Patent Application
20160381469
This application claims the priority, under 35 U.S.C. §119, of German application DE 10 2015 211 747.2, filed Jun. 24, 2015; the prior application is herewith incorporated by reference in its entirety.
The invention relates to a method for signal processing in a binaural hearing device having a first hearing aid and a second hearing aid. The first hearing aid has a first microphone for producing a first signal and a first sound generator, and the second hearing aid has a second microphone for producing a second signal and a second sound generator.
A binaural hearing device usually contains one hearing aid per ear of the user, each of the two hearing aids containing a microphone and an electroacoustic transducer for sound generation. The signals picked up by the two microphones can then be used to form a directional characteristic on the basis of their spacing, it being possible for the directional characteristics in the respective output signal that is output to the electroacoustic transducer to be different for both hearing aids so as to attain better spatial perception. In this case, the term microphone used here and below is intended to be understood to mean both an individual microphone, which is then particularly and essentially an omnidirectional microphone, that is to say a single microphone, and a microphone that contains a plurality of single microphones, particularly two single microphones, and that as such already has a directional characteristic through the interconnection of the single microphones, that is to say is a directional microphone. In other words, in the latter case, a hearing aid then has a monaural directional microphone system.
In a conversation situation with multiple interlocutors that surround the user of a hearing device in various directions, it is possible, given a significant noise level in the surroundings, for adaptation of the binaural directional characteristic to contribute to isolating the voice signals from individual interlocutors and hence improve comprehension of the voice signal that comes from a particular target interlocutor for the user of the binaural hearing device. Interfering noise in proximity to the target interlocutor can be reduced in this case in the same way as conversation contributions from other interlocutors if they are identified as not being the relevant useful signal at that moment. This is achieved by virtue of the directional characteristic being oriented in a comparatively narrow angle range, for example with a total angular aperture of 90 degrees or even just 45 degrees, in the frontal direction of the user.
Usually, adaptation of the binaural directional characteristic requires precise knowledge of the position of the target sound source or target interlocutor, or, synonymously, of the exact direction from which the useful signal comes. For operation of hearing devices, it is usually assumed in this case that the position of the target sound source is in the frontal direction, which is to say approximately 0 degrees relative to the line of vision of the user. A binaural directional characteristic with an orientation of 0 degrees and an angular aperture of 90 degrees or even just 45 degrees then suppresses a large proportion of the background noise, which first of all improves the signal-to-noise ratio for the useful signal coming from the target interlocutor. Furthermore, voice signals from other interlocutors outside the angle range in which the directional characteristic has a high sound sensitivity are also suppressed as appropriate. Such a hearing situation, which is generally referred to as a “cocktail party situation”, can arise for a user of a hearing device whenever a conversation is to be held in the presence of other people who are themselves holding conversations with one another.
In a normal conversation situation, the interlocutor makes small movements that may be in an angle range of up to 10 or even up to 20 degrees. Similarly, the user of the binaural hearing device may also make natural head movements by virtue of his gestures in a conversation, which head movements can lead to similar angular deviations in his frontal direction relative to the interlocutor. These movements can noticeably attenuate the voice signal from the interlocutor in the conversation on account of the narrow angular aperture of the directional characteristic. Since particularly head movements by the user of the binaural hearing device during a conversation situation are often fast and barely perceptible to the user himself, and this means that fluctuations in the signal level from the binaural hearing device are correspondingly brief, the auditory perception by the user can be significantly impaired, and, particularly on account of the relatively high concentration that is required as a result of the fluctuations in the sound level, may even be perceived as disagreeable.
Published, non-prosecuted German patent application DE 10 2013 207 149 A1 discloses a hearing aid system having at least two hearing aids and also a method for operation of said hearing aid system. The hearing aid system contains particularly a signal processing apparatus for processing audio signals and a signal connection for transmitting a first audio signal from each hearing aid to the signal processing apparatus. The signal processing apparatus rates a signal component from the preferential direction in relation to the head of the wearer in the first audio signals, and the signal processing device uses the first audio signals to produce a first binaural directional microphone signal and sets the directional characteristic thereof on the basis of the rating. In this situation, other useful signals, e.g. vocal contributions in a discussion, are prevented from being suppressed.
Published, European patent application EP 2 928 210 A1 proposes a binaural hearing system, containing a left and a right hearing aid, with noise suppression and with a user interface. The left and right hearing aids contain at least two input units that, in a number of frequency bands and time entities, provide a representation of an input signal in the time domain, and a noise suppression system having a multichannel “beamformer” that is connected to the input units and produces a directional signal. The wearer of the hearing system can use the user interface to prescribe directivity for a target signal.
The invention is based on the object of specifying for a binaural hearing device a method that, in a hearing situation with a main sound source against a noisy background, prevents fluctuations in the signal level on the basis of relative movements by the main sound source in relation to a user of the binaural hearing device as quickly and efficiently as possible.
The invention achieves the cited object by method for signal processing in a binaural hearing device having a first hearing aid and a second hearing aid. The first hearing aid has a first microphone for producing a first signal and a first sound generator, and the second hearing aid has a second microphone for producing a second signal and a second sound generator. The first signal and the second signal are taken as a basis for at least approximately ascertaining a direction of a main sound source. A deviation in the direction of the main sound source from a frontal direction of the binaural hearing device prompts the hearing aid that is closer to the main sound source in each case to be defined as the local hearing aid and the hearing aid that is more remote from the main sound source to be defined as the remote hearing aid.
In this case, in the local hearing aid, at least in one frequency band, the first signal is filtered using at least one first angle-dependent filter, that is to say by applying an angle-dependent filter factor, and this produces a filtered first signal. The first signal and/or the second signal and/or the direction of the main sound source is/are taken as a basis for determining an adaptation coefficient. An adapted first signal is produced from the first signal and the filtered first signal on the basis of the adaptation coefficient, and a local directional characteristic of a reproduction signal that is to be output by the sound generator of the local hearing aid is determined from the adapted first signal and the second signal. Refinements that are advantageous and in some cases inventive in themselves are the subject matter of the subclaims and of the description below.
In this context, at least approximate ascertainment of the direction of the main sound source is intended to be understood to mean particularly that a plurality of angle ranges are determined or prescribed and that the angle ranges with which the direction of the main sound source can be associated are ascertained. In particular, this can involve one of the angle ranges being associated with the frontal direction of the binaural hearing device, and hence containing a zero angle on the angle scale on which the division into the angle ranges is based. In this case, a deviation in the direction of the main sound source from the frontal direction of the binaural hearing device can occur by virtue of the direction of the main sound source being associated with a different angle range than the one corresponding to the frontal direction. In this case, the frontal direction of the binaural hearing device can be defined preferably by the plane of symmetry of the two hearing aids during operation when worn properly by the user.
In particular, it is also possible for a frontal angle range and a right and left deviation range to be defined in this case, so that the approximate determination of the direction of the main sound source in this case means the association of the main sound source either with the frontal angle range or with one of the two deviation ranges. Filtering of the first signal using at least one angle-dependent first filter factor particularly includes multiplication of the first signal by an angle-dependent first filter factor in a relevant frequency band. However, the first signal can likewise also be convoluted with a vector-value filter factor in the time domain or in a frequency domain, in order to produce the first filtered signal.
To determine the adaptation coefficient in a frequency band, the first signal needs to be associated either with the local hearing aid or with the remote hearing aid, and the second signal needs to be associated with the hearing aid that then remains. The adaptation coefficient in a local hearing aid specifies whether and, if need be, to what extent in the local hearing aid a first signal needs to be adapted to suit the second signal by producing the first adapted signal from the first signal and the first filtered signal. By way of example, this involves the first adapted signal being formed by mixing the first signal and the first filtered signal, e.g. in the form of a weighted sum. In particular, the first adapted signal will be formed by a convex sum of the first signal and the first filtered signal with the adaptation coefficient as convexity parameter. In this case, the adaptation coefficient lies between 0 and 1, the first filtered signal being forwarded directly for a value of 1, corresponding to complete adaptation of the first signal, whereas no adaptation takes place for a value of 0.
In this case, the angle-dependent first filter factor that is used to filter the first signal in order to produce the first filtered signal takes account of the fact that, in the case of a deviation in the direction of the main sound source from the frontal direction, a minimum time shift caused by the propagation time difference firstly arises between the first signal and the second signal. This shift can initially be approximated as a phase shift. In addition, secondly, shadowing effects arise in the cited case of the angular deviation as result of the head of the user of the binaural hearing device, as result of which the remote hearing aid has a weaker signal level. The filtering of the first signal using the first filter factor compensates for the phase difference and the volume differences between the first signal and the second signal for the direction of the main sound source, as result of which the main sound source can initially be considered to be a frontal source for the further signal processing in the local hearing aid on account of the aligned level and the aligned phase in the first signal and in the second signal.
In a hearing situation for the user of the binaural hearing device in which a voice signal from an interlocutor needs to be reproduced as clearly as possible while background noise, such as other voice signals, need to be suppressed as far as possible, the binaural directional characteristic is in most cases oriented very narrowly to the front in order to reduce the undesirable sound signals as result of the lower sound sensitivity in those directions that do not correspond to the voice signal from the interlocutor, which is considered to be the useful signal. Since the binaural directional characteristic, that is to say the directional characteristic formed on the basis of signal components from the two microphones, is now formed in the reproduction signal from one of the sound generators at a time, on the basis of the first adapted signal and the second signal, and the directional deviation of the main sound source from the frontal direction has effectively been corrected in the two cited signals, a voice signal from an interlocutor is no longer attenuated by the narrow binaural directional characteristic given relative movements between the interlocutor and the user of the binaural hearing device.
In the case of such relative movements as can arise as result of brief head movements by the user of the binaural hearing device or as a result of spontaneous gestures by the interlocutor, the interlocutor is, in simple terms, always “returned to the frontal direction” by virtue of the adaptation of the first signal to suit the second signal. A particular advantage in this case is that the adaptation of the first signal to suit the second signal, which corrects the effects of the relative movements between the user and an interlocutor, can be effected regardless of the formation of the binaural directional characteristic, and hence does not influence the algorithm of the latter.
If need be, after the formation of the binaural directional characteristic in the local hearing device from the first adapted signal and the second signal, the reproduction signal can still be appropriately reworked in order to achieve as realistic as possible a spatial perception of the signal from the main sound source.
Preferably, the approximate direction of the main sound source and/or the adaptation coefficient is/are ascertained in at least one frequency band on the basis of the signal levels of the first signal and the second signal. In particular, this can be affected on a frequency band by frequency band basis, with only a subset of frequency bands being used for the final determination of the direction of the main sound source. Direction determination on the basis of the differences in the level of the first signal and of the second signal can be implemented particularly simply and quickly. In particular, it is also possible for the signal level of the sum of the first signal and the second signal to be used in order to be able to take account of any phase cancellation effects too.
Alternatively, in at least one frequency band, the value of an angle-dependent second filter factor can be prescribed at least for a left deviation angle, a zero angle and a right deviation angle. A signal can be selected from the first signal and the second signal and needs to be multiplied by the second filter factor for the purpose of orientation, which produces a respective oriented signal. An angle-dependent interference power can be ascertained from the difference between the oriented signal and the other signal, and an angle-dependent total power can be ascertained from the sum of the oriented signal and the other signal. A normalized angle-dependent interference power can be ascertained from the angle-dependent interference power and the angle-dependent total power, and a comparison of the normalized angle-dependent interference powers can be taken as a basis for ascertaining the approximate direction of the main sound source and/or an adaptation coefficient at least for the left deviation angle, the zero angle and the right deviation angle.
In this case, prescribing the second filter factor for a left deviation angle, a zero angle and a right deviation angle corresponds to splitting the space into three angle ranges, the direction of the main sound source being determined approximately by classification into one of the three angle ranges. This classification is now performed as described below.
Since the selected signal is multiplied by the second filter factor—at least for the left deviation angle, the zero angle and the right deviation angle in each case—it is oriented in line with the other signal. In this case, this orientation is effected such that, for the angle that is used as a parameter in the angle-dependent second filter factor, the two signals do not have any significant phase and volume differences. If the difference is now formed from the oriented signal and the other signal, then any sound signal from that direction that is used as an angle parameter in the second filter factor is approximately cancelled out. For this angle, the angle-dependent interference power is close to zero.
So as now to be able to compare the angle-dependent interference power for the three cited angle ranges with one another, the interference power still needs to be normalized beforehand using the total power that is formed from the summed signal containing the adapted signal and the other signal. In this case the angle-dependent interference power and the angle-dependent total power preferably need to be computed as the absolute value or square of the absolute value of the difference or the sum of the oriented signal and the other signal.
The direction of a main sound source can then be identified as that direction in which the normalized angle-dependent interference power is at a minimum. A priori, this can also be performed for more than three angle ranges, but for most applications it is sufficient to distinguish between the frontal region and two deviation regions distributed symmetrically to the left and right.
A further advantage in this case is found to be when, in a plurality of frequency bands, the comparison of the normalized angle-dependent interference powers is taken as a basis in each case for determining a frequency-band-dependent direction parameter at least for the left deviation angle, the zero angle and the right deviation angle, wherein the approximate direction of the main sound source is ascertained from the frequency-band-dependent direction parameters. By way of example, this can be affected by a possibly weighted mean value of the direction parameters. Preferably, determination of the direction of the main sound source is performed only in those frequency bands in which a significant directional dependency of the sound signal from the main sound source can be expected. By using direction parameters from multiple frequency bands, it is possible to compensate for or minimize errors in the direction determination, which can arise on account of fluctuations, for example.
Preferably, the first filter factor for producing the first filtered signal and the second filter factor have the same frequency dependency for the respective frequency band and have the same angle dependency at least for the left deviation angle, the zero angle and the right deviation angle. This means that adapting the first signal to suit the second signal, that is to say angle-dependent orientation, involves the use of filter factors having the same functional dependency as for ascertaining the direction of the main sound source.
In an advantageous refinement of the invention, in the local hearing aid, a first filtered signal is produced in each of a plurality of frequency bands and the first filtered signal is taken as a basis for producing a first adapted signal, wherein, as the frequency of the frequency band for producing the first filtered signal for the first filter factor rises, the angle increases monotonously in each case in the direction of the main sound source. What is meant by a monotonous increase is particularly that the angle that is used for the first filter factor can be chosen to be constant for multiple successive frequency bands, with a smaller angle for the first filter factor being chosen for a group of low frequency bands and a larger angle for the first filter factor being chosen for a group of higher frequency bands. This takes account of the circumstance that the directional dependency of signal components in lower frequency bands is usually lower, which means that the need for angle-dependent adaptation has less of an effect in this case than in higher frequency bands, in which standard useful signal components has a much higher directional dependency that should preferably be taken into account.
Expediently, the absolute value of the first filter factor is ≦1 in each case. This is favorable particularly for the case in which the first signal, which can be adapted to suit the second signal by the first filter factor, is formed by the signal from the local hearing aid. The signal that is recorded by the microphone of the remote hearing aid has a lower volume level on account of shading effects due to the head of the user. This is represented by a first filter factor, the absolute value of which is chosen to be ≦1. The first filtered signal undergoes no raising of the level as result of the filtering process. The selection concerning whether the signal from the microphone of the local hearing aid or the signal from the microphone of the remote hearing aid is to be chosen as the first signal that needs to be adapted to suit the second signal can be made particularly on the basis of the current hearing situation in this case. Specifically in noisy surroundings, the “local signal” preferably needs to be adapted to suit the “remote signal”, since, in this case, the adaptation process does not raise the level of the signal still further, which means that oversaturation can be avoided.
A further advantage is found to be when, in a reproduction signal for the local hearing device, the first angle-dependent filter factor and/or the adaptation coefficient is/are taken as a basis for performing compensation for a volume and a phase difference when the first signal is produced by the microphone of the local hearing aid. In this case of adaptation, the “local signal” is effectively provided with the phase and volume reference of the remote hearing aid in the local hearing aid. For the reproduction signal from the local hearing aid, this referencing to the remote hearing aid needs to be compensated for in order to be able to obtain a spatial perception that is as realistic as possible for the user of the hearing aid.
The invention further cites a binaural hearing device having a first hearing aid and a second hearing aid. The first hearing aid has a first microphone for producing a first signal and a first sound generator, and the second hearing aid has a second microphone of a second signal and a second sound generator. At least one signal processing unit is provided that is set up to perform the method described above. The advantages cited for the method and developments thereof can be applied to the binaural hearing device mutatis mutandis in this case.
Other features which are considered as characteristic for the invention are set forth in the appended claims.
Although the invention is illustrated and described herein as embodied in a method for signal processing in a binaural hearing device, it is nevertheless not intended to be limited to the details shown, since various modifications and structural changes may be made therein without departing from the spirit of the invention and within the scope and range of equivalents of the claims.
The construction and method of operation of the invention, however, together with additional objects and advantages thereof will be best understood from the following description of specific embodiments when read in connection with the accompanying drawings.
Corresponding parts and parameters are provided with the same reference symbols throughout the figures.
The influence of the movements of a user of the binaural hearing device in a conversation, or of the movements of an interlocutor, on the signal level of the voice signal from the interlocutor is shown in
In the direction identifier 30, the direction of the interlocutor 4 is taken as a basis for determining an adaptation coefficient 44, and then a first adapted signal 46 is produced from the first signal 26, the first filtered signal 42 and the adaptation coefficient 44. In this case, the first adapted signal 46 is formed as a weighted overlay containing the first filtered signal 42 and the first signal 26, for example, the adaptation coefficient 44 being used for the weighting. The first adapted signal 46 and the second signal 28 are then used to form the directional characteristic 48 that the reproduction signal 50 that is to be output via the sound generator of local hearing aid 36 needs to have in the relevant frequency band.
To this end, the first signal 26 is first of all multiplied by the respective second filter parameter 40a, 40b, 40c for each of the cited angles 56a, 56b, 56c, as result of which an angle-dependent oriented signal 60a, 60b, 60c is formed in each case. For each of the cited angles 56a, 56b, 56c, an angle-dependent interference power 62a, 62b, 62c and an angle-dependent total power 64a, 64b, 64c are now formed from the oriented signal 60a, 60b, 60c with the second signal by difference and sum formation. The angle-dependent interference power 62a, 62b, 62c is normalized in each case by dividing the associated angle-dependent total power 64a, 64b, 64c thereby, as result of which a normalized angle-dependent interference power 66a, 66b, 66c is obtained for each of the angles 56a, 56b, 56c from the first signal 26 and the second signal 28.
The normalized angle-dependent interference powers 66a, 66b, 66c are then compared with one another. The orientation of the first signal 26 to the second signal 28 is such that a sound that comes each time from the direction of the angle 56a, 56b, 56c used for orientation does not lead to a significant angle-dependent interference power 62a, 62b, 62c on account of the difference formation. Therefore, a direction parameter 68 that is used for approximate direction determination for the voice signal from the interlocutor can be ascertained on the basis of that angle among the left deviation angle 56a, the zero angle 56b and the right deviation angle 56c for which the normalized angle-dependent interference power 66a, 66b, 66c is at a minimum. On the one hand, the direction parameter 68 can then be used in the relevant frequency band directly to determine the adaptation coefficient 44, and on the other hand, direction parameters 68 for multiple frequency bands 39 can be averaged in order to use this to associate the voice signal from the interlocutor, which is the main sound signal for the user 1, with one of the three cited angles.
Although the invention has been illustrated and described in more detail by the preferred exemplary embodiment, the invention is not restricted by this exemplary embodiment. Other variations can be derived therefrom by a person skilled in the art without departing from the scope of protection of the invention.
The following is a summary list of reference numerals and the corresponding structure used in the above description of the invention:
| Number | Date | Country | Kind |
|---|---|---|---|
| 102015211747.2 | Jun 2015 | DE | national |
