EP1453348A1 - Self-calibration of microphone arrays - Google Patents

Abstract

The array microphone has individual microphones (1-4) connected with a signal processor (11) having digital filters. A loudspeaker (5) is placed in an acquisition area of the microphones. An electronic circuit applies a signal to the loudspeaker to emit a preset periodic noise signal. The processor evaluates the response signals from the microphones and the digital filters as a response to the reception of the noise signal. An independent claim is also included for a method for checking array microphones.

Description

The invention relates to array microphones. Array microphones are used in environments such as Cars, lectern, stage or the like used to sound sources and speakers, in short Signal to selectively record and suppress ambient noise. In the car Array microphones used on the one hand as a hands-free microphone for calls and on the other hand in systems such as Navigation systems using voice recognition to be served.

Array microphones consist of an arrangement of individual microphones that are used for signaling purposes are interconnected. In principle, the arrangement of the microphones distinguish between one, two and three-dimensional array microphones become. In a one-dimensional arrangement, the microphones are positioned along a line, e.g. a straight line or an arc. When using microphones with the orientation of the individual microphones is insignificant, since they only act as pressure receivers and therefore not directional in the room. Using The orientation of the individual microphones of gradient microphones is essential: the Overall directional characteristic and thus the entire bundling of the array microphone is created by combining the directional characteristics of the individual microphones together with the Application of the algorithm described below with which the microphone signals processed together.

There are two types of one-dimensional array microphones: Broadside Array microphones and Endfire array microphones. They differ in the direction of the preferred direction of sound incidence relative to the arrangement of the microphones: with Endfire Array microphones are the preferred longitudinal direction of sound incidence Microphones, i.e. for sound incidence directions with  = 0 degrees. For broadside array microphones the preferred direction of sound incidence is  = 90 degrees. The mutual The distances between the microphones can be constant or different from one another. In the second Different groups of microphones are used for different frequency ranges used for beamforming as described in [1].

The signal connection of the individual microphones can be on the analog or digital level. The following is the implementation in the digital area to be viewed as. The individual microphone signals are processed using analog-to-digital converters digitized and fed to a signal processing unit. By means of the Signal processing unit is a suitable algorithm (keyword "beamforming") applied to the microphone signals. With the help of this algorithm the The degree of concentration of the microphone is increased and lateral sound sources are suppressed. A good overview of array microphones can be found in [1] and in the one cited there Literature.

Part of the algorithm are filter coefficient sets that are used for the arrangement, type, Sensitivity and characteristics of the microphones used, the acoustic environment and the locations of the sound sources are characteristic. In these filter coefficient sets different properties of the individual microphones can be taken into account, such as those caused by production variations, aging effects etc. become. A frequently used filter structure is in the literature under "Filter and Sum Beamformer "(see e.g. [1], page 159) Microphone signals after analog-digital conversion with suitable FIR filters (finite Filter impulse response filters) and then added. An embodiment with 4th Microphones are shown in Fig. 1 reflecting the prior art:

1 shows a simple linear microphone array with equal distances d between the individual microphones. The sound incidence angle  is related to the longitudinal axis of the microphone array. The incident sound wave hits the individual microphones of the array with different transit times. The runtime differences correspond to the path differences d * cos (). The in Fig.1. FIR filters FIR ₁ to FIR _{4 shown} contain filter coefficient sets which correspond to frequency-dependent amplitude and phase differences. After filtering, the signals are added (filter and sum beamformer). Due to the aforementioned differences in amplitude and phase, sound waves that come from specific directions of incidence are amplified by constructive superimposition and sound waves from other sound incidence directions are weakened by destructive superimposition. As the simplest special case, the FIR filters FIR ₁ to FIR ₄ can be thought of as so-called all-pass filters, which all have the same frequency-independent delay. In this case, sound waves with an angle of incidence  = 90 degrees are amplified and sound waves from other directions of incidence are weakened, ie a so-called broadside array is present.

The filter coefficient sets mentioned above are fixed in many applications predefined standard situation is calculated and in operation of the array microphone as a constant Sizes used.

The checking of individual microphones in the array is currently carried out in such a way that during installation or in the event of service, the current consumption of the individual microphones is checked. The value the current consumption is checked to determine whether it is between two predetermined Limits come to lie. This allows you to check the basic functionality of the Detect single microphones. No more is happening.

Various problems arise, particularly, but not exclusively, in the car:

The first problem concerns the failure of a single microphone. That can be the degree of bundling of the entire array microphone and reduce the directional characteristic in unintentionally change. The user notices a deterioration from that Array microphone-controlled function without being able to localize the exact cause e.g. the speech recognizer suddenly works badly, when calling the speaker poorly understood.

In general, these deteriorations can have various causes that are not have to hang together with the array microphone. For example, the GSM transmission link used be disturbed while talking on the phone. It is therefore for fault diagnosis essential to know whether at least the array microphone is fully functional as a subsystem is. According to the state of the art, the current consumption of the microphone can only be in the laboratory or in the event of service.

The second problem is rather insidious: by scattering the properties of the Individual microphones in the course of production or different aging processes or different responses to changing environmental conditions the directional and frequency characteristics of the individual microphones differ greatly differ. This allows the above-mentioned algorithms for signal processing no longer work in the desired way.

In addition, by installing the array microphone, for example in a Vehicle cabin, the acoustic conditions compared to the laboratory at the Development, changed because of reflections, diffractions and interferences Multiple sound paths occur. This allows the directional characteristic of the array microphone disadvantageously changed and the degree of bundling can be reduced.

Similar changes in microphone characteristics occur when the number and Distribution of people in the vehicle change when a sunroof or window is open or is closed, etc.

The aim of the invention is to eliminate these problems, or at least theirs To significantly reduce effects without having to remove the array microphone or complicated and therefore expensive retrofitting is necessary.

According to the invention this goal is achieved in that at least one loudspeaker in the Detection range of the array microphone is provided, which is an acoustic test signal emits, and that the signals of the individual microphones from a signal processor (DSP) evaluated and their consistency in relation to the desired signal character and the wanted signal consistency can be checked.

The loudspeaker can either be permanently installed or part of a portable one Test device, the signal processor can be that of the array microphone or also Part of the test fixture. If multiple speakers are provided, next to the Control of the individual microphones and control of the beamforming are particularly accurate possible.

die Fig. 1 eine Prinzipskizze der Anordnung und Signalverbindung gemäß dem Stand der Technik,

die Fig. 2 ein erfindungsgemäßes Ausführungsbeispiel mit vier Mikrofonen,

die Fig. 3 eine Variante der Ausführungsform der Fig. 2,

die Fig. 4 ein Ausführungsbeispiel zur Messung der Lautsprecherimpedanz,

die Fig. 4a ein Schaltschema für ein Verfahren und

die Fig. 5 ein Ausführungsbeispiel des Verfahrensablaufes.

The invention is explained in more detail in the following description using an example. It shows

1 is a schematic diagram of the arrangement and signal connection according to the prior art,

2 shows an embodiment according to the invention with four microphones,

3 shows a variant of the embodiment of FIG. 2,

4 shows an exemplary embodiment for measuring the loudspeaker impedance,

4a is a circuit diagram for a method and

5 shows an embodiment of the process flow.

Fig. 2 shows an embodiment with 4 microphones 1 to 4. The distances of the Microphones 1-4 are the same in this embodiment. The speaker 5 is from acoustically recorded on all microphones, i.e. a signal that the speaker 5 emits recorded by all microphones. The microphones 1 to 4 can be used as pressure receivers as well as a gradient receiver.

Another exemplary embodiment is shown in FIG. 3. This is in principle the same as in FIG. 2 constructed, but all acoustic transducers are in a common housing 6 accommodated. This housing can also contain electronic components, A / D and D / A converter. Of the microphones 1-4, only the spoken openings are to see.

As explained in more detail below, the device according to the invention can be constructed be, the inventive method using the speaker and the signal processor is carried out, for example, as an acoustic self-test of the array microphone, can be as follows:

A calibration speaker 5 - is placed in, on or near the array microphone. preferably a small loudspeaker based on the dynamic principle - mounted, the acoustic Connects to the individual microphones 1-4 of the array in the sense that the Speaker signal can be recorded by any of the microphones. The optimal one There is space for the positioning of the (individual) calibration speaker in the middle of the Microphone arrangement where the sum of all paths calibration speaker microphone Minimum results. However, other speaker positions are also conceivable, e.g. on the edge of the array or somewhat removed, as in the illustrated embodiments. The Calibration speaker 5 is connected to an amplifier.

• Das Mikrofon ist eingeschaltet,
• Das Mikrofon hat die richtige Polung,
• Das Mikrofon hat die gewünschte Empfindlichkeit,
• Das Mikrofon weist den gewünschten Frequenzverlauf der Empfindlichkeit auf,
• Das Mikrofon weist keine zu großen Verzerrungen auf und
• Die Richtwirkung der Mikrofone

The aim of the self-test is in particular to check one or more of the parameters of the individual microphones listed below:

• The microphone is switched on
• The microphone has the correct polarity,
• The microphone has the desired sensitivity,
• The microphone has the desired frequency response curve,
• The microphone is not too distorted and
• The directionality of the microphones

Before starting the acoustic self-test, the calibration speaker is checked. there it is determined whether its electrical impedance is within predetermined limits. Only when this condition is met does the acoustic self-test of the microphones began. This speaker impedance check can be done by Loudspeaker signal is applied directly to one of the A / D converters (analog-digital converter). Fig. 4 shows an embodiment for measuring the speaker impedance Loudspeaker is operated in parallel with the input impedance of the A / D converter. should that Ratio of the speaker impedance to the input impedance of the A / D converter too far deviate from the value 1, there can be an additional series resistor in front of the loudspeaker be switched.

The loudspeaker impedance is measured using a technique known to the technician Method for measuring complex impedances. For example, a Constant current source placed on the speaker and the voltage on the Speaker terminals measured.

A method is described below as a preferred exemplary embodiment. The associated circuit diagram is shown in Fig. 4a. A signal is sent to the power amplifier 2 via the D / A converter 6. This power amplifier has a defined output impedance R _a . The amplified signal goes to the loudspeaker 8 with the impedance R _LS and further to the input of the A / D converter 9, which has a defined input impedance R _i . R _a and R _LS form a voltage divider. The voltage is measured on the A / D converter and compared with a reference measurement in which a known reference impedance is used instead of the loudspeaker as the impedance. The data of the reference measurement are determined only once and recorded in a non-volatile data memory (eg in a ROM). The unknown loudspeaker impedance R _LS can be determined from the two voltage values determined in this way. A measurement without loudspeaker can also be used as the reference measurement, ie the reference impedance is infinite ohms.

The microphone signals can be evaluated in various ways. As suitable Measurement signals can be sinusoidal signals, stochastic noise signals or periodic ones Noise signals such as Maximum sequence noise can be used. Some procedures are to be described as examples:

Method 1) In the simplest case, some sinusoidal signals with different frequencies spent in a row. The levels on the individual microphones will turn up checked their coherence, i.e. whether the measured voltages within preselected There are limits. The results are used to determine whether the microphone is functional or not.

Method 2) The loudspeaker sends a periodic noise signal, e.g. Maximum follow-noise out. By averaging the signal responses of the individual microphones the signal-to-noise ratio is improved. From the averaged microphone signal responses can by using the so-called Discrete Fourier Transformation (DFT) calculates the impulse responses of the respective speaker-microphone system become. This procedure is analogous to that from the literature ([2]: Vorländer, M .: Applications of maximum follow technique in acoustics. Progress in acoustics - DAGA 94, pp. 83-102) known methods for measuring speakers and microphones. The impulse responses measured in this way loudspeaker-microphone are checked whether you Maximum is within the pre-selected terms. The measured Amplitude transfer functions are checked whether they are within preselected Tolerance ranges lie. These amplitude transfer functions are a measure of that Microphone sensitivity. The comparison can be made with a reference measurement Change in microphone sensitivity, e.g. caused by aging or Determine environmental influences.

• Eine Ausführungsvariante der akustischen Selbstkalibrierung besteht darin, das Messsignal unhörbar für die Personen in der Nähe, z.B. für die PKW-Insassen auszusenden. Das Messsignal wird dabei im Audiobereich mit geringem Pegel ausgesendet. Durch Mittelung der aufgezeichneten Mikrofonsignale im Zeitbereich kann auch bei Signal-Rauschverhältnissen < 0 dB gemessen werden, ähnlich wie dies bei raumakustischen Messungen, z.B. in voll besetzten Konzertsälen, während der Vorstellung erfolgt. Erst durch die Mittelung der Signalantworten werden die korrelierten Signalanteile verstärkt und die nichtkorrelierten Hintergrundgeräusche eliminiert.
• Eine weitere Ausführungsvariante besteht darin, mehrere Kalibrierlautsprecher zu verwenden. Dadurch können die oben genannten Mikrofonparameter genauer gemessen und zusätzlich Informationen über die Richtwirkung der Mikrofone erhalten werden.
• Ein andere Ausführungsvariante der akustischen Selbstkalibrierung besteht darin, dass die Überprüfung der Arrays im Ultraschallbereich erfolgt, d.h. in einem für den Benutzer unhörbaren Frequenzbereich erfolgt. Die verwendeten akustischen Wandler müssen zu diesem Zweck zumindest in einem Teilfrequenzbereich, der über 20kHz liegt, genügend hohe Übertragungsfaktoren aufweisen.

The self-test is triggered, for example, by a control signal to the signal processing unit. From this a measurement signal is sent to the amplifier and on to the calibration speaker. This measurement signal is recorded by the individual microphones and then evaluated by an evaluation unit. The microphone parameters listed above can be taken from the recorded measurement signals.

• One variant of the acoustic self-calibration is to transmit the measurement signal inaudibly to the people nearby, for example to the car occupants. The measurement signal is emitted in the audio area at a low level. By averaging the recorded microphone signals in the time domain, measurements can also be carried out with signal-to-noise ratios <0 dB, similar to what happens with room acoustic measurements, for example in fully occupied concert halls, during the performance. Only by averaging the signal responses are the correlated signal components amplified and the uncorrelated background noise eliminated.
• Another variant is to use multiple calibration speakers. This allows the above-mentioned microphone parameters to be measured more precisely and additional information about the directional effect of the microphones can be obtained.
• Another embodiment variant of the acoustic self-calibration is that the arrays are checked in the ultrasound range, ie in a frequency range that is inaudible to the user. For this purpose, the acoustic transducers used must have sufficiently high transmission factors at least in a partial frequency range that is above 20 kHz.

Evaluation of the errors found

• Der Fehler wird im Fehlermanagementsystem des Fahrzeuges gespeichert. Beim nächsten Besuch einer Fachwerkstätte kann das defekte Mikrofonmodul getauscht werden.
• Der Fehler kann im Fahrzeug angezeigt werden beispielsweise in einer Systemkonsole, in einer Kontrollleuchte, in einem Pop-Up-Menü auf dem Bildschirm des Fahrzeugcomputers etc.
• Der Fehler kann im Fahrzeug akustisch gemeldet werden durch Ausgabe einer geeigneten Warnung über die Autolautsprecher oder den Kalbrierlautsprecher des Arraymikrofons.

The errors that may have been determined from the evaluation methods are preferably processed further in one or more of the following ways:

• The fault is saved in the fault management system of the vehicle. The next time you visit a specialist workshop, the defective microphone module can be replaced.
• The error can be displayed in the vehicle, for example in a system console, in a warning light, in a pop-up menu on the vehicle computer screen, etc.
• The error can be reported acoustically in the vehicle by issuing a suitable warning via the car speakers or the calibration speaker of the array microphone.

The method according to the invention has, apart from the possibility of detection a whole series of previously undetectable defects still have the advantage that the Measurement can be carried out while the microphone is in operation. To A successful display can, for example, be an automatic display "Microphone OK".

In addition, it is also possible to address the second problem area mentioned above To get: The acoustic self-test is carried out exactly as above described. Then the results of the recorded microphone signals are added used to recalculate and implement the above-mentioned coefficients.

The type of adaptation of the filter coefficients can take place, for example, in that the age-related change in microphone sensitivity determined according to the above procedure is taken into account when calculating the filter coefficient sets. Thereby changes in the microphone properties, especially the sensitivity frequency curve compensated. The process is shown in the block diagram in Fig. 5. shown.

The implementation of this adaptation is known to the person skilled in the field of electroacoustics knowledge of the invention possible without problems. It is preferred that self-test, Recalculation and implementation carried out at regular intervals become. This also enables an improvement in the microphone bundling, because it can react to changing environmental conditions, such as: opening or Closing windows, getting on or off people, changing the Microphone properties as a result of changes in environmental parameters such as Air temperature, air pressure or humidity, direct sunlight on a part of the array microphone with the resulting different heating of the Single microphones, etc.

Literature:

[1] M. Brandstein, D. Wards (Eds), Microphone Arrays, Springer Verlag, 2001

[2] Vorländer, M .: Applications of the maximal follow technique in acoustics. progress der Akustik - DAGA 94, pp. 83-102.

Claims (6)

Array microphone with several individual microphones (1 - 4), characterized in that a loudspeaker (5) is arranged in the detection area of each of the microphones and that an electronic circuit is provided which acts on the loudspeaker (5) in such a way that it emits a predetermined acoustic test signal , and that evaluates the test response signals coming from each of the microphones in response to the reception of the test signal. Method for testing array microphones, which consist of a plurality of individual microphones (1 - 4), characterized in that at least one common loudspeaker (5) is provided in the detection range of each individual microphone, which is connected to test electronics to which each individual microphone is also connected, that the test electronics emit a predetermined acoustic test signal via the loudspeaker, that the test electronics evaluate the test response signals that then come from each individual microphone and compare them with signal models that are stored in the test electronics or externally, and that the properly functioning individual microphones correspond, and that the test electronics depend displays and / or saves a corresponding message from the result of the comparison. Method according to Claim 2, characterized in that the test electronics carry out a check of the loudspeaker (5) before the acoustic test signal is emitted, the loudspeaker signal being applied directly to one of the A / D converters and the loudspeaker parallel to the input impedance of the A / D Converter is operated and the loudspeaker forms a voltage divider together with the output resistance of the power amplifier operating the loudspeaker, which is recorded and evaluated by the signal present at the A / D converter, by comparing this signal with a reference signal that results from the measurement with a reference impedance instead the speaker impedance comes from. A method according to claim 3, characterized in that the ratio of the speaker impedance to the input impedance of the A / D converter is checked and, if it deviates too far from the value 1, is adapted by an additional series resistor which is connected in front of the speaker. Method for automatically calibrating array microphones, which consist of a plurality of individual microphones (1-4), the microphone signals being digitized by means of analog-digital converters and fed to a signal processing unit which, using a suitable algorithm which is applied to the microphone signals, determines the degree of concentration of the Array microphones increased and lateral sound sources suppressed, filter coefficient sets, which are characteristic of the arrangement, type, sensitivity and characteristics of the microphones used, the acoustic environment and the locations of the sound sources, being part of the algorithm, characterized in that at least one loudspeaker (5) In the detection range of each individual microphone is provided, which is connected to test electronics, to which each individual microphone is also connected, that the test electronics emits a predetermined acoustic test signal via the loudspeaker (5) that the test electronics are there Thereupon, test signal responses coming from each individual microphone (1-4) are evaluated and compared with signal models which are stored in the test electronics or externally and which correspond to properly functioning individual microphones, and that the test electronics, depending on the result of the comparison, change the value of individual or all Performs filter coefficients of the filter coefficient set and carries out the test again until the test signal responses are in the range of the signal models. A method according to claim 6, characterized in that after performing a predetermined number of test repetitions, the test is terminated and an error message is displayed and / or saved.

Images