EP3337189A1 - Method for determining a position of a signal source - Google Patents

Abstract

The invention specifies a method (10) for determining at least one direction of a useful signal source (6, 8) using an acoustic system (4) which comprises at least a first input transducer (16) and a second input transducer (18), the first input transducer ( 16) a first input signal (20) is generated from a sound signal from the environment and the second input converter (18) generates a second input signal (22) from the sound signal, with a plurality of angle-dependent directional characteristics (34) each with a fixed central angle (aj) and a given angular expansion (”) are formed, with the signal components for the individual directional characteristics (34) being examined for the presence of a useful signal from a useful signal source (6, 8). , And in a specific directional characteristic (34) determined useful signal source (6, 8) of the corresponding central control Angle (aj) is assigned as the direction of the useful signal source (6, 8).

Description

The invention relates to a method for determining at least one direction of a useful signal source by an acoustic system comprising at least a first input transducer and a second input transducer, wherein the first input transducer from a sound signal of the environment generates a first input signal and the second input transducer from the sound signal a second Input signal generated.

For the operation of a hearing aid, the algorithms which are used for user-specific amplification and more generally for sound adaptation to input signals of the hearing aid which are obtained from the ambient sound are often to be selected as a function of a respective hearing situation. The individual listening situations are given as frequently recurring patterns of superimpositions of a useful signal sound due to noise or noise in general, the pattern u. a. be typified by the type of noise occurring, the signal-to-noise ratio, the frequency response of the useful signal sound as well as temporal variations and averages of the sizes mentioned.

On the basis of a detected hearing situation, in particular the signal-to-noise ratio in the input signals can be improved in an efficient and resource-conserving manner, since in this case a reduction of the noise is achieved by means of its statistically predictable properties as part of typing as a hearing situation.

Especially in particularly complicated real acoustic environments - for example, with multiple noise sources, which also result in a high overall noise level - may be required to improve the sound quality, the position of a useful signal source, such as a speaker in a conversation to locate directly.

Currently known in binaural hearing aids is a localization of useful signal sources via a measurement of the transit time difference, which results from the different propagation times of a sound signal to the respective hearing aids worn on both ears of the user of the binaural hearing device. However, such a measurement requires on the one hand a high computational effort, on the other hand the fastest possible yet detailed transmission of the signal components from a hearing aid to the other hearing aid for evaluation. Both requirements have a negative effect on the energy consumption of the hearing aids.

The invention is therefore based on the object for an acoustic system to specify a method which locates a useful signal source with the least possible calculation and system complexity as accurately as possible.

The above object is achieved by a method for determining at least one direction of a useful signal source by an acoustic system comprising at least a first input transducer and a second input transducer, wherein the first input transducer from a sound signal of the environment generates a first input signal and the second input transducer generates a second input signal to the sound signal, the first input transducer from a sound signal of the environment generates a first input signal and the second input transducer from the sound signal generates a second input signal, wherein based on the first input signal and the second input signal, a plurality of angle-dependent directional characteristics with a given each Central angle and a respective same angular expansion are formed, wherein the signal components to the individual directivity to the presence of a useful signal from a Nutzsignalq uelle are examined, and wherein a determined in a specific directional characteristic Nutzsignalquelle the corresponding central angle is assigned as the direction of the useful signal source. Advantageous and partly inventive in themselves embodiments are the subject of the dependent claims and the following description.

In this case, an acousto-electrical converter is generally understood to include an input transducer which generates an electrical signal from a sound signal, that is to say in particular also a microphone.

Preferably, the individual directional characteristics each have a minimum or in each case a maximum of the sensitivity with respect to a test signal from the corresponding angular direction at their respective central angle. Directional characteristics include, in particular, directional cones which have the greatest sensitivity in the direction of their respective central angle, the sensitivity decreasing with increasing angular distance from the central angle. The degree of decrease of the sensitivity with increasing angular distance from the central angle can be taken in this case as a measure of the angular expansion of the directional characteristic. Alternatively, the individual directivity characteristics may each also have a minimum sensitivity at the respective central angle with respect to a given test signal, whereby the sensitivity with respect to the test signal increases with increasing angular distance from the central angle. The degree of increase in sensitivity with increasing angular distance from the central angle can then be used as a measure of the angular expansion. Preferably, the central angle of each two adjacent directional characteristics at a fixed angular distance from each other. This means that a kind of scan of an angular range is carried out by means of the individual directional characteristics, wherein in the case of a transition from a directional characteristic to its adjacent directivity, the central angle changes in each case by a constant amount.

The investigation of the signal components, including the presence of a useful signal, from a useful signal source can in particular in the individual directional characteristics based on the signal level or on the basis of one or more variables derived from the signal level.

The assignment of a central angle of a directional characteristic as the direction of a useful signal source determined in the corresponding directional characteristic now has the advantage that the useful signal source can be localized in the presence of other useful signal sources, for example, if a new, further useful signal source to an already existing useful signal source with a corresponding useful signal addition, which is spatially separated from the first, already existing useful signal source. The detection or localization of the individual useful signal sources is in this case not affected by the presence of other useful signal sources, as it might occur in a localization via a phase measurement, where each further useful signal source at least as a noise deteriorates the signal-to-noise ratio, and thus could affect the robustness and also the angular resolution of the localization. In particular, it is possible in this case to achieve a temporal resolution of the localization of a useful signal source by means of a permanent application or a repeated use of the localization at short time intervals. As a result, the spatial tracking of a movable useful signal source is thus possible. The direction of the useful signal source determined as a result of the method can be used in particular for aligning a directional cone during the input conversion of the acoustic system to the useful signal source in order to improve the signal-to-noise ratio of the useful signal source relative to the background noise by means of the directional cone.

An angular distance between two directivity characteristics adjacent to their central angle preferably corresponds to half the angular expansion. In particular, both adjacent directional characteristics have the same angular expansion. In the event that the individual directional characteristics are formed by directional cone whose sensitivity in the direction of the central angle is maximum, and decreases with increasing angular distance from the central angle, this means in particular that for each individual directional characteristic, an angle can be given for which the sensitivity of a test signal has dropped by a certain factor from the maximum value at the central angle, for example 6 dB or 10 dB. Such an angle is now assigned to the corresponding directional characteristic as a half angle widening, and the central angle of the adjacent directional characteristic is selected correspondingly at an angular distance of half an angular expansion. In the case in which notch-shaped attenuation of the sensitivity with a minimum at the central angle are selected as individual directivity characteristics, the same applies for the definition of the angular spread instead of the attenuation of the sensitivity to the maximum value at the central angle raising the sensitivity to the minimum value at the central angle is used. As a result, for a desired, wider angle range, a substantially complete coverage by the individual directional characteristics can be achieved, as a result of overlap of the individual directional characteristics to each next central angle a Nutzsignalquelle always at least one of the directional characteristics can be clearly assigned, whereby by the overlap and angular positions between two adjacent central angles are resolvable.

The central angle and the angular expansion of the directional characteristic are favorably determined by at least two conditions for the individual directional characteristics. In particular, the directional characteristics can each be formed according to the "linearily constrained minimum variance" method on the basis of the conditions. This makes it possible to align the conditions, for example in the form of relative attenuation of the sensitivity in a certain angular direction, and to be able to directly form the respective directivity characteristics.

It proves to be advantageous if the individual directional characteristics are respectively predetermined by a notch-shaped sensitivity characteristic which is determined by at least two conditions, so that the central angle and the angular expansion of the sensitivity characteristic are determined by the at least two conditions. Under a notch-shaped sensitivity characteristic is here to understand a directional characteristic, which has the maximum attenuation of sensitivity with respect to a test signal of given volume at the central angle, the sensitivity increasing with increasing angular distance from the central angle. The degree of this increase in sensitivity as a function of the angular distance to the central angle then defines the angle widening. Is a useful signal source in the direction of a central angle of such a directional characteristic, or within the angular resolution in the immediate vicinity of the central angle, ie within the "notch" of the sensitivity characteristic, the signal components of the useful signal are substantially attenuated by the directional characteristic, while signal components of other useful signal sources, which are outside the angular expansion about the central angle of said directional characteristic, largely maintained. This can now be used to determine a presence of a useful signal source in the region of the corresponding directional characteristic.

Expediently, an acoustic characteristic variable is formed from the signal components in each case for the individual directional characteristics, the acoustic characteristics of the directional characteristics being used to determine a useful signal in at least one of the directional characteristics. In this case, in particular, the maximum signal level over a suitable time window or the signal level averaged over the time window can be used as an acoustic parameter, and in particular only signal components in specific frequency bands can be used for the formation of the acoustic parameter.

Preferably, in this case the acoustic characteristic of the directional characteristic is compared with the total value of the corresponding characteristic for the first input signal and / or the second input signal, thereby forming a relative characteristic for the directional characteristic, wherein the presence of a useful signal in at least one of the relative characteristics of the directional characteristics the directional characteristics is determined. This means that in each case the acoustic characteristic, for example a time-averaged signal level, is first formed for the individual directional characteristics, and then the acoustic characteristic of each directional characteristic is normalized via a corresponding parameter which is derived from the first input signal and / or the second input signal. By normalizing the relative characteristic is then formed for each directional characteristic, which is ultimately used as a measure of the presence of a useful signal in the directional characteristic. As a parameter for normalization, for example, the reference level of the first input signal or the second input signal or the overall level of the first input signal and the second input signal can be used. By normalization can be achieved that in listening situations in which the respective sound level or the number of individual useful signal sources may change, the resulting changes in the overall level have no effect on the resolution of the localization of the useful signal sources.

In this case, the relative parameters are preferably compared with each other and / or with a predetermined limit value, and from this the presence of a useful signal in at least one of the directional characteristics is determined. The comparison of the relative characteristics with one another proves to be advantageous in that, depending on the nature of the directional characteristics, an extreme value (maximum or minimum) of a relative parameter can be used to directly deduce the presence of a useful signal in the corresponding directional characteristic. However, to localize a plurality of useful signal sources, it may also be advantageous to compare the relative characteristic quantities with a predetermined limit value whose exceeding or falling short of the presence of a useful signal source can be concluded.

It proves to be advantageous if a temporal mean value of the signal level is used as the characteristic variable, and from an attenuation, which experiences the temporal mean value of the signal level in a directional characteristic, standardized over the temporal mean value of the total level, by the respective sensitivity characteristic, the presence of a Useful signal is determined in the directional characteristics. The time window for averaging can depend in particular on the expected nature of the useful signal sources. If, for example, at least one of the useful signal sources is a conversation partner, then so the time window is preferably to be selected so that certain frequency ranges, eg format frequencies, are sufficiently strongly excited.

In an advantageous embodiment, a further input transducer generates a further input signal from the sound signal, wherein the directional characteristics are formed on the basis of the first input signal, the second input signal and the further input signal. Preferably, the further input transducer is spatially separated from the first input transducer and the second input transducer. By adding in the further input signal, the entire available phase information on the sound signal increases, so that a higher angular resolution can be achieved via the directional characteristics.

It proves to be particularly advantageous here if the individual directional characteristics are in each case given by a plurality of conditions which is equal to the number of input signals, so that at least the central angle and the angular widening are determined by the plurality of conditions. This means that, for example, in the case of four input signals from four spatially separate input transducers, the individual directional characteristics can be determined in each case by up to four conditions. This makes it possible to achieve particularly narrow angular widening and thus a particularly high angular resolution.

The invention further provides an acoustic system, comprising at least a first input transducer for generating a first input signal from a sound signal of the environment, a second input transducer, a second input transducer for generating a second input signal from the sound signal, and a signal processing unit, which for carrying out the method described above is set up. In particular, the acoustic system is designed as a hearing aid. The advantages mentioned for the method and its developments can be transferred analogously to the acoustic system.

Fig. 1

in einer Draufsicht eine Hörsituation für einen Benutzer eines binauralen Hörgerätes, in welcher sich die Anzahl seiner Gesprächspartner verändert,

Fig. 2

in einem Blockdiagramm der Ablauf eine Verfahrens zum Bestimmen einer Richtung einer Nutzsignalquelle,

Fig. 3

in einer Draufsicht eine Richtcharakteristik des Hörgerätes nach Fig. 1 mit einem gegebenen Zentralwinkel und einer gegebenen Winkelaufweitung, und

Fig. 4

gegen eine Zeitachse der Verlauf von relativen Kenngrößen zur Lokalisierung eines Sprechers für eine Hörsituation nach Fig. 1.

An embodiment of the invention will be explained in more detail with reference to a drawing. Here are shown schematically in each case:

Fig. 1

in a plan view of a hearing situation for a user of a binaural hearing aid, in which the number of interlocutors changes,

Fig. 2

in a block diagram the sequence of a method for determining a direction of a useful signal source,

Fig. 3

in a plan view of a directional characteristic of the hearing aid Fig. 1 with a given central angle and angular spread, and

Fig. 4

against a time axis, the course of relative characteristics for locating a speaker for a listening situation Fig. 1 ,

Corresponding parts and sizes are provided in all figures with the same reference numerals.

In FIG. 1 is shown schematically in a plan view of a listening situation 1. A user 2 of an acoustic system 4, which is embodied in the present case as a binaural hearing aid, is located in the left-hand representation of FIG. 1 in a conversation with a first conversation partner 6, which is positioned frontally to the user 2 with respect to its viewing direction, ie in FIG is an angle of 0 degrees to the user 2. The binaural hearing aid has localized the position of the interlocutor 6 for this hearing situation 1 as part of its resolution options. After some time of the conversation, a second conversation partner 8 is added, resulting in a new hearing situation 1 '. The second party 8 is positioned approximately at an angle of -45 degrees with respect to the viewing direction of the user 2. Is now in the binaural hearing aid, for example, for better amplification of the original signal, that of the first party 6, a focused on the first party 6 Directional characteristic formed, so conversation shares of the second party 8 are suppressed by such a directional characteristic or not sufficiently detected. In order to be able to perform a direction-dependent suppression of background noise in the changed hearing situation 1 ', and also to suppress conversational contributions of the second interlocutor 8 against those of the first interlocutor 6 is not as accurate as possible localization of the position of the second interlocutor 8 of considerable Advantage.

In FIG. 2 is a block diagram of the flow of a method 10 for determining a direction of a useful signal source by an acoustic system 4 is shown. In the present case, the useful signal source is given by the first call partner 6 or the second call partner 8 in one of the two listening situations 1, 1 ' FIG. 1 , The acoustic system 4 is formed in the present case by a binaural hearing aid. The binaural hearing device comprises a first local unit 12 and a second local unit 14, which are to be worn in the intended use of the binaural hearing device by the user 2 respectively on the left and on the right ear. The first local unit 12 or the second local unit 14 has a first input transducer 16 and a second input transducer 18, respectively, which generates a first input signal 20 and a second input signal 22 from an incoming sound signal of the environment. The first input transducer 16 and the second input transducer 18 are each given in the present case by a microphone. Possibly. can also be provided another input transducer 19, which generates a further input signal 23.

For a plurality of central angles aj, which cover the front hemisphere of the user 2 starting from 0 degrees in steps of 15 degrees in both directions, a first filter parameter F1 (aj) and a second filter parameter F2 (FIG. aj). The first filter parameter F1 (aj) and the second filter parameter F2 (aj) are designed such that they form a notch-shaped one with a convolution with the first input signal 20 and the second input signal 22, respectively, in a manner to be described Sensitivity characteristic 24 form. Thus, if the first input signal 20 and the second input signal 22 are respectively filtered with the first filter parameter F1 (aj) and the second filter parameter F2 (aj), the sum of the resulting signals for signal components whose signal source points in the direction of the central angle aj of the sensitivity characteristic 24 has a much lower sensitivity than for signals whose signal source is in another direction. For the signal 26 resulting from the described filtering, an acoustic characteristic 28 is now formed by averaging the signal level of the resulting signal 26 over a predetermined time interval. The acoustic parameter 28 formed from the averaged signal level is normalized via a reference level 30, and thus a relative characteristic 32 is formed. The reference level 30 is given in the present case by a time average of the level of the first input signal 16. Alternatively, as a reference level 30, a time average of the total level, ie the level of the sum of the first input signal 16 and the second input signal 18, are used. The relative core sizes 32 thus formed at each central angle aj are now compared with each other. In the comparison 34, each center angle aj whose sensitivity characteristic 24 has the relative characteristic 32 with the lowest value is now defined as the direction of a useful signal source.

In FIG. 3 is a plan view of a directional characteristic 34 with a central angle aj = -10 degrees and a predetermined expansion Δ shown. The directional characteristic is formed by a sensitivity characteristic 24 whose central angle aj is defined by the direction with the lowest sensitivity. By means of the concepts of "linearly constrained minimum variants" directional characteristics (LCMV), the central angle aj and the widening Δ can each be defined by a given attenuation of the signal components at the central angle aj itself and at another angle. In the present case, the angular spread Δ is determined by attenuating the signal level by 20 dB at a central angle aj of -10 degrees and attenuating the signal level by 6 dB in the frontal direction, ie at 0 degrees. As a result, a notch-shaped sensitivity characteristic 24 is defined, which for signals whose Source is in the range of the central angle aj, has a significantly lower sensitivity, and suppressed such signals accordingly.

Will now be in Fig. 2 illustrated method for a signal whose signal source is in the direction of the central angle aj = -10 degrees, so formed, for example, in a correspondingly positioned interlocutor, according to the sensitivity characteristic 24 of the time-averaged signal level, it has as acoustic characteristic 28 to the present signal source of all angle-dependent sensitivity characteristics 24 the lowest possible value. The acoustic characteristic 28 thus obtained is then normalized over the time-average overall level or the time-averaged level of one of the two input signals 20, 22. This leads to the fact that in the case of an absence of a speaker, ie a useful signal source, in the area of the central angle aj, all contributions to the standardization are detected essentially without attenuation in the acoustic characteristic 28. Only in the presence of a useful signal source in the region of the central angle aj, the contribution of the useful signal to the acoustic characteristic 28 is significantly suppressed by the sensitivity characteristic 24, whereas the contribution to standardization by the useful signal is retained. By the corresponding reduction in the relative characteristic 32, it is then possible to safely conclude the presence of a useful signal source in the direction of the central angle aj.

In FIG. 4 are shown against a time axis t to three different central angles respectively the relative characteristics 32a to 32c, which the in FIG. 2 Hearing situations 1, 1 'are assigned. By a time of about 4.5 seconds, the user 2 is only in conversation with the first party 6. The relative characteristics 32a through 32c, which are prescribed for sensitivity characteristics with central angles of -30 degrees (32a), -45 degrees (32b) and -60 degrees (32c) are formed, logically have no evidence for a useful signal source in the corresponding angular range. At about 4.5 seconds, the transition from listening situation 1 to listening situation 1 'takes place, the second conversation partner 8 joins in and in doing so pronounces conversational messages, which are now used as a sound signal in the processes described above received. By the contributions to the conversation of the second party 8 and the further presence of the first party 6 first changes the normalization for the three relative characteristics shown 32a to 32c. The most pronounced attenuation of the contributions of the second interlocutor 8 is expected to be due to the sensitivity characteristic with the central angle at -45 degrees, so that for this purpose also the relative characteristic 32b occupies the lowest value. For the two other central angles -30 degrees and -60 degrees, the speech characteristics of the second party 8 are still somewhat attenuated by the corresponding sensitivity characteristics. However, this attenuation is not as pronounced as in the sensitivity characteristic to -45 degrees. Accordingly, a drop in the value can be seen in the relative characteristics 32a and 32c, which, however, does not approach the reduction for -45 degrees (32b). From this it is now possible to close the second party 8 at about 4.5 seconds at an angle of -45 degrees.

Although the invention has been illustrated and described in detail by the preferred embodiment, the invention is not limited by this embodiment. Other variations can be deduced therefrom by those skilled in the art without departing from the scope of the invention.

LIST OF REFERENCE NUMBERS

1

listening situation

2

user

4

acoustic system

6

first interlocutor

8th

second interlocutor

10

method

12

first local unit

14

second local unit

16

first input converter

18

second input converter

19

further input converter

20

first input signal

22

second input signal

23

another input signal

24

sensitivity characteristic

26

resulting signal

28

acoustic characteristic

30

reference level

32

relative characteristic

32a-c

relative characteristic

34

directivity

B1, B2

condition

F1

first filter parameter

F2

second filter parameter

aj

central angle

Δ

widening

Claims (12)

Method (10) for determining at least one direction of a useful signal source (6, 8) by an acoustic system (4) comprising at least a first input transducer (16) and a second input transducer (18),
wherein the first input transducer (16) generates a first input signal (20) from a sound signal of the environment and the second input transducer (18) generates a second input signal (22) from the sound signal,
wherein on the basis of the first input signal (20) and the second input signal (22) a plurality of angle-dependent directional characteristics (34) having a respective fixed central angle (aj) and a respective given angular expansion (Δ) are formed,
wherein the signal components to the individual directional characteristics (34) are examined for the presence of a useful signal from a useful signal source (6, 8), and
wherein a corresponding in a specific directional characteristic (34) determined useful signal source (6, 8) of the corresponding central angle (aj) is assigned as the direction of the useful signal source (6, 8). Method (10) according to claim 1,
wherein an angular distance of two with respect to their central angle (aj) of adjacent directional characteristics (34) of half the angular expansion (Δ) corresponds. Method (10) according to claim 1 or claim 2,
wherein in each case the central angle (aj) and the angular expansion (Δ) of the directional characteristic (34) are predetermined by at least two conditions (B1, B2) for the individual directional characteristics (34). Method (10) according to claim 3,
wherein the individual directional characteristics (34) are each predetermined by a notch-shaped sensitivity characteristic (24) which is determined by at least two conditions (B1, B2), so that by the at least two Conditions (B1, B2) are respectively set the central angle (aj) and the angular expansion (Δ) of the sensitivity characteristic. Method (10) according to one of the preceding claims,
wherein an acoustic characteristic (28) is formed in each case from the signal components to the individual directional characteristics (34),
wherein the acoustic characteristics (28) of the directional characteristics (34) are used to determine a useful signal in at least one of the directional characteristics (34). Method (10) according to claim 5,
wherein the acoustic characteristic of the directional characteristic (34) is compared with the total value of the corresponding characteristic (30) for the first input signal (20) and / or the second input signal (22), and thereby a relative characteristic (32) for the directional characteristic (34 ) is formed,
and wherein from the relative characteristics (32) of the directional characteristics (34) the presence of a useful signal in at least one of the directional characteristics (34) is determined. Method (10) according to claim 6,
wherein the relative characteristics (32) are each compared with each other and / or with a predetermined limit, and from this the presence of a useful signal in at least one of the directional characteristics (34) is determined. Method (10) according to one of claims 4 to 7,
wherein as an acoustic characteristic (28) each a time average of the signal level is used, and
wherein the presence of a useful signal in the directional characteristic (34) is determined from an attenuation, which learns the time average of the signal level in a directional characteristic (34) normalized over the time average of the total level by the respective sensitivity characteristic (24). Method (10) according to one of the preceding claims,
wherein a further input transducer (19) from the sound signal generates a further input signal (23), and
wherein the directional characteristics (34) are formed from the first input signal (20), the second input signal (22) and the further input signal (23). Method (10) according to claim 9,
wherein the individual directivity characteristics (34) are respectively given by a plurality of conditions (B1, B2) equal to the number of input signals (20, 22), such that at least the central angle (aj) and the Angular expansion (Δ) of the sensitivity characteristic (24) are fixed. An acoustic system (4) comprising at least a first input transducer (16) for generating a first input signal (20) from a sound signal of the environment, a second input transducer (18) for generating a second input signal (22) from the sound signal, and a signal processing unit, which is set up to carry out the method (10) according to one of the preceding claims. Acoustic system (4) according to claim 11, which is designed as a hearing aid.

Images