EP2590437A1 - Periodic adaptation of a feedback suppression device - Google Patents

Abstract

The method involves activating an adaptation procedure of the feedback suppression device. The adaptation procedure of the feedback suppression device is carried out, and the activation of the adaptation procedure is carried out periodically. The adaptive filter having a variable increment is adopted during the adaptation procedure. The on-state and off-state for different lengths of time is continued. An independent claim is included for a feedback suppression device for hearing device.

Description

The present invention relates to a method for adapting a feedback suppression device of a hearing device to a predetermined situation by activating an adaptation process of the feedback suppression device and performing the adaptation process of the feedback suppression device. Moreover, the present invention relates to a corresponding feedback suppression device. The term hearing device here is understood to mean any device which can be worn in or on the ear and triggers a stimulus, in particular a hearing device, a headset, 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 (ITE), 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 standard integrated into 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 present invention can be applied not only to hearing devices, but generally to audio systems that include at least one microphone for receiving ambient sound, subsequent signal processing of the microphone signal (e.g., gain), and output of the processed signal by a transducer (e.g. Speakers) to the environment. A hearing aid is, for example, such an audio system.

A very unpleasant condition of such an audio system is so-called "feedback whistling". In a known manner, namely acoustic feedback arises when the sound emitted by the transducer again enters through the microphone in the audio system and is amplified there again. This results in a closed loop (microphone-> gain-> transducer-> microphone, etc.), and feedback whistling occurs when the gain exceeds a certain threshold.

The unwanted whistling can be reduced or even eliminated by a feedback suppression device. In known feedback cancellation systems, an adaptive filter models the time-variant impulse response g of the acoustic feedback path. A common example of one Adaptation rule for updating filter coefficients h is a normalized least squares (NLMS) algorithm: $$\mathrm{H}\left(\mathrm{k}\mathrm{+}\mathrm{1}\right)\mathrm{=}\mathrm{H}\left(\mathrm{k}\right)\mathrm{+}\mathrm{\mu }\left[\left({\mathrm{e}}^{\mathrm{*}}\left(\mathrm{k}\right)\mathrm{x}\left(\mathrm{k}\right)\right)\mathrm{/}\left({\mathrm{x}}^{\mathrm{*}}\left(\mathrm{k}\right)\mathrm{x}\left(\mathrm{k}\right)\right)\right]\mathrm{,}$$

In this equation, k represents the discrete time index, x the input signal of the feedback cancellation system, e = mc an error signal defined by the difference of the microphone signal m and the feedback compensation signal c, μ the adaptation parameter step size parameter and * a conjugate complex operation. A block diagram for this is in FIG. 2 played. For the sake of simplicity, the representation of the time dependence (discrete time index k) is dispensed with. A desired signal s is picked up by a microphone 10. This results in the microphone signal m. A signal processor 11 further processes the signal. From the microphone signal m, a compensation signal c is subtracted, whereby an error signal e is obtained. This error signal e is supplied to a main processing unit 12 (eg, including a filter bank). The output signal x of the main processing unit 12 (SP) is supplied to a converter (eg loudspeaker 13) and to a feedback compensator 14. The feedback compensator 14 (FBC), together with its specific circuitry, constitutes a feedback compensation device. It has the transfer function h, which serves as an estimate of the acoustic path g from the loudspeaker 13 to the microphone 10. The feedback compensator 14 outputs the compensation signal c = h * x. In addition, the feedback compensator 14 is controlled by the error signal e.

The acoustic signal a reaches not only the eardrum of the user but is, as already indicated, fed back via the feedback path 15 to the microphone 10. This feedback path 15 has the said transfer function g.

Further information can be found for example in S. Haykin, Adaptive Filter Theory Englewood Cliffs, NJ: Prentice-Hall, 1996 , as in Toon van Waterschoot and Marc Moonen, "Fifty years of acoustic feedback control: state of the art and future challenges", Proc. IEEE, Vol. 99, No. 2, Feb. 2011, pages 288-327 ,

The parameter μ is also called step size. With it, the adaptation speed of a filter can be controlled. A suitable, time-dependent control of the step size μ is important for effective and stable feedback suppression. When μ is large, the filter quickly adapts to changes in the position of the acoustic feedback path g, thereby preventing feedback whistles. Otherwise, if the step size is constantly too high, mis-adaptation to tonal signals (eg, music) may result.

There are several concepts for controlling the step size μ in a suitable manner. However, compromises have to be made in all concepts. In the following two known concepts are presented:

a) Optimal step size estimation

By meeting some simplifications, a theoretical optimal step size μ is determined under certain conditions as follows: μ _opt ~ = E { | c | ² } / E {| e | ² },
where E {} means the expectation operator. The above formula helps to stabilize the adaptation, but it does not help to solve the above problem, to find a suitable step to avoid mis-adaptions. In practice, the prerequisites necessary for the derivation of the above estimate are not met (eg the assumption of uncorrelated receiver and microphone signals and the assumption that the unknown acoustic feedback path has the transfer function g = 0). Due to these defects, malfunctions arise.

b) Feedback triggered by feedback detector

In a normal mode, the adaptation is "frozen" in that the step size μ has a very small value. Adaptation is only allowed when the feedback detector becomes active, indicating a change in the acoustic feedback path (g). Consequently, the need for a new adaptation of the feedback suppression filter with the transfer function h is triggered by the temporal increase of the step size parameter μ. On the one hand, freezing the adaptation guarantees stability (no mis-adaptions for tonal excitation signals) unless the feedback detector is activated in error. On the other hand - and this is a massive drawback - the feedback suppression device only adapts when the feedback detector becomes active at all. In practice, it often happens that the acoustic feedback path (g) changes only slightly (eg when passing through a door or when sitting in a couch). This usually leads to no feedback detection. Since there is no adaptation, the instantaneous transfer function h of the feedback cancellation system does not represent the current acoustic path (g), resulting in an audible, rougher sound quality. The reduced soundness will continue until the feedback detector becomes active, which is usually accompanied by a feedback whistle. As a countermeasure, the sensitivity of the feedback detector could be adjusted so that even minor changes in the acoustic path are detected. However, this leads to increased erroneous feedback detections and thus to more artifacts.

Of course, both control methods for the step size μ can be used together. A corresponding scenario is in FIG. 3 shown. The block diagram shown is based essentially on that of FIG. 2 , Reference is made in this regard to the above description. In addition to the system of FIG. 2 Here, in the signal processing device 11, a feedback detector 16 (FD) is integrated, which receives the microphone signal m. Its output signal becomes one Step size control unit 17 (SWS) supplied with the step size μ is controlled in the adaptation of the feedback compensator 14. The step size μ is therefore not here as in the example of FIG. 2 controlled by the error signal e. As additional input variable, the feedback compensator 14 receives the error signal e as well as the step size control unit 17. The latter receives the compensation signal c of the feedback capacitor 14 as a further input variable.

However, the user of such a feedback suppression system must live with a compromise. It either tolerates harsh sound quality in undetected path changes, or it accepts false feedback detections with the risk of mis-adaptions, resulting in tonal glitches or other processing artifacts.

The object of the present invention is therefore to reduce artifacts in the automatic adaptation of feedback suppression devices.

According to the invention, this object is achieved by a method for adapting a feedback suppression device of a hearing device to a predetermined situation by activating an adaptation process of the feedback suppression device and carrying out the adaptation process of the feedback suppression device, wherein the activation of the adaptation process takes place periodically.

In addition, the invention provides a feedback suppression device for a hearing device with an adaptation device for adapting the feedback suppression device to a predetermined situation and an activation unit for activating the adaptation device, wherein the adaptation device with the activation device is periodically activatable.

Advantageously, therefore, the adaptation process of the feedback suppression device is activated periodically. It can thus be ensured that adaptations of the feedback suppression system take place even in the case of minor path changes in which the feedback detector does not respond. However, the adaptation is not always immediate, but at least in the foreseeable future.

Preferably, an adaptive filter with a variable step size is adapted in the adaptation process. This allows the adaptation to be carried out faster and less quickly as required.

In this case, the activation of the adaptation process can take place in that the step size is increased by leaps and bounds. The erratic increase leads to a short mismatch, which initiates an immediate readjustment.

The adaptation process can also be activated by a feedback detector in parallel or independently of the periodic activation if a feedback is detected by it. As a result, it can also be adapted as required during a predetermined trigger period.

Activation may be by a periodic binary enable signal representing an on state (eg, "high") and an off state (eg, "low"). In this case, the activation of the adaptation process can also be effected solely by the edge during a transition from "low" to "high".

The on state and the off state can each last different lengths. Ie. the binary enable signal need not be symmetric with respect to the length of the inputs and outputs.

According to a specific embodiment, the duration of the on state or the duration between two temporally successive states depending on a current one Hearing situation are changed. In particular, a current hearing situation can be represented by a classification result of a classifier. Depending on this, then a "high" state and / or a "low" state is changed in its duration.

The activation of the adaptation process can, however, also take place with a periodic activation signal which has more than two values, wherein the values are changed as a function of a current hearing situation. Ie. the activation signal can also have several states (multiple discrete states) or even have a steady course. Thus, so-called soft decisions for the initiation of the adaptation process are possible.

Moreover, by activating the adaptation process of the feedback suppression device, a frequency-shifting algorithm or a frequency-compression algorithm can be started. This further improves the stability of the feedback suppression.

FIG 1

den prinzipiellen Aufbau eines Hörgeräts gemäß dem Stand der Technik;

FIG 2

ein Blockdiagramm zu einer einfachen Rückkopplungsunterdrückung gemäß dem Stand der Technik;

FIG 3

ein Blockdiagramm zu einer Rückkopplungsunterdrückung mit Schrittweitensteuerung gemäß dem Stand der Technik; und

FIG 4

ein Blockdiagramm zu einer Rückkopplungsunterdrückung mit periodischer Aktivierung gemäß der vorliegenden Erfindung.

The present invention will be explained in more detail with reference to the accompanying drawings, in which:

FIG. 1

the basic structure of a hearing aid according to the prior art;

FIG. 2

a block diagram to a simple feedback suppression according to the prior art;

FIG. 3

a block diagram of feedback control with step size control according to the prior art; and

FIG. 4

a block diagram to a periodically activated feedback suppression according to the present invention.

The embodiments described in more detail below represent preferred embodiments of the present invention.

The feedback cancellation, as described below, can be used in any audio system and in particular in any hearing device, especially in hearing aids.

In the normal mode described above, the adaptation is frozen in known systems by the step size μ is chosen to be very small for the adaptation. According to the invention it is now proposed that the step size μ is periodically set to a higher or predetermined high value. This increase in the step size independent of the current acoustic situation leads to a spontaneous readjustment of the feedback suppression. This means that the feedback suppression is transferred from a frozen state to an adaptation state. This periodic triggering of the adaptation can be used in addition to or in parallel with existing step size control methods.

A schematic block diagram of a system with inventive feedback suppression device is shown in FIG FIG. 4 played. This figure also reproduces corresponding method steps for adapting the feedback suppression. The basis is the system of FIG. 3 , It is with regard to the system according to the invention thus expressly to the description of FIG. 3 respectively. FIG. 2 pointed. Same components have in FIG. 3 and in FIG. 4 The same reference numerals and they also exercise the same function, unless otherwise described.

FIG. 4 shows that in the signal processing device 11 of the hearing device, an additional activation device 18 is located. Its output signal is supplied to the step size controller 17. The activation device 18 is formed in the simplest case so that it is a periodic binary signal 19 with unchanged structure. This binary signal 19 has only two different states, namely an on state (eg "high") and an off state (eg "low"). As soon as the activation signal 19 is in the on state or on an edge from the off state to the on state, the step size μ in the step size control 17 for the feedback compensator 14 (jump) is increased significantly.

The activation device 18 can also be designed such that it generates other activation signals 20, 21. It can generate only one of these activation signals 19 to 21 or even several of them.

Representative of other activation signals here are the activation signal 20, in which the signal period is variable, and called the activation signal 21, which is not purely binary and can also assume intermediate values. Optionally, the activation device 18 is actuated by other components of the signal processing device 11 in order to change the activation signal to be output as a function of current signal processing variables. Such driving possibilities are in FIG. 4 not shown.

If in a specific embodiment, the activation signal (activation trigger) for the step size μ z. B. is "low" (OFF state), the feedback suppression device remains in the frozen state. On the other hand, if the activation signal is "high", the filter adapts to the feedback suppression device so that it adapts to the current acoustic feedback situation (g). When the periodic enable signal is in a "low" state and the feedback detector 16 becomes active, the step size μ is also adapted.

1. a) Die Zeitdauer des "low"-Zustands und die Zeitdauer des "high"-Zustands sind entweder gleich oder voneinander verschieden. Beispielsweise kann das Aktivierungssignal die Struktur besitzen: 1 Sekunde "low", 1 Sekunde "high", 1 Sekunde "low", 1 Sekunde "high", etc. Entsprechend einem anderen Beispiel besitzt das Aktivierungssignal die Struktur: 5 Sekunden "low", 1 Sekunde "high", 5 Sekunden "low", 1 Sekunde "high", 5 Sekunden "low", etc.
2. b) Die Zeitdauer zwischen zwei aufeinanderfolgenden "high"-Zuständen (Periodendauer) und die Zeitdauer des "high"-Zustands selbst kann während des Betriebs entweder fest oder variabel sein. Die variable Zeitdauer kann beispielsweise in Abhängigkeit von einer Entscheidung des Rückkopplungsdetektors und/oder in Abhängigkeit von einer Klassifikation der aktuellen Hörsituation ermittelt werden (vergleiche Aktivierungssignal 20).
3. c) Das periodische Aktivierungssignal kann nur einen "low"-und "high"-Zustand besitzen, was eine harte Entscheidung zwischen keiner Adaption und Adaption darstellt. Das periodische Aktivierungssignal kann aber auch so gestaltet sein (vergleiche Aktivierungssignal 21) dass eine sogenannte weiche Entscheidung möglich wird. Bei einer weichen Entscheidung erfolgt der Übergang der Schrittweite µ kontinuierlich, wodurch eine flexiblere und situationsangepasste Steuerung der Anpassgeschwindigkeit möglich wird.
4. d) Es besteht die Möglichkeit, die periodische Aktivierung mit anderen Stabilitätsmaßnahmen bei der Adaption zu kombinieren. Beispielsweise kann eine Frequenzverschiebung oder Frequenzkompression angeschaltet werden, wenn sich das periodische Trigger- bzw. Aktivierungssignal im "high"-Zustand befindet.

In the following several possibilities are shown how the adaptation of the feedback suppression can be activated or the step size μ can be increased.

1. a) The duration of the "low" state and the duration of the "high" state are either the same or different. For example, the activation signal may have the structure: 1 second "low", 1 second "high", 1 second "low", 1 second "high", etc. According to another example, the activation signal has the structure: 5 seconds "low", 1 second "high", 5 seconds "low", 1 second "high", 5 seconds "low", etc.
2. b) The time between two consecutive "high" states (period) and the duration of the "high" state itself may be either fixed or variable during operation. The variable time duration can be determined, for example, as a function of a decision of the feedback detector and / or as a function of a classification of the current hearing situation (compare activation signal 20).
3. c) The periodic activation signal can only have a "low" and "high" state, which is a hard decision between no adaptation and adaptation. However, the periodic activation signal can also be designed (compare activation signal 21) that a so-called soft decision is possible. In the case of a soft decision, the transition of the step width μ takes place continuously, as a result of which a more flexible and situation-adapted control of the adaptation speed becomes possible.
4. d) It is possible to combine the periodic activation with other stability measures during adaptation. For example, a frequency shift or frequency compression may be turned on when the periodic trigger signal is in the "high" state.

According to the invention, the step size control of the feedback suppression device is thus triggered periodically. Thus, the filter coefficients are renewed from time to time, which softens the frozen state somewhat. Thus, it is not absolutely necessary for the feedback detector to detect a corresponding feedback event. This has the advantage that during most of the time of a typical use, the periodic activation signal can be kept in a "low" state so that the adaptation is frozen. Thus, no processing or mismatch artifacts of the feedback suppression device arise.

With minor changes in the acoustic path, known step size controllers had not left the frozen state, resulting in harsh, metallic sound quality. This condition was maintained until a change in the acoustic path became so severe that a feedback detector struck or until the hearing aid wearer himself provoked feedback with his hand. Only then could the situation of the lack of adaptation be remedied. But this meant that a feedback whistle was inevitable for the situation change. With the periodic activation of the fitting process according to the invention, the time span in which the feedback suppression device dwells in the uncomfortable state is automatically restricted. With a correspondingly short period of the activation signal, this uncomfortable state is over before the wearer of the hearing device is even aware of the harsh sound quality. Moreover, the re-adaptation to the change of the acoustic path by the periodic activation does not cause any additional feedback whistle as before. The reduction in the frequency of feedback whistles thus increases the wearing comfort of hearing devices and confidence in a well-functioning instrument.

Claims (10)

A method for adapting a feedback suppression device of a hearing device to a predetermined situation Activating an adaptation process of the feedback suppression device and Performing the adaptation process of the feedback suppression device,
characterized in that - The activation of the adaptation process occurs periodically. The method of claim 1, wherein in the adaptation process, an adaptive filter is adapted with a variable step size. The method of claim 2, wherein the activation is effected by increasing the step size by leaps and bounds. Method according to one of the preceding claims, wherein the adaptation process is activated in parallel with the periodic activation of a feedback detector (16) when a feedback is detected by it. Method according to one of the preceding claims, wherein the activation is effected by a periodic, binary activation signal (19, 20, 21) representing an on state and an off state. The method of claim 5, wherein the on state and the off state persist for different lengths. Method according to Claim 5 or 6, wherein the duration of the on state or the time duration between two temporally successive individual states is changed as a function of a current auditory situation. Method according to one of the preceding claims, wherein the activation with a periodic activation signal (19, 20, 21) takes place, which has more than two values, and wherein the values are changed depending on a current hearing situation. Method according to one of the preceding claims, wherein upon activation of the adaptation process of the feedback suppression device, a frequency-shifting algorithm or a frequency-compression algorithm is started. Feedback suppression device for a hearing device with - An adaptation device for adapting the feedback suppression device to a predetermined situation and an activation unit (18) for activating the adaptation device,
characterized in that - The adaptation device with the activation device (18) is periodically activated.

Images