EP1648197B2 - Method and device for reducing the feedback in acoustic systems - Google Patents

Abstract

A feedback signal (RS) is detected in an incoming signal (ES), which is processed by relying on a detected feedback signal in an outgoing signal (AS), which is modulated (MO) so that the feedback signal is also modulated correspondingly. This modulation detects the feedback signal and has to be unheard by hearing-aid wearers. An independent claim is also included for a signal-processing device for an acoustic system.

Description

Method and apparatus for reducing feedback in an acoustic system

The present invention relates to a method of reducing feedback in an acoustic system by detecting a feedback signal in an input signal and processing the input signal in response to the detected feedback signal to produce an output signal. Moreover, the present invention relates to a corresponding signal processing device for an acoustic system. The acoustic system is, for example, a mobile radio device, a headset, a public address system and in particular a hearing aid or middle ear implant.

Acoustic feedback, hereafter referred to as feedback, is common in hearing aids, especially when it comes to high gain devices. These feedbacks are expressed in strong oscillations of a certain frequency and can be heard as whistling. This "whistling" is usually very uncomfortable for the hearing aid wearer himself as well as for people in his immediate vicinity. Feedback can be z. B. occur when sound that was recorded on the hearing aid microphone, amplified by a signal amplifier and output via the handset, gets back to the microphone and is reinforced again.

The simplest approach to reducing feedback is to permanently reduce hearing instrument gain so that the loop gain remains below the critical limit even in adverse situations. The decisive disadvantage, however, is that this limitation can no longer achieve the reinforcements required for severe deafness. Other approaches see a measurement of loop gain during the hearing aid fitting and reduce with the help of so-called notch filters (narrow-band notch filters) the gain targeted in the critical area. However, since the loop gains can change constantly in everyday life as described above, the utility is also limited.

• Die erste Klasse umfasst die so genannten KompensationsAlgorithmen, die mit Hilfe adaptiver Filter den Feedbackanteil im Mikrofonsignal schätzen und durch Subtraktion neutralisieren und somit die Hörgeräteverstärkung nicht beeinträchtigen. Allerdings setzen diese Kompensationsverfahren unkorrelierte, d. h. idealerweise weiße, Eingangssignale voraus. Tonale Eingangssignale, die immer eine hohe zeitliche Korrelation aufweisen, führen zu einer fehlerhaften Schätzung des Feedbackpfads, was dazu führen kann, dass irrtümlicherweise das tonale Eingangssignal selbst subtrahiert wird.

For the dynamic reduction of feedback, a number of adaptive algorithms have been proposed which automatically adjust to the respective feedback situation and take appropriate measures. These procedures can be roughly divided into two classes:

• The first class comprises the so-called compensation algorithms, which use adaptive filters to estimate the feedback component in the microphone signal and neutralize it by subtraction, thus not impairing hearing aid amplification. However, these compensation methods assume uncorrelated, ie ideally white, input signals. Tonal input signals, which always have a high temporal correlation, lead to an erroneous estimation of the feedback path, which can lead to the erroneous subtraction of the tonal input signal itself.

The second class contains the algorithms that only become active when feedback whistles are present. They generally include a feedback whistle detection mechanism that continuously monitors the microphone signal for feedback oscillation. If feedback-type oscillations are detected, the hearing aid gain at the corresponding point is reduced to such an extent that the loop gain drops below the critical limit. The gain reduction can, for. B. by lowering a frequency channel or by activating a suitable narrow-band notch filter (Notchfilter) done. The disadvantage is that the oscillation detectors in principle do not distinguish between tonal input signals and feedback whistles can. The result is that tonal input signals are held for feedback oscillations and then inadvertently lowered in level by the reduction mechanism (eg, notch filter).

In summary, it can be said that the functioning of all adaptive feedback reduction methods is impaired by input signals which have a tonal character characterized by dominant sinusoidal signal components (eg triangular tones, alarm signals). This often leads to unacceptable sound degradation of the input signal. This is where the present invention disclosure begins.

In the compensation algorithms decorrelating delay elements are often introduced into the signal processing chain in order to prevent that tonal signal sections are not significantly attacked with a length characteristic of speech signals. However, due to echo effects and irritations caused by desynchronized visual and auditory information, only millisecond delays are allowed. Therefore, for example, the reduction of music signals, which are often correlated over a much longer period of time, can not be avoided.

Another countermeasure is to slow down the adaptation of the filter so that all relevant environmental tonal signals are not attacked. However, this also has the consequence that the compensation filter can no longer follow rapid changes in the feedback path quickly, so that feedback whistles develop for a certain time, which only disappears again when the feedback path has stabilized and the filter has again been adapted with sufficient accuracy ,

The negative consequences of the misdetections of oscillation detectors are counteracted by the fact that the resulting gain reduction takes place only to a limited extent that z. B. erroneously held for feedback oscillations tonal useful signals (eg alarm signals) remain audible. However, this entails the risk that in the feedback case, the gain reduction is not sufficient to fall below the critical limit and the feedback whistling is not eliminated.

From the publication WO 2001/06746-A2 is a step size control of the compensation filter known, the feedback detector operates on the principle of bandwidth detection. If a narrow bandwidth of the input signal of the hearing device is detected by the bandwidth detector in the frequency band prone to feedback whistles, it is assumed that feedback whistling is present. A distinction of natural, narrow-band signals with spectral components in this frequency band, such. As music, but is not possible. In addition, the feedback whistle must represent a dominant signal component in order to be recognized.

Furthermore, from the document EP 1 052 881-A2 an oscillation detector for detecting feedback known. Again, the feedback whistles must be very clear to be recognized.

In the publication WO 2001/95578-A2 For example, a detection of feedback whistles is described by estimating the variance of the frequency estimate of the hearing aid input signal. This method also has the above-mentioned disadvantages.

Furthermore, in the document DE 199 04 538-C1 the optional damping of individual frequency bands proposed. Frequency bands in which there is feedback whistling, through a damping element introduced a stronger attenuation than would be expected for useful signals. The engagement in the forward signal path may be audible to the hearing aid wearer and is likely to occur Slow detection takes place, since the bands are ideally examined sequentially.

Another method for reducing feedback in acoustic systems is from the document US 6,347,148 B1 known. In the process, the spectrum of an input signal is estimated and a control signal is generated on the basis of a psychoacoustic model. The control signal is used to drive a noise source, with which a non-audible noise signal in response to the noise signal can be generated. In addition, the possibility is described there to impose short noise signals of predetermined duration on the output signal. The noise signals in the input signal reduce feedback signals.

The publication US 5,748,751 A describes echo reduction by decorrelating the input signal and the output signal. The decorrelation takes place with a phase modulator, which is operated with a certain frequency.

Furthermore, the document shows US 5,412,734 and WO 00/44113 A1 a feedback reduction using AM, FM or QPSK. The correspondingly modulated signal is added to the basic signal and detected again during feedback.

The object of the present invention is thus to further improve the reduction of feedback of a hearing aid.

According to the invention, this object is achieved by a method according to claim 1 and a signal processing device according to claim 7.

The underlying idea is to impose imperceptible features on the output signal of the acoustic system and in particular the hearing aid for the hearing aid wearer. This allows, by appropriate analysis of the input signal too determine whether the input signal is feedback or a "normal" external input signal (useful signal). The determination of the characteristic of the feature in the input signal also allows conclusions to be drawn about the corresponding proportions of feedback and useful signal. This can then be used directly to control feedback reduction algorithms.

In an advantageous manner, it is thus possible during operation to determine continuously and absolutely inconspicuously or inaudibly the extent to which feedback signals are present at a microphone or at the hearing device microphone, whereby the control and operation of the known feedback reduction algorithms can be significantly improved.

The processing of the input signal preferably takes place with an adaptable filter whose adaptation speed and / or strength depends on the quantity of the detected feedback signal. In particular, it is advantageous if the adaptation speed increases in proportion to the quantity of the detected feedback signal. If, for example, the feature analysis of the input signal is negative, d. H. it contains no feedback signal, so the adaptation speed of the compensation filter mentioned above can be slowed down so that the filter is not adjusted by tonal input signals and they are not attacked. If, on the other hand, the feature is detected in the input signal, the effectiveness and / or speed of the feedback compensator is set to the value at which feedback is optimally suppressed.

In the case of detection of a feedback signal, at least one notch filter may be activated for processing the input signal.

The phase modulation shows no particular susceptibility to misdetection in narrowband signals.

A feedback situation can already be detected before there is a dominant expression of the feedback whistle in the signal mixture.

The detection of feedback can be performed separately in several subbands. As a result, the gain, but also the reduction of feedback in the individual subbands can be adjusted individually.

A closed loop in the signal processing device can be used for a signal modification. In this case, the modulated signal passes through the loop several times, so that the corresponding signal modification is caused.

• FIG 1 ein Hörgerätesystem gemäß dem Stand der Technik;
• FIG 2 ein Hörgerätesystem gemäß einer ersten Ausführungsform der vorliegenden Erfindung;
• FIG 3 ein Hörgerätesystem gemäß einer zweiten Ausführungsform der vorliegenden Erfindung; und
• FIG 4 einen Rückkopplungsdetektor mit Filterbank.

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

• FIG. 1 a hearing aid system according to the prior art;
• FIG. 2 a hearing aid system according to a first embodiment of the present invention;
• FIG. 3 a hearing aid system according to a second embodiment of the present invention; and
• FIG. 4 a feedback detector with filter bank.

The embodiments described in more detail below represent preferred embodiments of the present invention. For a better understanding of the invention, reference will first be made to FIG FIG. 1 the prior art explained in more detail.

FIG. 1 shows a hearing aid HG, whose input forms a microphone M. The recorded signal is applied as an input signal ES a processing unit V forwarded. There it is processed and possibly reinforced. The resulting output signal AS is delivered to a handset H. Via a feedback path RP, the output signal of the handset H is fed back to the microphone M. With open supply, there is primarily an acoustic feedback path. In general, however, electromagnetic, electrical, magnetic and other feedback are also conceivable. The feedback signal RS resulting from the feedback path is added to a useful signal NS, and the sum signal is picked up by the microphone M.

The signal path from the microphone M via the hearing aid processing V, the handset H, the feedback path RP back to the microphone M represents a loop. H. the gain experienced by a signal passing through this loop is at least 1.0 at least at one frequency, and when the phase condition is met, feedback whistling occurs. Even if the loop gain is just below this limit, audible feedback effects, e.g. B. sound changes on.

One successful method for suppressing the feedback effects is the digital replica of the feedback path RP. This is simulated by an adaptive filter AF, which is fed by the output signal of the processing unit V. A corresponding compensation signal KS, which originates from the compensating, adaptive filter AF, is subtracted from the input signal ES of the microphone M and the resulting difference signal is fed to the processing unit V.

There are thus two paths, on the one hand the outer feedback path RP and on the other hand the adaptive filter AF simulated digital compensation path. The resulting signals of both paths are subtracted from each other at the input of the device as shown in FIG FIG. 1 through the two addition units is shown. Ideally, the effect of the external feedback path RP is thereby canceled.

An important component in the adaptive algorithm for determining the feedback path is its step size control. It indicates the speed with which the adaptive compensation filter adapts to the outer feedback path RP. Since there is no meaningful compromise for a fixed step size, it must be adapted to the current situation in which the system is located.

In principle, a large step size for a fast adaptation of the adaptive compensation filter AF to the outer feedback path RP is to be aimed for. A disadvantage of a large step size, however, is the generation of perceptible signal artifacts.

In the event that there is no feedback situation, the step size should be vanishingly small. In this case, the situation in which the loop gain is just below 1 or greater than / equal to 1 and the phase condition is fulfilled at least at one frequency is referred to as the feedback situation. If, on the other hand, a feedback situation occurs, the step size should or will be large. This ensures that the algorithm adapts the adaptive compensation filter AF only if it differs significantly in its characteristic from the characteristic of the feedback path RP, ie. H. if there is a need for post-adaptation. For this purpose, a feedback detector is provided.

In order to be able to reliably detect a feedback, a modulation device MO is provided according to the invention FIG. 2 between the processing unit V and the handset H is connected. It modulates the output signal AS to a modulated output signal AS '. The modulation of the output signal AS is not perceptible. In the event of a feedback situation, a significant proportion of the sound signal emitted by the receiver H returns to the microphone M and is recorded in the device together with the ambient signal.

In FIG. 2 is indicated that the feedback path RP can basically be designed arbitrarily. Ie. it does not have an acoustic feedback signal RS, as in FIG. 1 is indicated, present, which is added with an acoustic useful signal NS before the microphone M. Rather, the feedback in the microphone M can also be done for example via structure-borne noise or electromagnetic coupling.

The input signal ES of the microphone M is analyzed by a feedback detector RD. Thus, the feedback signal RS can be detected due to its modulation. A downstream controller S drives the adaptive compensation filter AF according to the detection result of the feedback detector RD. As a result, for example, the adaptation speed of the adaptive filter AF is changed.

The embodiment of FIG. 3 corresponds essentially to that of FIG. 2 , Here is the feedback path as in the example of FIG. 1 purely acoustic nature, so that the feedback signal is added to the useful signal in front of the microphone M.

Another difference to the circuit of FIG. 2 is that the signal for the feedback detector RD is not picked up immediately after the microphone M, but after the subtraction of the compensation signal of the adaptive filter AF at the point A. The strength of the modulation of the signal at the point A is an image of the difference between the effect of the feedback path RP and the effect of the adaptive compensation filter AF. An essential difference to the embodiment according to FIG. 2 in which the signal to be analyzed is tapped immediately behind the microphone M, however, does not exist.

In addition, in FIG. 3 indicated that in the feedback detector RD a step size control can be integrated, so that can be dispensed with a separate control module. The remaining components of the embodiment of FIG. 3 correspond to those of the embodiment of FIG. 2 , In this respect, therefore, the description to FIG. 2 directed.

In the embodiment according to FIG. 3 the phase of the output signal AS is modulated since the human ear is largely insensitive to phase changes. In a concrete example, the phase of the output signal AS with a specific frequency, referred to here as the modulation frequency f_mod, is linearly rotated back and forth between two phase values. For example, the phase values are α and α + π / 2, where α is any solid phase. In the feedback situation, a detectable tremolo component with a frequency of f_mod is formed in the signal loop.

The tremolo component can be detected by means of a frequency demodulator in the feedback detector RD. It is advantageous to construct the feedback detector RD with a filter bank, as in FIG. 4 is shown, the z. B. the input signal ES with multiple bandpasses BP1, BP2, ..., BPn divided into subbands. After each bandpass an analysis unit AE and a threshold value SW is arranged in each case. The output signals of the signal paths for each subband are optionally supplied to an OR gate OR. The respective analysis units AE and threshold SW can be identical to each other. Thus, the analysis in this example is done in each subband path in the same way. If the analysis result in a band exceeds a certain threshold, the associated threshold switch SW responds, ie a feedback situation is detected for this band.

This information can be used for an adaptive compensation filter AF, which is adapted in subbands for step size control. If, on the other hand, an adaptive filter AF is used in the entire band, the results of the subband detections must be combined by means of a logical OR operation into an overall band detection statement. Even the special case that the entire band is analyzed uniformly, where n = 1, leads to a functional system. However, the error detection rate is lower for a larger n, e.g. B. n = 16.

The step size control of the adaptive filter AF, in addition to the simple threshold decision according to FIG. 4 According to which only the presence or absence of feedback is detected, also differentiated occur. For example, the step size can be determined by proportional conversion of the estimated strength of the signal modulation at point A. This can also be done again via a subband approach. The greater the detected signal modification, the higher would be the need for a post-adaptation, ie the higher the necessary step size would have to be selected. The step size can thus be continuously adapted to the signal modulation. In a pure threshold decision, however, the step size is set high for a certain fixed time or for the time frame in which feedback is detected. Otherwise, it takes on a small value.

Claims (12)

1. Method for reducing feedback in an audio system (HG) by

   - detecting a feedback signal (RS) in an input signal (ES),

   - processing the input signal (ES) to produce an output signal (AS) on the basis of the detected feedback signal (RS) and

   - modulating (MO) the output signal (AS) by impressing a feature onto the output signal (AS), so that the feedback signal (RS) is also correspondingly modulated,
   where

   - the feedback signal (RS) is detected on the basis that the impressed feature is present
   characterized in that

   - the modulation (MO) is effected by phase modulation and this involves the phase being shifted forward and back at a predetermined frequency between two phase values.
2. Method according to Claim 1, where the input signal (ES) is processed using an adaptable filter (AF) whose adaptation speed and/or level of action is dependent on the quantity of the detected feedback signal (RS).
3. Method according to Claim 2, where the adaptation speed and/or level of action rises in proportion to the quantity of the detected feedback signal (RS).
4. Method according to one of the preceding claims, where if a feedback signal is detected then at least one notch filter for processing the input signal (ES) is activated.
5. Method according to one of the preceding claims, where the detection is performed separately in a plurality of subbands.
6. Method according to Claim 5, where the adaptive filter (AF) is individually adapted in the subbands.
7. Signal processing apparatus for an audio system (HG) having

   - a processing device (V, AF) for producing an output signal (AS) from an input signal (ES) by taking into account a feedback signal (RS),

   - a modulation device for modulating (MO) the output signal (AS) by impressing a feature onto the output signal (AS), so that feedback results in a correspondingly modulated feedback signal (RS), and

   - a detection device (RD) for detecting the modulated feedback signal (RS), where

   - the feedback signal can be detected by the detection device on the basis that the impressed feature is present, characterized in that

   - the output signal (AS) can be modulated with the modulation device (MO) by phase modulation
   and

   - this involves the phase being able to be shifted forward and back at a predetermined frequency between two phase values.
8. Signal processing apparatus according to Claim 7, where the processing device (V, AF) has an adaptable filter (AF) whose adaptation speed and/or level of action is dependent on the quantity of the feedback signal (RS).
9. Signal processing apparatus according to Claim 8, where the adaptation speed and/or level of action rises in proportion to the quantity of the feedback signal (RS).
10. Signal processing apparatus according to one of Claims 7 to 9, where the processing device can be used to activate at least one notch filter if a feedback signal is detected.
11. Signal processing apparatus according to one of Claims 7 to 10, which has a respective detection device (RD) for a plurality of subbands.
12. Signal processing apparatus according to one of Claims 7 to 11, which utilizes a closed loop to produce a signal modification.

Images