EP2182739A1 - Adaptive microphone system for a hearing aid and accompanying operating method - Google Patents

Abstract

The method involves filtering two microphone signals of respective microphones (1, 2) by finite impulse response filter units (3, 4), respectively. Directivities of the microphones are adjusted by changing an adaptation parameter to minimize the sum of interference power by a control unit (6). Noise floor of ambient noise is compared with a microphone noise number by the control unit, and a limit of value range of the adaptation parameter is determined based on the comparison result of the control unit. An independent claim is also included for a microphone system comprising an onmidirectional directional microphone.

Description

The invention relates to a specified in claim 1 method for suppressing microphone noise and specified in claim 7 associated microphone system.

In acoustic systems and in particular in hearing aids, it is advantageous to combine several microphone signals and to spatially and spectrally filter them so that the output signal contains the lowest possible interference components. Disturbances are defined on the one hand as signals that are incident from undesired directions, for example outside a predetermined angular range about the 0 ° direction and on the other hand as microphone noise, which is amplified when forming the directivity, especially in low-frequency areas. In particular, there is the problem that the microphone noise increases when the directivity of a directional microphone is increased.

In the DE 10 2004 052 912 A1 An acoustic system and a method are specified which suppress interference power in directional microphones as far as possible. For this purpose, the microphone signals of several microphones are adaptively filtered as a function of at least one parameter. The directivity of the directional microphone obtained in this case is adjusted by changing the at least one parameter such that the sum of noise power including microphone noise is reduced or minimal. Thus, depending on the noise distribution between a directional operation and an omnidirectional operation blended.

That in the DE 10 2004 062 912 A1 described method leads to a minimization of the total power of microphone noise and ambient noise. The residual noise consists in half of residual ambient noise and residual microphone noise. In the mathematical sense, the overall disturbance is minimal, but not for the subjective sound impression of a user of the acoustic system. Due to rapidly changing signal components and wide subband signals, disturbing microphone noise is noticeable to the user over and over again. Particularly non-stationary interferers, such as speech, cause a brief transition to directional operation. If the interferer is then again not active, there is a delayed fading to the omnidirectional operation, so that noise flags are audible for a short time.

It is an object of the invention to overcome these disadvantages and to provide a device and an associated method which prevent perceptible microphone noise.

According to the invention, the stated object is achieved with the method for operating a microphone system of independent claim 1 and the microphone system of independent claim 7.

• adaptives Filtern der mindestens zwei Mikrofonsignale mit mindestens einem Adaptionsparameter,
• Einstellen der Richtwirkung durch Verändern des mindestens einen Adaptionsparameters derart, dass die Summe von Störleistungen minimiert wird, und
• Begrenzen des Wertebereichs des mindestens einen Adaptionsparameters, wobei die Grenzen aus einem Vergleich des Rauschteppichs des Umgebungsrauschens mit einer Mikrofonrauschzahl ermittelt werden.

Dies bietet den Vorteil, dass der Adaptionsbereich eines adaptiven differentiellen Richtmikrofons abhängig vom stationären Anteil des Hintergrundrauschens ist, wodurch die Richtwirkung immer so gewählt wird, dass das durch die Richtwirkung entstehende, instationäre Mikrofonrauschen fast immer durch den stationären Anteil des Hintergrundrauschens maskiert wird. Damit wird situationsangepasst ein ruhiger Klangeindruck ohne Rauschartefakte bei gleichzeitig maximal möglicher Richtwirkung erzielt.The invention provides a method for operating a microphone system with at least two omnidirectional microphone-emitting microphones, wherein the microphones are electrically interconnected to form a directivity. The method comprises the following steps:

• adaptively filtering the at least two microphone signals with at least one adaptation parameter,
• Setting the directivity by changing the at least one adaptation parameter such that the sum of noise power is minimized, and
• Limiting the range of values of the at least one adaptation parameter, the limits being determined from a comparison of the noise floor of the ambient noise with a microphone noise figure.

This offers the advantage that the adaptation range of an adaptive differential directional microphone is dependent on the stationary portion of the background noise, whereby the directivity is always chosen so that the transient microphone noise resulting from the directivity is almost always masked by the stationary portion of the background noise. This situation adapted to a quiet sound impression achieved without noise artifacts with maximum possible directivity.

In a further development, the method can be carried out separately for a plurality of frequency subbands. This provides improved directivity with simultaneous noise suppression.

In another embodiment, the noise carpet may be determined using Wiener filters or non-linear power estimators. An advantage of this is a simple and robust noise power determination.

Furthermore, the value of the microphone noise number can be predetermined depending on the microphone, with a data sheet value of the microphone noise of the microphones and at least one distance of the microphones from one another being taken into account. An advantage of this is the use of microphone-specific parameters.

In a further development, the interference power by directivity can include amplified microphone noise and power from unwanted signal sources.

Advantageously, the range of values can be chosen such that the microphone noise amplified by the directivity is masked by the stationary portion of the background noise.

The invention also provides a microphone system having at least two omnidirectional microphones emitting microphones, the microphones being electrically interconnected to form a directivity. The microphone system comprises at least one filter unit with at least one adaptation parameter for adaptively filtering the at least two microphone signals to obtain the directivity and a control unit with which the at least one adaptation parameter is variable such that the sum of interference power is reduced. The value range of the at least one adaptation parameter is limited, wherein the control unit the limits are determined from a comparison of the noise floor of the ambient noise with a microphone noise figure.

In a refinement, the at least one filter unit can have separate filters for a plurality of frequency subbands, so that the change in the at least one adaptation parameter can be carried out separately in a plurality of frequency subbands.

In a further embodiment, the noise carpet may be detected by means of Wiener filters or non-linear power estimators in the control unit.

Advantageously, the value of the microphone noise number in the control unit can be predetermined as a function of the microphone, whereby a data sheet value of the microphone noise of the microphones and at least one distance of the microphones from one another are taken into account.

Furthermore, the spurious power by directivity may include amplified microphone noise and power from unwanted signal sources.

In a development, the value range can be selected by the control unit such that the stationary portion of the background noise masks the microphone noise amplified by the directivity.

The invention also claims a hearing aid with a microphone system according to the invention for carrying out a method according to the invention. This offers the advantage that hearing aid users no longer perceive the resulting microphone noise perceptually.

Other features and advantages of the invention will become apparent from the following explanations of an embodiment with reference to schematic drawings.

Figur 1:

ein Prinzipschaltbild eines Mikrofonsystems erster Ordnung,

Figur 2:

ein Diagramm zur Optimierung des Adaptionsparameters,

Figur 3:

einen Verlauf des Rauschteppichs und des Mikrofonrauschens in Abhängigkeit der Frequenz und

Figur 4:

einen Verlauf des Grenzwerts des Adaptionsparameters in Abhängigkeit der Frequenz.

Show it:

FIG. 1:

a schematic diagram of a microphone system first order,

FIG. 2:

a diagram for optimizing the adaptation parameter,

FIG. 3:

a course of the noise carpet and the microphone noise as a function of frequency and

FIG. 4:

a profile of the limit value of the adaptation parameter as a function of the frequency.

FIG. 1 shows a differential microphone of the first order. Two microphones 1, 2 receive a time-dependent sound signal s (t). In each case, a microphone noise signal n ₁ (t) or n ₂ (t) mixes with the ideal microphone signals. The respective sum signals are digitized with an analog-to-digital converter and thus provide the digital, noisy microphone signals x ₁ (k) and x ₂ (k).

zugeführt. Die Entzerrung liefert ein Monoausgangssignal y(k).Known for achieving a directivity, but in FIG. 1 not shown, is to cross-subtract the two microphone signals x ₁ (k) and x ₂ (k). The signals are delayed in the corresponding paths with timers and a difference signal is multiplied by an adaptation parameter a. The resulting signals are added and for equalization in Nutzsignalrichtung an equalizer 5 with a transfer function $\mathrm{H}\left(\mathrm{z}\right)=\frac{1}{1-z^{-2}}$

fed. The equalization provides a mono output signal y (k).

The differential microphone of the first order can also be used as in FIG. 1 represented by two FIR filter units 3, 4 with the transfer functions 1 + az ^-1 and -az ^-1 realize. The filter coefficients can not be chosen freely, but depend on the adaptation parameter a. This dependence, which results from conversion of the filtering from the differential microphone, ensures that the output signal after the directional microphone processing contains the signal from the 0 ° direction (useful signal direction) unchanged, independently of the choice of the parameter a. In order to optimize the adaptation parameter a, this is adapted to the respective one adapted acoustic situation. At a = -1 there is no directivity, the microphone system has an omnidirectional character, at a = 0 the sound is attenuated from the direction 180 ° and with increasing a the notches (= directions of greatest attenuation) move forward in the directional diagram. The value of the adaptation parameter a is supplied from an output of a control unit 6 to the filter units 3, 4.

With larger directivity, that is with increasing a, but also increases the microphone noise. However, it is desirable that the total interference power of a directional microphone is as low as possible. Therefore, on the one hand to adjust the directivity of the directional microphone so that the sound of a source of interference is suppressed as well as possible and on the other hand to keep the microphone noise as low as possible. In FIG. 2 For better understanding, the power of the interference signal ST and the microphone noise MR are qualitatively plotted against the adaptation parameter a. A sum signal SUM from the two signals ST and MR represents the total interference power for the directional microphone. By known methods, such as in DE 10 2004 052 912 A1 discloses, it is possible to find the minimum of the sum curve SUM and to use the corresponding parameter value a _min for the adaptive filter 3, 4.

The adaptation of the directional microphone to a specific interference source or the optimization of the parameter a can be carried out, for example, by a gradient method comparable to the LMS method (least mean squares). But there are also other embodiments conceivable. With the gradient method, the adaptation condition is very simple. It can be determined by minimizing the average output signal power of the directional microphone. This will be like in FIG. 1 represented the output signal y (k) of the control unit 6 supplied.

For the adaptation of the directional microphone, the minimization of the average output signal power is only possible because it is ensured by the special choice of the filter coefficients in dependence of the parameter a, that the useful signal from the 0 ° direction is not changed. The minimization of the total power (= useful signal + interference) is therefore equivalent to minimizing the power of the disturbance. The disturbance consists of two components: microphone noise and interference from signal sources that come from unwanted directions. An attenuation of direction-dependent signal sources can be achieved by selecting the parameter a> 0. By limiting to a maximum value, for example a = 2, the range in the 0 ° direction is determined - in this case +/- 60 ° - in which incident signal sources are not or only slightly attenuated. If the adaptive method is also allowed to select the parameter a smaller than 0, the directivity is reduced, but the microphone noise is also reduced. At a = -1 no directivity occurs and the microphone system of the microphones 1, 2 acts exclusively omnidirectional.

By adapting the parameter a in individual frequency bands, the method achieves that the sum of the interference powers, i. of microphone noise and signal sources from undesired directions, in each frequency band is minimized.

A disadvantage of this adaptation is that due to a finite processing time with rapidly changing interference signals, such as voice from undesired direction, the adaptation parameter a can not be readjusted so quickly to suppress unwanted microphone noise. As a result, for a short time microphone noise as so-called noise flags is disturbing audible. This is where the invention starts. At the price of a reduced directivity, the microphone noise is suppressed by limiting the range which the adaptation parameter a can assume depending on the ambient noise. This makes it possible to mask the disturbing noise flags by ambient noise. The limitation of the adaptation parameter a is in FIG. 2 clarified by a _zul .

With the help of the representations of the FIGS. 3 and 4 the invention will be explained in more detail. According to FIG. 3 is according to the invention First, a stationary noise carpet NF of ambient noise in 48 signal sub-bands determined. This is plotted as a bar graph with the signal power P in dB. To determine the ambient noise NF, as in FIG. 1 represented the microphone signals x ₁ (k) and x ₂ (k) inputs of the control unit 6 supplied. From data sheet values of the microphones 1, 2 and the distance of the two microphones 1, 2 from each other, a theoretical value of the microphone noise MN, also referred to as microphone noise number, is determined as a function of the frequency f.

In a further step, the range of the adaptation of the parameter a as a function of the frequency f is now limited so that it is no longer possible for the adaptation to select the directional microphone setting so that the resulting microphone noise is above the measured noise carpet NF, ie perceptually perceived by the user. In FIG. 4 the limit value A of the adaptation parameter a is represented in the form of vertical bars as a function of the 48 signal subbands. For the lower bound always a = -1. From the FIGS. 3 and 4 It can be seen that for smaller differences from ambient noise NF and microphone noise MN, the upper limit A of the adaptation parameter a becomes smaller.

The inventive step is to use the noise carpet NF for activating the directional microphone mode in each band and not the overall signal level or noise level. This ensures that short-term transient interferers do not lead to a fading into the directional microphone mode and thus to perceptible microphone noise, among other things by noise flags. For the calculation of the noise floor NF in the individual bands, methods which are known from the Wiener filter-based, single-channel noise reduction can be used, or non-linear power estimators, which follow increasing level values more slowly than falling ones.

An analogue structure and an analogous method are used for directional microphones of higher orders. Preferred application finds the microphone system and the associated method in hearing aids.

LIST OF REFERENCE NUMBERS

1, 2

microphone

3, 4

filter unit

5

equalizer

6

control unit

a

adaptation parameter

a _min

minimal adaptation parameter a

a _perm

permissible adaptation parameter a

A

Limit value of the adaptation parameter a

f

frequency

MR

microphone noise

MN

Microphone noise figure

n ₁ (t), n ₂ (t)

Microphone noise signal

NF

noise floor

P

interference power

SUM

Cumulative noise

ST

interference

x ₁ (k), x ₂ (k)

microphone signal

y (k)

output

Claims (13)

Method for operating a microphone system having at least two omnidirectional microphones emitting microphones (1, 2), wherein the microphones are electrically interconnected to form a directivity by: adaptively filtering the at least two microphone signals with at least one adaptation parameter (a) and Adjusting the directivity by changing the at least one adaptation parameter (a) such that the sum (SUM) of interference power (ST, MR) is minimized, characterized by : Limiting the range of values of the at least one adaptation parameter (a), the limits (A) being determined from a comparison of the noise floor (NF) of the ambient noise with a microphone noise figure (MN). Method according to claim 1,
characterized,
that it is carried out separately for several frequency subbands. Method according to claim 1 or 2,
characterized,
that the noise floor (NF) is determined using Wiener filters or non-linear power estimators. Method according to one of claims 1 to 3,
characterized,
that the value of the microphone noise number (MN) is predetermined depending on the microphone, with a data sheet value of the microphone noise of the microphones (1, 2) and at least one distance between the microphones (1, 2) are considered each other. Method according to one of the preceding claims,
characterized,
that the stray powers (SUM) comprise directivity enhanced microphone noise (MR) and undesired signal source (ST) power. Method according to one of the preceding claims,
characterized,
in that the value range (A) is selected so that the microphone noise (MR) amplified by the directivity is masked by the stationary portion of the background noise (LF). Comprising microphone system at least two omnidirectional microphones emitting microphones (1, 2), the microphones (1, 2) being electrically interconnected to form a directivity, - At least one filter unit (3, 4) with at least one adaptation parameter (a) for adaptively filtering the at least two microphone signals to obtain the directivity and a control unit (6) with which the at least one adaptation parameter (a) can be changed in such a way that the sum (SUM) of interference power is reduced, characterized, - That the range of values of the at least one adaptation parameter (a) can be limited, wherein by the control unit (6) the limits (A) from a comparison of the noise carpet (NF) of the ambient noise with a microphone noise figure (MN) can be determined. Microphone system according to claim 7,
characterized,
the at least one filter unit (3, 4) having separate filters for a plurality of frequency sub-bands, so that the change in the at least one adaptation parameter (a) can be executed separately in a plurality of frequency subbands. Microphone system according to claim 7 or 8,
characterized,
that the noise floor is determined (NF) with the aid of Wiener filters or non-linear power estimators in the control unit. Microphone system according to one of claims 7 to 9,
characterized,
that the value of the microphone noise number is depending on the microphone can be predetermined (MN), wherein a data sheet value of the microphone noise of the microphones (1, 2) and at least one distance between the microphones (1, 2) are considered each other. Microphone system according to one of claims 7 to 10,
characterized,
that the stray powers (SUM) comprise directivity enhanced microphone noise (MR) and undesired signal source (ST) power. Microphone system according to one of claims 7 to 11,
characterized,
in that the value range (A) can be selected by the control unit (6) such that the stationary portion of the background noise (NF) masks the microphone noise (MR) amplified by the directivity. Hearing aid with a microphone system according to one of claims 7 to 12 for carrying out the method according to one of claims 1 to 6.

Images