EP2595414B1 - Hearing aid with a device for reducing a microphone noise and method for reducing a microphone noise - Google Patents

Description

The invention relates to a hearing device in which at least one microphone is coupled to a device for reducing a microphone noise. The invention also includes a method for reducing microphone noise in an input signal of a hearing device. The term "hearing device" is understood here in particular as a hearing device. In addition, the term includes other portable and non-portable acoustic devices such as headsets, headphones and the like.

Hearing aids are portable hearing aids that are used to care for the hearing impaired. In order to meet the numerous individual needs, different types of hearing aids such as behind-the-ear hearing aids (BTE), hearing aid with external receiver (RIC: receiver in the canal) and in-the-ear hearing aids (IDO), e.g. Concha hearing aids or canal hearing aids (ITE, CIC). The hearing aids listed by way of example are worn on the outer ear or in the ear canal. In addition, bone conduction hearing aids, implantable or vibrotactile hearing aids are also available on the market. The stimulation of the damaged hearing takes place either mechanically or electrically.

Hearing aids have in principle as essential components an input transducer, an amplifier and an output transducer. The input transducer is usually a sound receiver, z. As a microphone, and / or an electromagnetic receiver, for. B. an induction coil. The output transducer is usually used as an electroacoustic transducer, z. As miniature speaker, or as an electromechanical transducer, z. B. bone conduction, realized. The amplifier is usually integrated in a signal processing unit. This basic structure is in FIG. 1 shown using the example of a behind-the-ear hearing aid. In a hearing aid housing 1 for carrying behind the ear, one or more microphones 2 for receiving the sound from the environment are installed. A signal processing unit 3, which is also integrated in the hearing aid housing 1, processes the microphone signals and amplifies them. The output signal of the signal processing unit 3 is transmitted to a loudspeaker or earpiece 4, which outputs an acoustic signal. The sound is optionally transmitted via a sound tube, which is fixed with an earmold in the ear canal, to the eardrum of the device carrier. The power supply of the hearing device and in particular the signal processing unit 3 is effected by a likewise integrated into the hearing aid housing 1 battery. 5

The microphones 2 may be condenser microphones. The disadvantage of this type of microphone is that condenser microphones produce an inherent noise. This microphone noise is always superimposed on the sound signal detected by means of the condenser microphone and can be perceived by the user of the hearing device as an unwanted artifact in a quiet environment via the handset 4. If the hearing aid compensates for hearing loss by means of frequency-selective amplification of an input signal, the amplified frequencies are more likely to raise the level of microphone noise above the audiometric user's hearing threshold, so that the user will always hear unwanted noise even in a quiet environment. The microphone noise has a typically characteristic frequency response similar to that of a pink noise.

In order to avoid that a user perceives the microphone noise in a silent environment, it is desirable to suppress the microphone noise in the input signal of the hearing device whenever the microphone noise is not superimposed by a signal of ambient sound and thereby masked or covered. For this purpose, the input signal is known a hearing device in response to a level of the input signal by means of a compressor whose characteristic for input signals of low level, as typically result for the microphone noise alone, causes an attenuation of the input signal. On the other hand, for input signals that significantly exceed a certain minimum level, the characteristic curve of the compressor has a slope of one, ie microphone signals with a high input level are not influenced by the compressor. The characteristic of the processor can be adapted to a type of microphone, but is usually fixed.

However, due to a temperature change or due to aging of the microphone, the power density spectrum of the microphone noise may change such that in at least some frequency channels of the compressor the level of the microphone noise lies in the region of the transition of the characteristic from the compressor to the neutral region with the gain unity , As a result of this, relative level fluctuations of the microphone noise are amplified by the gain factor of the compressor, which is then level-dependent on the input signal, in the output signal of the compressor. Thus, the noise for a user of the hearing aid is particularly noticeable. A temperature dependence of the power density spectrum of the microphone noise and a dependence on an age of the microphone can not be compensated without costly additional measures by means of a compressor.

From the publication US 2003/0147538 A1 a method for reducing wind noise is known. The acoustic input signal is filtered by means of a Wiener filter if the ratio of difference to sum of the input signal powers exceeds a threshold value.

From the publication DE 10 2008 055760 A1 For example, a method of reducing microphone noise using a directional microphone array is known. The directivity is doing so adapted that non-stationary microphone noise is masked by stationary background noise.

An object of the present invention is to provide an input signal with a low microphone noise in a hearing device.

The object is achieved by a method according to claim 1 and a hearing device according to claim 13. Advantageous developments of the invention are given by the dependent claims.

In the method according to the invention, a microphone noise contained therein is reduced in an input signal by filtering the input signal by means of a Wiener filter if a noise power determined for the input signal is smaller than a predetermined limit value. If the noise power is greater than the threshold or equal to the threshold, on the other hand, the Wiener filter is deactivated.

Accordingly, it is provided in the hearing device according to the invention to couple a microphone with a device for reducing a microphone noise. This device comprises a Wiener filter and an estimator coupled to it for determining an estimate of a noise power. By means of the Wiener filter is thereby an input signal received by the device, e.g. a microphone signal, with a damping applied, the value of which is determined on the basis of the estimate of the noise power. The thus filtered input signal then forms an output signal of the device for further processing in the hearing device.

In the hearing device according to the invention, the device for reducing a microphone noise is additionally configured to monitor the noise power estimate and disable the Wiener filter if the estimate is greater than a predetermined threshold. In the context of the invention, the deactivation of the Wiener filter is understood to mean that its influence on the input signal is reduced completely or at least to an extent which is insignificant for further processing.

The method according to the invention and the device according to the invention have the advantage that the microphone noise in the quiet environment, when the noise power contained in the input signal is below the limit value, can be very flexibly suppressed by means of the Wiener filter. The Wiener filter is able due to the time-dependent determination of the noise power, temperature or age-related To follow changes in the power density spectrum of the microphone noise and thus always adapt the attenuation to the current course of the power density spectrum. By deactivating the Wiener filter upon detection of a noise level exceeding the threshold, it is also effectively prevented that the microphone noise reduction device undesirably alters microphone signals that are not generated by the microphone itself but by environmental sound.

In order to be able to deactivate the Wiener filter as a function of noise power, an embodiment of the method according to the invention provides for weighting an attenuation of the Wiener filter, the so-called "gain", acting on the input signal with a weighting factor which is a function of the determined noise power , This results in the advantage that it is possible to use a Wiener filter structure known per se from the prior art, the damping or gain of which then acts on the input signal of the hearing device as a function of the noise power or not.

The increase in microphone noise fluctuation described in the context of the compressor in the event that its power is close to the limit can be easily avoided in the inventive method with an embodiment in which the attenuation of the Wiener filter attenuated in a gradual transition so that there is a transition between fully active damping and fully deactivated damping. In this case, a transition according to a ramp function and a hyperbolic tangent function have proven particularly suitable.

Furthermore, it has proven expedient to limit the determined noise power to a predetermined maximum value. Then the estimator for determining the noise power is able to determine a current value for the noise power very quickly, even if the Wiener Filter was deactivated for a while and then activated again in a quiet environment. By limiting the noise power to the maximum value, a period of time which the estimator needs to converge to the actual value of the noise power is significantly reduced.

The estimation of the noise power is expediently for a signal component of the input signal, ie for z. For example, at least one channel of a filter bank through which the input signal is spectrally analyzed, based on this signal component itself. For the estimation of the noise power, a statistical estimation method can be used as it is for the estimation of noise power of the prior art per se numerous variants is known.

Since the microphone noise is a noise inherent to the microphone, which is generated independently of the ambient noise, can be used to determine the noise power for at least a signal component of the input signal on a characteristic of the microphone. This has the advantage that no uncertainty estimation of the noise power is necessary in this signal component. The characteristic can be determined, for example, in the production of the hearing device or the microphone.

According to a further embodiment of the method according to the invention, it is provided to set the already described limit value for the activation or deactivation on the basis of a characteristic curve of a microphone. This makes it possible to specify very precisely for the different types of microphone and for individual frequency bands for which noise level the Wiener filter should be activated or deactivated.

The use of a Wiener filter to attenuate the microphone noise has the further advantage that on his Based on a processed microphone noise can be generated, which has no disturbing fluctuations, such as the well-known musical noise phenomenon. For this purpose, the inventive method can be further developed simply by limiting an attenuation of the Wiener filter acting on the input signal when the Wiener filter is active to a predetermined maximum attenuation value.

The method according to the invention can also be combined particularly advantageously with a beamformer in which a directional effect can be set by means of a directional parameter. This may be any type of adaptive beamformer available in the prior art. For the combination of the beamforming with the method according to the invention, the individual microphone signals of the microphones of the beamformer do not have to be processed individually either individually. In the method according to the invention, the input signal for the device for reducing the microphone noise is first formed from the plurality of microphone signals of the microphones by means of the beamformer, ie. H. only the (single) output signal of the beamformer has to be processed. In order to adapt the method according to the invention to the signal properties of the output signal of the beamformer, it is sufficient to first scale the input signal, ie the beamformer output signal, in response to a current value of the directional parameter of the beamformer when determining the noise power. This effectively compensates for erratic changes in the noise power density of the microphone noise contained in the input signal, as typically caused by the beamformer when setting new values for the directional parameter. Thus, again, a standard estimator can be used to estimate the noise power.

In order to use the thus determined noise power for calculating the attenuation of the Wiener filter, provides a development of the method, depending on the current Value of the straightening parameter to rescale the detected noise power. As a result, the estimated value for the noise power is followed by the sudden change of the microphone noise in the input signal.

In addition, according to another development, it is also intended to limit the attenuation of the Wiener filter acting on the input signal as a function of the current value of the directional parameter to a maximum value. As a result, a virtually flat power density distribution of the processed microphone noise can be achieved, that is, a far less disturbing white residual noise for a user.

In connection with the noise power-dependent deactivation of the Wiener filter, according to another embodiment of the method according to the invention, it is also provided to set the limit value for the deactivation as a function of a current value of the directional parameter. This results in the advantage that the microphone noise is suppressed by the Wiener filter even if it is amplified by an unfavorable setting of the Beamformers so far that it would otherwise exceed the limit.

The hearing device belonging to the invention has further developments which include features which have already been described in connection with the developments of the method according to the invention. For example, a further development of the hearing device according to the invention provides that a plurality of microphones is coupled to the device for estimating the noise power via a beamformer which is designed to generate an input signal for the device from the microphone signals of the microphones. In the case of the beamformer, as already described, a directivity can be set using at least one directional parameter. The noise power estimation means is arranged in the described manner to determine the noise power estimate from the microphone signals scaled input signal depending on a value of the directional parameter of the beamformer.

Since the features of the other developments of the hearing devices according to the invention result in a similar manner from the developments of the method according to the invention, they will not be explained again here.

FIG 1

eine schematische Darstellung eines Hinter-dem-Ohr-Hörgeräts,

FIG 2

ein Blockdiagramm zu einer Einrichtung zum Reduzieren eines Mikrofonrauschens, welche sich in einer Hörvorrichtung gemäß einer Ausführungsform der erfindungsgemäßen Hörvorrichtung befindet,

FIG 3

eine Kennlinie, gemäß welcher eine Dämpfung eines Wiener-Filters der Einrichtung von FIG 2 gewichtet wird,

FIG 4

ein Blockdiagramm zu einer Hörvorrichtung gemäß einer weiteren Ausführungsform der erfindungsgemäßen Hörvorrichtung,

FIG 5

einen Signalflussgraphen zu einem Beamformer wie er in der Hörvorrichtung von FIG 4 eingebaut sein kann,

FIG 6

ein Blockdiagramm zu einer Einrichtung zum Reduzieren eines Mikrofonrauschens, wie sie in der Hörvorrichtung von FIG 4 vorgesehen sein kann,

FIG 7

einen zeitlichen Verlauf eines Schätzwerts für eine Rauschleistung, wie er sich bei einer Rauschleistungs-Schätzeinrichtung der Hörvorrichtung von FIG 4 ergeben kann,

FIG 8

ein Diagramm zu einer Einstellung von Maximaldämpfungswerten, wie sie bei der Hörvorrichtung von FIG 2 und FIG 4 vorgesehen sein kann, und

FIG 9

ein Diagramm zu einer weiteren Einstellung von Maximaldämpfungswerten, wie sie bei der Hörvorrichtung gemäß FIG 4 vorgesehen sein kann.

The invention will be explained in more detail below with reference to exemplary embodiments. This shows:

FIG. 1

a schematic representation of a behind-the-ear hearing aid,

FIG. 2

1 is a block diagram of a device for reducing a microphone noise, which is located in a hearing device according to an embodiment of the hearing device according to the invention,

FIG. 3

a characteristic according to which a damping of a Wiener filter of the device of FIG. 2 is weighted,

FIG. 4

1 a block diagram of a hearing device according to a further embodiment of the hearing device according to the invention,

FIG. 5

a signal flow graph to a beamformer like him in the hearing aid of FIG. 4 can be installed

FIG. 6

a block diagram of a device for reducing a microphone noise, as in the hearing of FIG. 4 can be provided

FIG. 7

a temporal course of an estimate of a noise power, as in a Noise power estimation device of the hearing device of FIG. 4 can result

FIG. 8

a diagram of a setting of maximum attenuation values, as in the hearing device of FIG. 2 and FIG. 4 can be provided, and

FIG. 9

a diagram for a further setting of maximum attenuation values, as in the hearing device according to FIG. 4 can be provided.

The examples illustrate preferred embodiments of the invention.

In FIG. 2 a hearing device 10 is shown which may be, for example, a behind-the-ear hearing device or an in-the-ear hearing device. A microphone 12 detects an environmental sound and converts it into an analog electrical signal, which is converted by a preprocessing device 14 into a digital input signal x by means of an analog-to-digital conversion. In the preprocessing device 14 may also be provided to divide the signal of the microphone 12, for example by means of a filter bank in a plurality of frequency channels. The input signal x then comprises a corresponding number of narrowband sub-signals. From the input signal x, an output signal is generated by a signal processing device 16, which is converted by a receiver 18 into a sound signal and emitted to an ear of a user of the hearing device 10.

The microphone 12 may be, for example, a condenser microphone. In addition to the useful signal generated from the ambient sound (including a desired signal and environmental noise), the analog input signal always also contains a microphone noise, which is generated by the microphone 12 itself. In an environment where it is so quiet that in the input signal x, or at least in one of its frequency channels, the microphone noise is a significant has greater signal power than the signal component generated by the ambient sound, it is still not the case that the user of the hearing device 10 perceives the microphone noise on the handset 18. The microphone noise is suppressed by a damping W, which in the in FIG. 2 shown example as a multiplicative, optionally frequency-dependent attenuation factor W via a multiplier 20 acts on the input signal x or its individual frequency channels. The attenuation factor W is set by a device 22 for suppressing the microphone noise. When the ambient sound produces a level in the input signal x that is large enough to mask the microphone noise, the attenuation factor W for that time period and, optionally, for the corresponding frequency channel is set by the device 22 to a value of one or nearly unity. On the other hand, if the ambient sound is so quiet that the microphone noise could temporarily be heard via the receiver 18, the attenuation factor W is set to a value between zero and one for this period and, if appropriate, for the corresponding frequency channel, resulting in a noticeable attenuation. As a result, the microphone noise is then correspondingly reduced in the input signal x.

To set the attenuation factor W, the device 22 has a device 24 for calculating the power spectral density (PSD) of the input signal x and a Wiener filter 26 for calculating a gain W '. The gain W 'is calculated by the Wiener filter 26 from the power density PSD of the input signal x and an estimate of the noise power NPSD (Noise Power Spectral Density) according to a function f. The device 24 may comprise, for example, a simple squarer for determining an amplitude square of the input signal x or a squaring device and a downstream smoothing device for calculating a time average. Any other means of calculating a power density spectrum is also useful here. The function f for calculating the gain W 'can also be a calculation rule known per se from the prior art for attenuation of noise power contained in a signal. The function f results in a gain W 'with a value between zero and one, whereby the value tends the more against one, the larger the in FIG. 2 ratio shown is PSD / NPSD. The function f may also include an estimate of a signal-to-noise ratio (SNR).

The noise power NPSD is determined by an estimator 28 for a noise power contained in the input signal x and a characteristic curve 30 which describes to the microphone 12 the typical spectral noise power density of the microphone noise of this type of microphone. The characteristic 30 may, for example, have been created during the manufacture of the device 10 by measurements. The estimation device 28 is a device known per se from the prior art for determining a noise power in a signal.

In the device 22 is effected via a limiter 32, a switch 34 and a masking device 36, that the damping factor W acts only on those signal components of the input signal x, in which a level is so low that it is in the signal components with high probability exclusively or almost exclusively to the microphone noise of the microphone 12 is.

Depending on the switch position of the switch 34, either a fixed (frequency dependent) estimate of the noise power determined based on the characteristic 30 or an actual estimate of the noise power from the estimator 28 is provided to the Wiener filter 26 and the fader 36 In case the estimator 28 is used, the noise power estimate NPSD is limited by the limiter 32 to a predetermined maximum value. For the following explanations, assume that the maximum value is 40 dB. In the case of the decibel used here and below it is decibel for the sound pressure level (SPL - Sound Pressure Level). The maximum value for the noise power estimate can be derived from the characteristic curve 30, wherein an offset of, for example, 25 dB is added to the characteristic value.

The combination of the estimator 28 and the limiter 32 altogether forms a noise power estimate which operates exclusively within the microphone noise level region. This causes the function f to calculate a value for the gain W 'through the Wiener filter 26 which automatically tends to unity when the input signal x has a spectral power density PSD significantly greater than the maximum value of the limiter 32 , so in this example is greater than 40 dB. Even with the direct use of the characteristic curve 30 as an estimate for the noise power, as can be achieved by corresponding switching of the switch 24, this automatic deactivation of the Wiener filter 26 results.

In addition, in order to obtain the audio quality of the sound signal of the listener 18 in the range of levels of the input signal x near the limit imposed by the limiter 32, the skimmer 36 additionally provides a gradual transition. The operation of the masking device 36 will be described below with reference to FIG FIG. 3 explained in more detail. In FIG. 3 For this purpose, a diagram is shown which represents the dependence of a gain factor GF on a current value for the noise power NPSD. On the basis of the gain factor GF is determined by the masking device 36, to what extent the damping factor W is calculated from the gain W '. A gain factor of unity means W = W '. A gain factor with the value zero means W = 1, ie the Wiener filter 26 is deactivated with respect to its influence on the input signal x.

The damping factor W is a function of the noise power NPSD. For a value of the noise power NPSD <20 dB: W = W '. For a noise power NPSD ≥ G = 30 dB, W = 1. In between, a transition 38 is formed by the masking device 36, which can run, for example, according to a ramp function 40 or a hyperbolic tangent function 40 '. The value G represents a limit for the activation or deactivation of the Wiener filter.

To illustrate the mode of operation of the device 22, the expected noise power 42 determined by the characteristic curve 30 and the maximum value 44 defined by the limiter 32 are also shown in the diagram. Other than shown here, the maximum value 44 is expediently set equal to the value G.

On the basis of the measurement of the microphone noise, a noise level-dependent limitation of the gain W 'is performed by the masking device 36. The closer the noise power NPSD estimate approaches the G limit, the more damping is reduced. This ensures that signal components that are not dominated by microphone noise remain undamped. This prevents the device 22 from interacting with further signal-processing algorithms in the signal processing device 16. At the same time, the possibility is provided to parameterize the device 22 for suppressing the microphone noise independently of the other algorithms, e.g. in that a stronger maximum attenuation brought about by the gain W 'is exerted on the microphone noise than on an ambient noise by the signal processing device 16.

In FIG. 4 For example, a microphone 46, microphones 48, 50, filter banks 52, 54, a beamformer 56, a microphone noise reducer 58, and a multiplier 60 are shown. By the beamformer 56 are each separately in individual frequency channels of the filter banks 52, 54, the digitized microphone signals of the microphones 48 and 50 summarized to achieve a directivity. The beamformer 56 may be a beamformer known per se from the prior art. Exemplary is in this FIG. 5 shown a possible structure of the beamformer 56, as it may be provided for processing a single channel of the filter banks 52, 54. Delays having a delay time constant T0 delays the microphone signals of the microphones 48, 50 and then combines them into directional signals via adders, resulting in a cardioid signal and an anti-cardioid signal.

By means of a multiplier, one of the signals is weighted with the value of a directional parameter a, before the two signals are combined to form a directed beamformer signal x 'by means of a further adder. The arrangement described has a clearly perceptible high-pass characteristic. For this reason, low frequencies are amplified by an amplifier 62 to make the audio information contained therein audible again to the user. This amplification also acts on a microphone noise contained in the directional signal x ', which is caused by the two microphones 48, 50. In the input signal x for the hearing device 46, which the amplifier 62 generates, the microphone noise due to the amplification has a different spectral power density distribution than the original microphone noise of the microphones 48, 50 itself. In addition, by changing the value of the directional parameter a, the Spectral power density of the microphone noise in the input signal x is changed over time. In the case of the hearing device 46, these properties of the microphone noise are taken into account in the calculation of an attenuation W in the input signal x, so that a user of the hearing device 46 does not perceive any disturbing microphone noise even if the value of the directional parameter a changes with time.

The input signal x and the directional parameter a form input values for the device 58. By means of the device 58, comparable to the device 22, a damping factor W is calculated which acts on the input signal x of the hearing device 46 via the multiplier 60. By the damping factor W is achieved in the same way as in the case of the hearing device 10, the microphone noise is reduced to the input signal x, without a dominating component in the input signal x caused by ambient sound being likewise influenced by the damping factor W.

In the FIG. 6 is for the explanation of the operation of the device 58 this again shown in more detail. In FIG. 6 are components that function like components in FIG. 2 are shown, provided with the same reference numerals as in FIG. 2 , They are related to FIG. 6 not explained again.

The change in the spectral power density of the microphone noise in the input signal x caused by the value of the directional parameter a is compensated by calculating the change in the power density spectrum by a computing device 64 in the form of a wite noise gain WNG and by a divider 66 in the form of a scaling the input signal x is taken into account. The noise power NPSD calculated by the estimator 28 from the scaled input signal x / WNG is scaled back to a scaled-back noise power NPSD 'by means of a multiplier 68 and the value for the wide noise gain WNG. In connection with in FIG. 5 shown beamformer 56 results as a scaling value WNG for a frequency with the normalized center frequency Ω of the filter banks 52, 54 following calculation rule, in which Ω = 2 * n * f * Ts and Ts is the sampling time of the analog-to-digital converter of the hearing 46: $$\mathrm{WNG}\left(\mathrm{\Omega }\right)=\left[{\mathrm{a}}^{\mathrm{2}}+\mathrm{1}+\mathrm{2}\mathrm{a}*\cos \left(\mathrm{\Omega }*\mathrm{T}\mathrm{0}/\mathrm{ts}\right)\right]/\left[\mathrm{1}-\mathrm{<figure-callout\; id="2"\; label="cos"\; filenames="imgf0001.png,imgf0007.png"\; state="\{\{state\}\}">cos</figure-callout>}\left(\mathrm{2}*\mathrm{\Omega }*\mathrm{T}\mathrm{0}/\mathrm{ts}\right)\right]\mathrm{,}$$

Based on FIG. 7 is shown below for a single frequency component of the input signal x, which estimates for the noise power may result in the device 58.

This is in the in FIG. 7 bottom diagram shows how the value for the directional parameter a with the time t in a period of 14 seconds is changed gradually, so that at the beginning of an omnidirectional directional characteristic of the beamformer 56 results and then gradually by a zero Steering a notch (Notch) in the directional characteristic in different, in FIG. 7 specified angular directions is aligned, in order to be switched again from the twelfth second to an omnidirectional directional characteristic. For the corresponding values of the directional parameter a, the corresponding value of the wide noise gain WNG in decibels is shown in the graph above. For that the FIG. 7 underlying example is assumed that the microphone is operated in a quiet environment, so that the input signal x in the in FIG. 7 shown frequency component exclusively has the stationary microphone noise. The gradual change of the wide noise gain WNG, however, results in a course of the square of squares | x | ^{2 of} the input signal x, as shown in the top diagram of FIG. 7 is shown. Based on this history, the estimator 28, due to its inertia, would not be able to correctly understand the noise power at the discontinuities (eg, at 2 second). Due to the scaling by means of the divider 66, however, an input signal for the estimation device 28 results in a steady-state input signal x / WNG. The estimator 28 calculates a corresponding correct noise power NPSD for this scaled input signal x / WNG. The rescaling by means of the multiplier 68 then yields a corresponding correct estimate NPSD 'for the actual noise power contained in the input signal x. This is used to calculate a gain W 'suitable for the effective attenuation of the microphone noise by means of the Wiener filter 26.

In the device 58 may still further, in connection with FIG. 2 explained components for deactivating the Wiener filter 26 by means of a limiter, a switch and a masking device may be provided. These components are in FIG. 6 for the sake of clarity, not shown again.

In FIG. 8 and FIG. 9 Two alternative possibilities are shown for further improving the audio quality of the input signal x processed by means of the multiplier 20 or 60. The diagrams are made in the event that a beam former, such as the Beamformer 56, is used. The noise power of the microphone noise, especially for low frequencies, may be significantly enhanced by the beamformer 56, depending on the value for the directional parameter a relative to the original microphone noise of the microphones 48, 50. In FIG. 8 and FIG. 9 For this purpose, for several channels C of the filter banks 52, 54, the level of the microphone noise for a specific time and a specific setting of the parameter a before damping by means of the multiplier 60 (as a bar graph | x | ² ) and after the attenuation (as a bar graph | x | ² * W ² ). In order not to amplify a fluctuation of the level of the microphone noise by the time-variant attenuation W, the attenuation W is limited to a maximum attenuation value NF, here on a logarithmic scale in the in FIG. 8 shown example NF = - 10 dB. Correspondingly, in the low-frequency range (here, in particular, channels 0 to 6), if necessary, perceptible signal portions of the microphone noise in the input signal are retained even after the attenuation has been applied to a user. However, due to the limitation of the attenuation to the maximum attenuation value NF, these have a little disturbing modulation.

For the other channels (channels C = 6-47) results in such a strong attenuation of the microphone noise that the same is no longer perceptible to the user. The microphone noise also has the stationary behavior after the processing, which it also showed before the processing by the beamformer.

In order to achieve in the determination of the value for the maximum attenuation NF that the microphone noise is reduced to a comfortable level, in addition the beamformer characteristic z. B. be taken into account in the form of the value of the directional parameter a. Thus, by means of the white noise gain WNG a frequency-dependent maximum attenuation NF (C, a) can be set. The goal here is to obtain a muted microphone noise, in which the channels C have a nearly equal level of microphone noise and this level regardless of a current setting of the beamformer, d. H. of the value for the directional parameter a, is.

Such a frequency-dependent setting of the maximum attenuation NF (C, a) is in FIG. 9 shown in the right diagram. The values for the maximum attenuation NF (C, a) are frequency- and time-dependent and represent a function of the value of the directional parameter a. The left-hand diagram shows how such a maximum limitation NF (C, a) causes a spectrally nearly flat one History of the microphone noise is achieved. The limitation of the damping to a maximum damping can be carried out, for example, within the Wiener filter 26. It should be noted here that a limitation of the attenuation means that the Wiener gain W 'does not become smaller than a value corresponding to the value NF or NF (C, a). Due to the limitation of the attenuation factor W to small values, a so-called noise floor (noise floor) results in the processed input signal.

On the basis of the value for the aiming parameter a, the limit value G for the masking device 36 can also be set when it is provided in the device 58. As a result, it is then prevented that the Wiener filter is deactivated solely because the beamformer 56 results in a level of the microphone noise that is greater than the level of the microphone noise that is to be expected on the basis of the characteristic curve 30.

In summary, it can be stated that even with a beamformer with adjustable directivity an efficient reduction of the microphone noise to a comfortable level is possible. In addition, the approach shown has the advantage of avoiding so-called "noise flags" which are otherwise typically caused in a signal from a beamformer. Such noise flags may be followed by a signal from an external sound source, such as a speaker, when that sound source is silenced and then the microphone noise becomes audible to the user of the hearing device because it is not attenuated fast enough. The fast adaptation is among the approaches shown u.a. by the limiter 32, which keeps the noise power estimation of the microphone noise NPSD at a level already very close to the actual microphone noise.

Claims (14)

1. Method for reducing a microphone noise in an input signal (x) of a hearing apparatus (10, 46), by filtering the input signal (x) by means of a Wiener filter (26), if a noise power (NPSD) determined for the input signal (x) is smaller than a predetermined limit value (G), and deactivating the Wiener filter (26) if the noise power (NPSD) is greater than the limit value (G) or equal to the limit value (G).
2. Method according to claim 1, wherein, for noise power-dependent deactivation, an attenuation (W) of the Wiener filter (26) acting on the input signal (x) is weighted with a weighting factor (GF), which is a function (40, 40') of the determined noise power (NPSD).
3. Method according to claim 2, wherein the function forms a gradual transition (38) between a completely active attenuation and a completely deactivated attenuation, in particular a transition (38) according to a ramp function (40) or a tangens hyperbolicus function (40').
4. Method according to one of the preceding claims, wherein the determined noise power (NPSD) is limited to a predetermined highest value (44).
5. Method according to one of the preceding claims, wherein the noise power (NPSD) for at least one signal part of the input signal (x) is estimated on the basis of this signal part according to a statistical estimation method (28).
6. Method according to one of the preceding claims, wherein the noise power for at least one signal part of the input signal (x) is determined on the basis of a characteristic curve of a microphone.
7. Method according to one of the preceding claims, wherein the limit value is defined on the basis of a characteristic curve (30) of a microphone (12).
8. Method according to one of the preceding claims, wherein attenuation (W) of the Wiener filter (26) acting on the input signal (x) is limited to a predetermined maximum attenuation value (NF) with an active Wiener filter.
9. Method according to one of the preceding claims, wherein the input signal (x) is formed from a plurality of microphone signals by means of a beamformer (46), in which a directional effect can be set with the aid of a directional parameter, and when determining the noise power (NPSD), the input signal (x) is initially scaled as a function of the current value of the directional parameter (a).
10. Method according to claim 9, wherein the determined noise power (NPSD) is back-scaled as a function of the current value of the directional parameter.
11. Method according to claim 9 or 10, wherein an attenuation of the Wiener filter acting on the input signal is limited to a highest value (NF, (C, a)) as a function of the current value of the directional parameter (a).
12. Method according to one of the claims 9 to 11, wherein the limit value (G) is a function of the value of the directional parameter.
13. Hearing apparatus (10, 46) in which at least one microphone (12, 48, 50) is coupled to a facility (52, 58) for reducing a microphone noise in an input signal (x) of a hearing apparatus, which has a Wiener filter and an estimation facility (28) coupled hereto for determining an estimated value for a noise power (NPSD) of the input signal, wherein the input signal can be subjected to an attenuation (W) by means of the Wiener filter (26) in order to generate a processed input signal and a value of the attenuation can be determined on the basis of the estimated value for the noise power (NPSD),
    characterised in that
    the facility (22, 58) is set up to monitor the estimated value for the noise power, to filter the input signal (x) by means of a Wiener filter (26) if the estimated value is less than a predetermined limit value (G) and to deactivate the Wiener filter, if the estimated value is greater than a predetermined limit value (G).
14. Hearing apparatus (46) according to claim 13, characterised in that a plurality of microphones (48, 50) is coupled to the facility (58) via a beamformer (56), by means of which an input signal (x) can be generated from microphone signals of the microphone (48, 50) for the facility and in which a directional effect can herewith be set with the aid of a directional parameter (a), wherein the facility (58) is set up to reduce the microphone noise, to determine the estimated value by means of the estimation facility (28) for the noise power (NPSD), to scale the input signal as a function of a value of the directional parameter (a).

Images