exclusively in a specific direction, especially in problematic acoustic environments with multiple acoustic sources (for example, the so-called cocktail party scenario). In order to improve directional, focused hearing and speech understanding, it is known to use so-called beamformers in hearing aids to emphasize the respective acoustic source, e.g. a speaker, by amplifying other sounds less than the desired acoustic signal. The use of beamformers requires the presence of a directional microphone array, which requires at least two microphones in a spatially separated arrangement. Just two microphones on a single hearing instrument are sufficient to achieve directional, i.e. spatially directed, sensitivity of the microphone array.An extension of the directional capabilities of hearing instruments can be achieved by combining the microphones of both hearing instruments in a binaural hearing system into a directional microphone array. This requires a connection, preferably wireless (e2e = ear-to-ear), between the two hearing aids.

[0004] For hearing instruments with a directional microphone array and beamformer, the problem is determining the direction in which the beamformer should be directed and finding the optimal width, or aperture angle, of the beam. In other words, the problem is to find the spatial direction in which the directional microphone array should have the highest sensitivity and the angle, or aperture angle, over which the sensitivity should be increased. It is obvious that better directionality and sensitivity can be achieved by directing the beam as precisely as possible toward the acoustic source of interest and focusing it as tightly as possible.

[0005] Interesting acoustic sources can primarily be speakers or speech signals, but there are also a number of other possibilities, such as music or cue signals.

[0006] US 2011 / 0103620 A1 discloses a method for reproducing acoustic signals using multiple loudspeakers. By appropriately filtering the individual loudspeaker signals, a desired spatial reproduction characteristic is achieved.

[0007] US 2012 / 0020503 A1 discloses a hearing aid that uses a method for acoustic source separation. Using a binaural microphone array, the spatial direction of an acoustic source is determined. A binaural receiver array then generates an acoustic output signal dependent on the determined direction.

[0008] US 2007 / 0223754 A1 discloses a hearing aid that determines the spatial direction of acoustic signals. Based on the determined spatial-acoustic information, the acoustic environment is then classified, and the transfer characteristics of the signal processing are adjusted depending on the classification.

[0009] US 2010 / 0074460 A1 discloses a hearing aid that determines the spatial direction of acoustic sources. A beamformer is then directed toward a determined direction to focus on the acoustic source in question. The spatial direction can be determined based on, among other things, the user's head orientation or line of sight.

[0010] US 2010 / 0158289 A1 discloses a hearing aid that uses a method for "blind source separation" of different acoustic sources. The user can select the various detected sources one after the other by pressing a switch.

[0011] Hearing aids from the manufacturer Siemens are known under the name SpeechFocus A method is known in which the acoustic environment is automatically scanned for speech components. Once speech components are identified, their spatial direction is determined. The amplification of acoustic signals from this direction is then increased compared to signals from other directions.

[0012] Using known methods and devices, the simplest way to shape the beam is to assume that the desired source or speaker is located directly in front of the hearing instrument user and that the beam should therefore be directed forward, with the beam direction being changed by head movements of the user. Alternatively, the hearing instrument can direct the beam in a desired direction using an algorithm for processing the microphone signals, regardless of the orientation of the head. The beam direction can be controlled, for example, by a remote control. The disadvantage, however, is that the user cannot hear or can hardly hear sources outside the beam and thus cannot register them. Furthermore, having to control the beam via remote control is uncomfortable and unintuitive for the user.

[0013] Alternatively, the hearing instrument can automatically analyze the direction of potentially interesting acoustic sources and automatically align the beam in this direction, as in the procedure SpeechFocus from the manufacturer Siemens. However, this can be confusing for the user, as the hearing instrument can automatically and potentially unexpectedly jump between different sources without any control on the part of the user. Furthermore, a continuously adapting beamformer changes the binaural "cues," making it difficult or even impossible for the user to locate the source of interest.

[0014] In contrast to the beam direction, the beam width is usually constant or can be manually adjusted by the user between different preset opening angles.

[0015] Publication DE 10 2010 026 381 A1 describes a method in which an audio signal is classified before localizing an audio source. The classification can be based on features such as harmonic signal components or the characteristics of formants. The subsequent localization benefits from the previous classification.

[0016] The object of the invention is to enable an automatic adaptation of the beam width and / or the beam direction, which can be used comfortably and intuitively, which avoids unexpected focusing of the beam without intervention by the hearing instrument user, and which makes it possible to make acoustic sources outside the beam known to the user in a simple and easy-to-use manner.

[0017] This object is achieved according to the invention by a method having the features of claim 1.

[0018] Directivity is a property of the beamformer that can be represented as a metric, the higher the directivity the beamformer is focused, i.e., the smaller the solid angle of the beam. By increasing the directivity of a beamformer, for example, by increasing a beamformer parameter corresponding to the metric, signals within the beam are amplified more strongly than signals outside. The described method controls the aforementioned beamformer parameter.

[0019] This advantageously starts the direction-dependent, directional recording of acoustic signals automatically as soon as the user looks toward an acoustic source, such as a speaker, stops moving their head, and then focuses on the source, i.e., looks at it without looking away. For the detection of head movements, suitable tolerance values ​​or thresholds, such as at least 15° rotation, must be specified in order to distinguish unintentional or irrelevant minimal head movements from relevant head movements. Manual triggering of focusing, for example, by pressing a button on the hearing instrument or using a remote control, is not required, which significantly contributes to the practicality and comfort of the method.

[0020] By controlling or stopping the focusing based on an identified acoustic source, the probability is increased that the method actually focuses on a source of interest to the user, and not on a randomly set focus spatial angle depending on the source.

[0021] An advantageous design consists of the following steps: Identifying an acoustic source in the focus solid angle based on the acoustic signals from the focus solid angle, for example by using a frequency or frequency spectrum criterion, a 4Hz speech modulation detector, a Bayes detector, or a hidden Markov model detector, Determining the spatial direction in which the acoustic source is located, Centering the focus solid angle in this direction.

[0022] By aligning the focus solid angle, the focus is better aligned with the source of interest to the user. This subsequently enables sharper focusing due to a narrower focus solid angle, thus increasing directionality. Increasing directionality, in turn, results in a further enhancement of the source signal of interest.

[0023] A further advantageous design consists of the following steps: Subsequently, detect additional acoustic signals coming from spatial angles other than the focal spatial angle. Detect additional acoustic sources based on the additional acoustic signals, for example, using a frequency or frequency spectrum criterion, a 4 Hz speech modulation detector, a Bayesian detector, or a hidden Markov model detector. Upon detection of another acoustic source, increase the gain of the additional acoustic signals.Detecting the spatial orientation and / or position of the hearing instrument user's head after increasing the amplification of the additional acoustic signals. Upon detecting the absence of head movements within a predetermined period of time after increasing the amplification of the additional acoustic signals, lowering the amplification again. Upon detecting head movement within the predetermined period of time, defocusing by again enlarging the focus solid angle and then carrying out the method according to one of the preceding claims.

[0024] This means that while the process is in the focused stage on a single source, meaning that only the signals from that source are highlighted for the user's perception, the wider space around the user is scanned for additional sources. If such an additional source is found, it is made perceptible to the user by increasing the gain; the user is, as it were, alerted to the presence of the additional source. If the user reacts by moving or turning their head, the previous focus is automatically canceled and refocusing takes place. This also advantageously starts refocusing automatically and does not need to be initiated manually, which further contributes to the practicality and comfort of the process.

[0025] A further advantageous design consists of the following steps: If no further acoustic sources are detected, detect the spatial orientation and / or position of the hearing instrument user's head. If head movement is detected, defocus by increasing the focus spatial angle again or by switching from direction-dependent to direction-independent detection of acoustic signals.

[0026] This means that focusing is automatically stopped as soon as the user looks away from the currently focused source, which further contributes to the practicality and comfort of using the method.

[0027] A further advantageous embodiment is that the method is only performed if a head movement was detected before the absence of head movement was detected. This prevents, for example, automatic focusing from occurring even though the user has not turned toward an acoustic source, for example because the source is non-acoustic or because the user does not want to devote their increased attention to any one source.

[0028] A further advantageous embodiment is that the method is only performed if an acoustic source was detected within the focus solid angle before focusing. This prevents focusing despite the absence of acoustic sources, which would obviously be impractical.

[0029] Further advantages and developments emerge from the dependent patent claims and the following description of exemplary embodiments and figures. They show: Fig 1User with left and right hearing instrument Fig 2Hearing instrument including essential components Fig 3Signal processing components of the adaptive beamformer Fig 4User and multiple acoustic sources Fig 5Focused beam Fig 6Acoustic source outside the beam Fig 7Changing the beam direction Fig 8Refocused beam Fig 9Flow diagram, focusing and D-focusing

[0030] In Figure 1is a schematic view of a user 1 with left hearing instrument 2 and right hearing instrument 3 in plan view. The microphones of the left and right hearing instruments 2, 3 are each combined to form a directional microphone arrangement, so that it is possible to direct the respective beam essentially either forwards or backwards as seen from user 1. Furthermore, it is possible to connect the left and right hearing instruments 2, 3 with a wireless link (e2e) to enable a binaural configuration with a binaural microphone arrangement. This essentially enables directions to the right and left as seen from user 1 as additional beam directions of the arrangement. The automatic focusing of the beam can be carried out both for each monaural hearing instrument individually (front / back) and for the binaural arrangement (right / left) together.

[0031] In Figure 2The left and right hearing instruments 2, 3 are shown schematically, including the essential signal processing components. The hearing instruments 2, 3 are constructed identically and may differ in their external shape to accommodate their respective use on the left or right ear. The left hearing instrument 2 comprises two microphones 4, 5, which are arranged spatially separately and together form a directional microphone arrangement. The signals from the microphones 4, 5 are processed by a signal processing device 11, which emits an output signal via the receiver 8. A battery 10 supplies power to the hearing instrument 2. In addition, a motion sensor 9 is provided, the function of which in the automatic focusing process will be explained further below. The right hearing instrument 3 comprises the microphones 6, 7, which are also combined to form a directional microphone arrangement.For further components, please refer to the previous description.

[0032] In Figure 3 The essential signal processing components of the automatically focusing beamformer are shown schematically. The signals from microphones 4 and 5 of the left hearing instrument 2 are processed by the beamformer to create a beam directed straight forward (0°, "broadside") as seen from the user, which has a variable beam width. The variable beam width is synonymous with variable directionality (a smaller beam width means greater directionality and vice versa, with greater directionality being synonymous with greater directional dependence). The beamformer is constructed in a conventional manner, for example, as an array of fixed beamformers, as a mixture of a fixed beamformer with a direction-independent omnidirectional signal, as a beamformer with variable beam width, etc.

[0033] Output signals of the beamformer 13 are the desired beam signal, which contains all acoustic signals from the direction of the beam, the direction-independent omni signal (which contains all acoustic sources in all directions with under-distorted binaural cues) and the anti-signal, which contains all acoustic signals from directions outside the beam.

[0034] The three signals are fed to the mixer 19 and, in parallel, to the source detectors 15, 16, 17. The source detectors 15, 16, 17 continuously determine the probability (or a comparable measure) that an acoustic source of interest, for example a speech source, is present in the three signals.

[0035] The motion sensor 9 is designed to detect head movements of the hearing instrument user, including rotation, and also to determine the extent of each movement. A dedicated hardware sensor of the conventional type is the fastest and most reliable way to detect head movements. However, other options for detecting head movements are also available, for example, based on a spatial analysis of the acoustic signals or using additional alternative sensor systems. A head movement detector 14 analyzes the signals from the motion sensor 9 and uses them to determine the direction and extent of head movements.

[0036] All signals are fed to the focus control 18, which determines the beam width based on the signals. The determined beam width is then fed by the focus control 18 to the beamformer 13 as an input signal. In addition to the beam width, the focus control also controls the mixer 19, which mixes the three previously explained signals (omni, anti, beam) and forwards them to a hearing instrument signal processor 20. In the hearing instrument signal processor 20, the acoustic signals are further processed in the manner customary for hearing instruments and amplified, then output to the receiver 8. The receiver 8 generates the acoustic output signal for the hearing instrument user.

[0037] The focus control 18 is preferably designed as a finite state machine (FSM), the finite states of which will be explained further below.

[0038] The three signals (Omni, Anti, Beam) are mixed by the mixer 19 so that the user receives a natural-sounding spatial signal. This also includes ensuring smooth transitions rather than abrupt ones. Further processing steps take place in the hearing instrument signal processing 20, which primarily serve to compensate for or treat the user's hearing impairment.

[0039] In Figure 4An exemplary situation is shown schematically. It shows hearing instrument user 1 with left and right hearing instruments 2, 3 in a top view. An acoustic source 21 is located directly in front of user 1, toward which user 1 is looking. The beam of the respective hearing instrument 2, 3 is focused on acoustic source 21 by reducing the beam width to an angle α 1. Thus, the additional acoustic source 22 lies outside the beam, but would be within a beam with a beam width of α 2. The additional acoustic source 23 lies even further outside the beam and is located almost next to user 1.

[0040] In Figure 5 to 8 The functionality of the automatic focusing of the beam is explained schematically. In Figure 5 the beam is focused with the width β on the acoustic source 21. In Figure 6The user moves their head away from source 21 and toward source 23. The head movement is detected by the automatic focus control (or by the motion sensor). The automatic focus control then defocuses the beam by switching to the Omni signal. Alternatively, defocusing can also be achieved by setting the beam width to a predefined, significantly larger aperture angle than the focused state.

[0041] In Figure 7User 1 has turned their head completely towards the acoustic source 23. The head movement ends and user 1 looks towards the source 23. The end of the head movement is detected, whereupon the automatic focusing of the beam on the source 23 begins. If necessary, the signal is switched from the direction-independent omnidirectional signal to the direction-dependent beam signal and / or the greatly increased beam width is gradually reduced. The beam width is reduced until the signal source 23 is fully focused. Further reduction of the beam width results in the source no longer being completely within the beam, so that the signal from source 23 or its portion in the beam signal decreases. The focusing of the beam, i.e. the reduction of the beam's aperture angle, is ended as soon as the source 23 is sharply focused, which is the case with the Figure 8the angle β shown is the case. Any reduction of the beam angle that may exceed this value will be reversed.

[0042] In Figure 9 The finite states of the Finite State Machine (FSM) are explained. The FSM starts in the "Omni" state 40 (no directionality, the mixer outputs the Omni signal), in which the hearing instrument user hears normally and independently of direction. In this state, they are able to localize acoustic sources normally. They can move and turn their head in a normal and natural manner, for example, to search for an acoustic source of interest, such as a speaker.

[0043] As soon as the user turns their attention to a source and focuses on it, they turn their head toward that source and then stop moving their head. The loop 41 is exited. Instead, the FSM enters the "Focusing" state 42, and the directionality of the beamformer is gradually increased (the beam width is reduced, and a correspondingly more direction-dependent signal is output to the user). This increases the proportion of the source signal in the beam signal, and the mixer passes on the filtered signal by outputting exclusively or primarily the beam signal.

[0044] Once the maximum directionality (minimum beam width) is reached, which corresponds to the previous Figure 5 and Figure 8If the beam width corresponds to the state described above, the proportion of the source signal of interest in the beam signal cannot be increased any further. The directionality is not further changed (beam width is not further reduced), and the FSM leaves loop 43 and enters the "Focused" state 44. In the "Focused" state 44, the automatic beam control continuously monitors the user's head movements using the motion sensor (loop 47). As long as no head movements are detected, the FSM remains in the "Focused" state 44.

[0045] Furthermore, the system continuously monitors whether any acoustic sources of interest are present outside the beam in the Omni and Anti signals. If a new source is detected, the FSM switches to the "Glimpsing" state 45. In the "Glimpsing" state 45, a small portion of the Omni signal, which contains the potential additional source, is mixed into the output signal for the user by the mixer. This alerts the user that another source is present. If the user does not turn toward this new source, they do not move their head. The automatic focus control detects this with the help of the motion sensor and, after a certain period of time, adjusts the portion of the Omni signal back to zero (fade out), allowing the user to fully concentrate on the focused signal again.The described "glimpsing" is performed every time a new source appears in the acoustic environment or when the acoustic environment changes significantly.

[0046] However, if the user moves his head because he wants to focus on a new signal or simply wants to survey the acoustic environment, which is described in the previous Figure 6 As shown, the head movement is detected and the focus control immediately switches to the omni signal, meaning the beam width is greatly increased again and / or the mixer outputs the omni signal additionally or exclusively. This is represented in the illustration by element 46.

[0047] The Omni signal allows the user to survey the acoustic environment with all undistorted spatial cues that are distorted or missing in the beam signal. This allows the user to localize acoustic sources normally. As soon as the user focuses on a different acoustic source, which is similar to the previously explained Figure 7 , the FSM returns to the focusing state 42. Beam focusing then begins again.

[0048] It goes without saying that, to ensure a pleasant acoustic experience for the user, all states of both the beam focusing and the mixer are changed smoothly and without sudden steps.

[0049] The method explained above enables a function that closely mimics the human ability to focus on different sources by combining the various beamformer signals with the head movement detector. Head movement is used to provide natural feedback for automatic focusing and rapid defocusing on a target to control the beamformer. Focusing occurs gradually when the user is not moving their head. Defocusing upon head movement, or the transition from the beam signal to the omni-signal, occurs quickly to quickly provide an undistorted signal with all spatial information in the event of changes. The glimpsing function allows the user to remain focused on one source while maintaining an overview of new sources and changes.

[0050] A basic idea of ​​the invention can be summarized as follows: The invention relates to a method for focusing a beamformer of a hearing instrument. The object of the invention is to enable automatic adaptation of the beam width and / or beam direction that can be used conveniently and intuitively. A basic idea of ​​the invention is a method for focusing a beamformer of a hearing instrument, comprising the steps: Detecting the spatial orientation and / or position of the hearing instrument user's head. If the absence of head movements is detected, direction-dependent detection of acoustic signals. Then increasing the gain of acoustic signals coming from a focal solid angle in front of the hearing instrument user's head compared to acoustic signals from other spatial angles, and thereby activating or increasing the directivity. Then gradually focusing by reducing the focal solid angle, and thereby increasing the directivity, until the level of acoustic signals from the focal solid angle, actually the presence of the desired signals in the focal solid angle (purely theoretically the probability that the desired signal is present in the focal solid angle), decreases due to the reduction in the focal solid angle.

[0051] This advantageously starts the direction-dependent, directional detection of acoustic signals automatically as soon as the user looks in the direction of an acoustic source, for example a speaker, and then looks at the source without looking away.

Claims

1. Method for focusing a beam former (13) of a hearing aid (2, 3), comprising the following steps: - detecting the spatial orientation and / or position of the head of the hearing aid user (1), - detecting movements of the head of the hearing aid user (1) by a movement sensor (9) or on the basis of a spatial analysis of acoustic signals, - detecting acoustic signals in a directionally dependent manner when the hearing aid user (1) has moved the head in the direction of a source of an acoustic signal (21) and does not move the head any further, - thereafter increasing the gain of acoustic signals which come from a focus solid angle (α1, α2, β) frontally in front of the head of the hearing aid user (1) in the direction in which the hearing aid user (1) looks, in relation to acoustic signals from other solid angles, - thereafter repeatedly carrying out focusing by reducing the focus solid angle (α2) until the level of acoustic signals from the focus solid angle (α2) reduces due to a reduction in the focus solid angle (α2) and a minimum focus solid angle (α1, β) is obtained.

2. Method according to Claim 1, comprising the following further step: - identifying the acoustic source (21) in the focus solid angle (α2) on the basis of the acoustic signals from the focus solid angle (α2).

3. Method according to Claim 2, comprising the following further step: - focusing until the level of acoustic signals from the acoustic source (21) in the focus solid angle (α2) reduces due to the reduction in the focus solid angle (α2) .

4. Method according to Claim 2 or 3, comprising the following further steps: - establishing the spatial direction in which the acoustic source (21) is situated, - centring the focus solid angle (α2) in this direction.

5. Method according to one of the preceding claims, comprising the following further steps: - subsequently detecting further acoustic signals which come from solid angles (γ) other than the focus solid angle (α2), - detecting further acoustic sources (23) on the basis of the further acoustic signals.

6. Method according to Claim 5, comprising the following further steps: - increasing the gain of the further acoustic signals when a further acoustic source (23) is detected, - detecting the spatial orientation and / or position of the head of the hearing aid user (1) after increasing the gain of the further acoustic signals, - reducing the gain again if a lack of head movement within a predetermined time interval after increasing the gain of the further acoustic signals is detected, - carrying out defocusing by re-enlarging the focus solid angle (α2) when detecting a head movement within the predetermined time interval and subsequently carrying out the method according to one of the preceding claims.

7. Method according to Claim 5, comprising the following further steps: - detecting the spatial orientation and / or position of the head of the hearing aid user (1) in the absence of the detection of further acoustic sources (23), - carrying out defocusing by re-enlarging the focus solid angle (α2) or by changing from directionally dependent to directionally independent detection of acoustic signals if a head movement is detected.

8. Method according to one of the preceding claims, wherein the method is only carried out if a head movement was detected prior to detecting the absence of head movements.

9. Method according to one of the preceding claims, wherein the method is only carried out if an acoustic source (21) was detected in the focus solid angle (α2) prior to focusing.

Images