Hearing aid system and method of operating hearing aid system

Abstract

A hearing aid system (200) with improved adaptive feedback suppression based on improved feedback path variation measurements and a method (300) of operating such a hearing aid system.

Description

1. Field of the Invention

[0001] The present invention relates to hearing aid systems. The present invention more particularly relates to hearing aid systems that rely on adaptive feedback cancellation in order to reduce problems caused by acoustic feedback. The present invention also relates to methods of operating a hearing aid system. 2. Background technology

[0002] 2.1. What is a hearing aid system?

[0003] In the context of the present disclosure, a hearing aid may be understood as a small, battery-powered microelectronic device designed to be worn by a hearing-impaired user behind or inside the human ear. Before use, the hearing aid is adjusted by a hearing aid fitter according to a prescription. The prescription is based on a hearing test, producing a so-called audiogram of the individual hearing performance of the hearing-impaired user. The prescription aims to achieve a setting in which the hearing aid alleviates hearing loss by amplifying sounds within the audible frequency range of the user's hearing impairment. The hearing aid comprises one or more microphones, a battery, a microelectronic circuit and an acoustic output transducer, the microelectronic circuit comprising a signal processor adapted to provide amplification in those parts of the audible frequency range of the user's hearing impairment. The signal processor is preferably a digital signal processor. The hearing aid is enclosed in a housing adapted to be mounted behind or inside the human ear.

[0004] In the context of the present application, a hearing aid system may comprise a single hearing aid (a so-called monaural hearing aid system) or comprise two hearing aids, one for each ear of a hearing aid user (a so-called binaural hearing aid system). Furthermore, a hearing aid system may comprise an external device, such as a smartphone with a software application adapted to interact with the other devices of the hearing aid system. Therefore, in the context of the present application, the term "hearing aid system device" may denote either a hearing aid or an external device.

[0005] In general, a hearing aid system according to the invention is understood to mean any system which provides an output signal which can be perceived by a user as an acoustic signal or which contributes to the provision of such an output signal and which has means for compensating or contributing to compensating the user's individual hearing loss. These systems may include hearing aids which may be worn on the body or head, in particular on or in the ear, and which may be fully or partially implanted. However, some devices whose primary purpose is not to compensate for hearing loss may still be regarded as hearing aid systems, such as consumer electronic devices (such as headphones), as long as they have some provisions for at least partially alleviating an individual's hearing loss.

[0006] 2.2. What is acoustic feedback?

[0007] Acoustic feedback from the receiver to the microphone or microphones will limit the maximum amplification that can be applied in a hearing aid. Due to the feedback, the amplification in the hearing aid can lead to resonances, which shape the frequency spectrum of the hearing aid output in an undesirable way, and even worse, it can cause the hearing aid to become unstable, resulting in whistling or howling. Hearing aids often use compression to compensate for hearing loss; that is, the amplification gain decreases as the sound pressure increases. In addition, automatic gain control is often used at the output to limit the output level, thus avoiding clipping of the signal. In the case of instability, these compression effects will eventually stabilize the system slightly, thus producing a nearly constant level of whistling or howling.

[0008] Feedback suppression is commonly used in hearing aids to compensate for acoustic feedback. The acoustic feedback path can change dramatically over time due to, for example, the amount of ear wax, if the user wears a hat or holds a phone to their ear, or if the user is chewing or yawning. Therefore, an adaptive mechanism is often applied to feedback suppression to account for temporal variations.

[0009] 2.2.1. What is Adaptive Feedback Cancellation?

[0010] A widely used feedback suppression method in hearing aid systems is based on adaptive feedback cancellation. Figure 1 , which highly schematically illustrates a hearing aid 100 according to the prior art, comprising means for adaptive feedback cancellation.

[0011] The hearing aid 100 includes at least one acoustic-electric input transducer 101 providing an input signal 106 (which may also be represented as a microphone signal y(n) hereinafter), a digital signal processor 102 (which may also be represented as a hearing aid processor hereinafter) providing an output signal 111 (which may also be represented as a speaker signal u(n) hereinafter), an electroacoustic output transducer 103, an adaptive feedback filter 104 and an adaptive feedback estimator 105.

[0012] Acoustic feedback occurs when the microphone picks up a portion of the loudspeaker signal, forming an acoustic closed loop. When the amplitude of the signal traveling around the loop is not reduced with each round trip, and the feedback signal is added in phase with the microphone signal, the closed loop system becomes unstable. Feedback thus limits the maximum achievable stable gain, degrades the sound quality by creating distortion of the input signal, and causes howling when the system becomes unstable.

[0013] Feedback problems can be reduced by adaptive feedback cancellation techniques that attempt to model the feedback path response h(n) using an adaptive feedback estimator 105 and an adaptive feedback filter 104 and subtract the modeled feedback signal provided by the adaptive feedback filter 104 from the microphone signal 106 (y(n)). (which may be denoted as feedback cancellation signal 109 hereinafter).

[0014] exist Figure 1 In the example, the adaptive feedback filter 104 provides an estimate h(n) of the true acoustic feedback path response h(n). Ideally, And the feedback eliminates the signal will therefore be identical to the true feedback signal 108 (v(n). This means that the feedback cancellation signal is subtracted from the microphone signal 106 (y(n)) The residual signal 110 (e(n)) after the feedback operation will only contain the input signal 107 (x(n)) without feedback, that is, e(n)=x(n).

[0015] 2.2.2. Biased estimation problem

[0016] However, a common problem with adaptive feedback cancellation methods for acoustic feedback suppression is that With biased estimates of , adaptation will usually not converge to the optimal suppression of acoustic feedback.

[0017] It can be shown that the deviation is given by the cross-correlation vector E[u(n)x(n)] between the output signal 111 (u(n)) and the input signal 107 (x(n). Therefore, the correlation between the output signal 111 and the input signal 107 will cause The estimation of the feedback cancellation function may be biased, and thus the feedback cancellation performance may be reduced, and may cause the cancellation system to malfunction and howl to occur.

[0018] For most audio signals, the correlation between the input signal 107 and the output signal 111 is strong for shorter hearing aid signal processing delays, and becomes weaker as the hearing aid signal processing delay increases.

[0019] 2.2.3. Frequency shift as a solution to the biased estimation problem

[0020] Different techniques have been proposed to reduce the biased estimation problem.

[0021] One known technique applies decorrelation (e.g., pre-whitening techniques) in the adaptive estimation path. Generally, the advantage of performing decorrelation in the adaptive estimation path is that this does not modify the receiver signal, so that no sound quality degradation due to decorrelation occurs. However, this is often not sufficient to ensure good and reliable canceller performance. This is particularly the case when the input signal is tonal and therefore highly correlated, or when the hearing aid processing delay is low.

[0022] Alternatively (or additionally), the cancellation performance can be improved by decorrelating the loudspeaker signal 111 with respect to the input signal 107, for example by means of a frequency shift or a phase shift. However, this comes at the expense of increased delay in the main signal path and audible artifacts, these artifacts being caused in particular by interference between the directly transmitted sound (i.e. the ambient sound transmitted through the hearing aid and towards the eardrum) and the amplified and frequency-shifted sound provided by the hearing aid.

[0023] 2.2.4. Using probe noise as a solution to the biased estimation problem

[0024] Another technique for reducing the biased estimation problem is based on adding a detection noise signal to the main signal path.

[0025] EP-B1-1742509 describes a hearing aid system comprising a microphone unit adapted to convert said audio input signal into an electrical signal, a filter unit adapted to remove a selected frequency band of said electrical signal and to deliver a filtered signal, a synthesizer unit adapted to synthesize said selected frequency band of said electrical signal based on said filtered signal to generate a synthesized signal, a combiner unit adapted to combine said filtered signal and said synthesized signal to generate a combined signal, and an output unit adapted to convert said combined signal into an audio output signal. Such a hearing aid system is therefore limited in performance, since it requires that an entire frequency band be removed and then replaced by adding a synthesized signal.

[0026] 2.2.5. Problems to be solved by the present invention

[0027] Therefore, there is still a need to improve the performance of adaptive feedback cancellation systems, especially in achieving high performance adaptive feedback cancellation and high sound quality by avoiding audible artifacts.

[0028] It is therefore a feature of the present invention to provide a hearing aid system with improved adaptive feedback cancellation while maintaining high sound quality.

[0029] It is another feature of the present invention to provide a method of operating a hearing aid system that provides improved feedback cancellation while maintaining high sound quality. 3. Summary of the invention

[0030] The invention is set out in the accompanying claims. 4. Description of the drawings

[0031] Preferred embodiments of the present invention have been shown and described by way of example. As will be appreciated, the invention is capable of other and different embodiments, and its several details are capable of modifications in various obvious respects, all without departing from the invention. Accordingly, the drawings and description are to be regarded as illustrative in nature and not restrictive. In the drawings:

[0032] Figure 1 A hearing aid with adaptive feedback cancellation according to the prior art is illustrated highly schematically;

[0033] Figure 2 illustrates highly schematically a hearing aid system according to an embodiment of the invention; and

[0034] Figure 3 The method according to an embodiment of the invention is illustrated highly schematically. 5. Specific implementation methods

[0035] In the context of the present application, the term "feedback path change measure" may be used interchangeably with the shorter term "change measure", and similarly, the term "time / frequency / phase domain feedback path change measure" may be used interchangeably with the corresponding shorter term "time / frequency / phase domain change measure". Furthermore, terms including "change measure" should be interpreted as indicating a value, even though this may not be explicitly stated. Likewise, "feedback change" may be used interchangeably with "feedback path change", and, for example, the term "signal combiner" may be used interchangeably with "combiner". Finally, the term "change measure" may be abbreviated as CM.

[0036] 4.1 Hearing Aid System Implementation

[0037] Reference now Figure 2 , which highly schematically illustrates a hearing aid system 200 according to an embodiment of the present invention. Figure 2 In the embodiment shown, the hearing aid system 200 consists of only a single hearing aid, but in a binaural hearing aid system each of the two hearing aids would obviously have the same type of components in order to provide the functionality of the present invention.

[0038] The hearing aid system 200 comprises an acoustic-electric input transducer 201 that receives an audio input signal and converts the audio input signal into an input transducer output signal. The input transducer output signal is fed into a signal combiner 210. The combiner 210 is configured to subtract the feedback cancellation signal from the input transducer output signal and thereby generate a residual signal. The residual signal is then branched and provided to both an adaptive feedback estimator 205 and a digital signal processor (DSP) 202 that generates a processed residual signal.

[0039] 4.1.1 Decorrelation Unit 209

[0040] The processed residual signal is received by a decorrelation unit 209 which, when active, provides decorrelation with the processed residual signal. This can be achieved in a number of ways. According to one embodiment, this is achieved by applying a frequency shift to at least a portion of the received signal (i.e., the processed residual signal). According to another embodiment, decorrelation can be performed by adding a probe noise signal to the processed residual signal. The decorrelation unit 209 does not need to be enabled all the time, and it is generally advantageous to apply decorrelation only when needed to avoid feedback problems, as the decorrelation unit may provide undesirable acoustic artifacts when, for example, adding probe noise or frequency shifting the processed residual signal. According to this embodiment, the decorrelation unit 209 is at least partially controlled by the feedback path change measurement controller 208. It should be noted that if the decorrelation unit 209 is not enabled (also denoted as disabled hereinafter), it is transparent and the residual processed signal will be equal to the output transducer input signal received by the electroacoustic output transducer 203, which is configured to generate an acoustic (i.e., audio) output transducer output signal based thereon.

[0041] Therefore, according to one embodiment, the decorrelation unit 209 is adapted to frequency shift at least a part (frequency) of the processed residual signal so as to at least reduce the potential correlation between the output transducer input signal and the part of the input transducer output signal that does not include the feedback component (the sound received by the input transducer 210 and does not include the feedback component may be denoted as external sound in the following), thereby giving the adaptive feedback system including the adaptive feedback cancellation filter 204 and the adaptive feedback estimator 205 the best conditions for adapting to the current acoustic feedback path by reducing the deviation. The frequency shift may be a nonlinear transformation (however, the frequency shift does not need to be based on a nonlinear transformation), which would destroy the linearity of the feedback loop. This fundamentally improves the ability to distinguish between external sounds and feedback components, which is also referred to as reducing the deviation.

[0042] According to a specific embodiment, the frequency shift is achieved by applying a Hilbert transform, for example as described in European patent EP-B1_2671390, paragraphs

[35] -

[39] and references Figure 2 The public.

[0043] According to an alternative embodiment, the decorrelation unit 209 is adapted to add probe noise to the processed residual signal instead of providing a frequency shift.However, the desired effect of both decorrelation techniques is the same, namely reducing bias and thus improving the ability of the adaptive feedback system to estimate the feedback path.

[0044] According to yet other optional embodiments, the decorrelation unit 209 is adapted such that it can provide at least one of frequency shifting, phase shifting, probe noise, spectral replacement (wherein at least one complete frequency band (or frequency range) of the amplified signal is replaced by, for example, a known noise signal as opposed to replacing only a portion of the frequency band) and spectral replication, wherein some frequency bands of the signal to be decorrelated are replaced by different frequency bands of the same signal.

[0045] Therefore, according to a more specific embodiment, the decorrelation unit is adapted to provide at least two different forms of decorrelation depending on the situation. This is particularly advantageous when a large decorrelation is required.

[0046] 4.1.2CM Units 206 and 207

[0047] The main advantage of the time domain variation measurement unit 206 and the frequency domain variation measurement unit 207 is that they, at least in some cases, allow for the detection of unstable adaptive feedback cancellation before it is audible to the user (which would produce an undesirable, and very unpleasant, and possibly embarrassing howling from the hearing aid system). In conventional adaptive feedback cancellation systems, methods for detecting hearing aid howling are based on analyzing a signal representing the sound received by the hearing aid system, which makes it difficult to detect howling early in the build process and therefore before howling is audible. In contrast, by analyzing the filter characteristics of the adaptive feedback cancellation filter, it is more likely that feedback path changes will be detected before the associated sound artifacts become audible.

[0048] 4.1.3 Time Domain CM

[0049] According to one embodiment, the time-domain variation measurement unit 206 provides the time-domain feedback path variation measurement value as a time average value d(l) of the sign change of a single filter tap of the adaptive feedback filter using the following formula:

[0050]

[0051] where l is the tap index, p, n, and m are, respectively, the number of computational steps for increasing the tap value, the number of computational steps for decreasing the tap value, and the total number of computational steps for determining the time average d(l) of the sign transition.

[0052] According to an alternative embodiment, the time domain variation measure is provided as a time average d(N) of the sign changes of a plurality of N filter taps using the following formula: avg :

[0053]

[0054] Wherein N is the number of filter taps (preferably and according to this specific embodiment, it includes all filter taps of the adaptive feedback filter), l is the tap index, and p(l), n(l) and m(l) are respectively the number of calculation steps for increasing the tap value, the number of calculation steps for decreasing the tap value and the total number of calculation steps for determining the time average value d(l) of the sign change of a given single filter tap.

[0055] According to another alternative embodiment, the time domain variation measure is provided as the maximum of the time averages d(l) of the sign transitions of a plurality of individual filter taps. This embodiment is particularly advantageous in terms of early detection of feedback path variations (and thus enabling rapid response to variations) because detection of feedback path variations is more reliable / less noisy.

[0056] The time domain change measures given above are particularly advantageous because, as opposed to monitoring changes in filter tap values, they are based on monitoring the sign changes of (at least one) filter tap (i.e., monitoring the monotonicity of the change). This is because for a converging adaptive filter, the changes in filter tap values ​​for each tap will be around a static average. However, after the feedback path is changed, the adaptive filter will begin to adapt to the new feedback path. During the adaptation, the taps will show a systematic drift (drif). Depending on the update scheme of the adaptive feedback cancellation filter and the signal in a given situation, the step size of each tap will vary during the convergence process. In order to avoid a few large steps dominating, it is ideal to monitor the average sign change rather than the average change to improve the detection of feedback path changes.

[0057] 4.1.4 Single FD CM

[0058] According to one embodiment, the frequency domain variation measurement unit 207 provides a frequency domain feedback path variation measurement value, which may also be expressed as a frequency domain variation measurement value.

[0059] 4.1.4.1 Real part

[0060] According to a more specific embodiment, the frequency domain variation measurement value is determined as the maximum value of the time average value of the sign transformation of the real part of the frequency domain adaptive filter coefficient of the multiple real parts of the frequency domain adaptive filter coefficient k. Therefore, the maximum value can be determined as:

[0061]

[0062] It can be concluded that the frequency k 0 CM Max,FD,re (k 0 ) is determined to be around k 0 The width of the sliding frequency window △k is d re (k), and similar to the above time domain variation measurement d(l), d re (k) is given as follows:

[0063]

[0064] where k is the frequency domain filter coefficient index, and p re 、n re and m re are respectively the number of calculation steps for increasing the real part of the frequency domain filter coefficient, the number of calculation steps for decreasing said value and the time average value d for determining the sign change re (k) is the total number of computation steps.

[0065] 4.1.4.2 Imaginary part

[0066] According to a more specific embodiment, the frequency domain variation measurement value is determined as the maximum value of the time average value of the sign transformation of the imaginary part of the frequency domain adaptive filter coefficient of the multiple imaginary parts of the frequency domain adaptive filter coefficient k. Therefore, the maximum value can be determined as:

[0067]

[0068] It can be concluded that the frequency k 0 CM Max,FD,im (k 0 ) is determined to be around k 0 The width of the sliding frequency window △k is d im (k), where similar to the above time domain variation measurement d(l), d im (k) is given as follows:

[0069]

[0070] where k is the frequency domain filter coefficient index, and p im 、n im and m imare the number of calculation steps for increasing the imaginary part of the frequency domain filter coefficient, the number of calculation steps for decreasing the value, and the number of calculation steps for determining the time average value d of the sign change. im (k) is the total number of computation steps.

[0071] 4.1.4.3 Dual FD CM

[0072] According to another specific embodiment, the frequency domain variation measurement value is determined as the maximum value of the time average value of the sign transformation of the real part and the imaginary part of the frequency domain adaptive filter coefficient of a plurality of said real parts and imaginary parts of the frequency domain adaptive filter coefficient k. Therefore, the maximum value can be determined as:

[0073]

[0074] Where k, d re (k) and d im (k) has been given above.

[0075] Frequency domain variation measures are advantageous over time domain variation measures because they provide a frequency dependent variation measure, e.g., if the feedback path variation is limited to that frequency range, it may be possible to apply decorrelation only within a limited frequency range, thereby avoiding undesirable acoustic artifacts.

[0076] As is well known to those skilled in the art, the frequency domain adaptive filter coefficients are determined by performing a discrete Fourier transform on the time domain adaptive filter coefficients. Therefore, the frequency domain adaptive filter coefficients represent the frequency response of the adaptive filter.

[0077] 4.1.5 Phase Domain CM

[0078] According to one embodiment, the phase domain variation measure may be determined as a time average of the sign changes of the phase of the frequency response of the adaptive feedback cancellation filter for a given frequency. Thus, the phase domain variation measure is a special type of frequency domain variation measure.

[0079] As with the time domain filter taps, if the adaptive filter has converged, the phase changes in the frequency response at any given frequency are symmetrical about a static average phase. However, if the adaptive filter is not in a converged state, the filter's frequency response will show systematic drifts in the average phase, especially if there is drift in the impulse response. When the reflector moves (e.g., a hand moves around the ear), subtle drifts or lateral shifts in the impulse response can be observed in the feedback path, a common and difficult to detect scenario in standard adaptive feedback cancellation systems. Another difficult scenario arises when the earphones are repositioned slightly, meaning within 5mm. For these and similar scenarios, it is preferable to use phase domain change measurements.

[0080] Therefore, we now consider the update step size of the adaptive filter, which can be given as Among them, H k (f) is the frequency response of the adaptive feedback cancellation filter, where θ k is the phase change of the frequency response of the adaptive feedback cancellation filter with an update step size of k, and where the update step size is considered small, we can assume that

[0081] To eliminate the biasing effect of amplitude changes and observe only systematic changes in phase that are independent of step size, we can do this by observing The imaginary part of is positive or negative to monitor whether the phase changes positively or negatively.

[0082] More specifically, this is accomplished by first applying the unbiased average phase of the frequency response of the adaptive feedback cancellation filter to Determined as:

[0083]

[0084] Where R is the resultant length, v is the loop variance, and M is the number of calculation steps. Based on this, the time average of the sign change ρ(f) of the phase of the frequency response of the adaptive feedback cancellation filter can be determined based on the following expression:

[0085]

[0086] Where P(f) is the number of computational steps where the frequency response phase change is positive, N(f) is the number of computational steps where the frequency response phase change is negative, and M is the total number of computational steps used in the average. In the following, ρ(f) may also be denoted as a phase domain change measure. Phase domain change measures are particularly advantageous for detecting slight offsets in the feedback path, while time domain change measures are optimal for detecting more fundamental changes in the feedback path. Thus, the combination of time domain and phase domain change measures provides a combined feedback path change measure that is more suitable for detecting different types of feedback path changes.

[0087] Studying the behavior of R(f)ρ(f) it is apparent that patterns across frequencies emerge during changes in the feedback path, and in particular localized waves around zero are observed to reliably predict the changes.

[0088] 4.1.6CM controller 208

[0089] According to one embodiment, the feedback path variation measurement controller 208 is configured to control the adaptation speed of the adaptive feedback filter 204 (by providing an input to the adaptive feedback estimator 205) in addition to controlling the decorrelation unit 209 as described above, thereby enhancing the tracking of the feedback path and thus minimizing the risk of feedback howling. However, in principle, the feedback path variation measurement controller 208 can be configured to control only one of the two. More specifically, the amount of applied decorrelation is controlled based on the variation measurement, so that if a frequency-dependent variation measurement is monitored, the applied decorrelation can be limited to the frequency range in which the variation measurement detects a feedback path variation. In addition, the applied decorrelation can increase the adaptation speed of the adaptive feedback system, so that the adaptive feedback filter can converge faster and thus undesirable acoustic artifacts, such as howling, can be avoided.

[0090] It has been found that by using the change measure according to the invention, significant changes in the feedback path can be identified within 50 milliseconds (without too many false positives), which is usually sufficient to avoid acoustic artifacts.

[0091] Regardless, the feedback path variation measurement controller 208 operates based at least in part on receiving inputs including at least one of a time domain variation measurement value from the time domain variation measurement unit 206 and a frequency domain variation measurement value from the frequency domain variation measurement unit 207. However, it will be apparent to one skilled in the art that the feedback path variation measurement controller 208 may also be configured to additionally receive inputs directly from, for example, the adaptive feedback estimator 205. According to one embodiment, a measure of uncertainty in the feedback estimate is used to increase the variation measurement averaging length in accordance with the increased uncertainty measure.

[0092] It is also obvious to those skilled in the art that the time domain variation measurement unit 206 and the frequency domain variation measurement unit 207 need not be separate units, but may be, for example, integrated in the adaptive feedback estimator 205. Similarly, the feedback path variation measurement controller 208 may also be integrated in the adaptive feedback estimator 205.

[0093] 4.1.7 Combined CM

[0094] 4.1.7.1TD&FD CM

[0095] According to one embodiment, the feedback path variation measurement controller 208 receives inputs including time domain variation measurements from the time domain variation measurement unit 206 and frequency domain variation measurements from the frequency domain variation measurement unit 207 and provides combined time domain and frequency domain variation measurements based thereon.

[0096] A particular advantage of such a combined variation measure is that the frequency domain variation measure can be determined as a function of frequency (e.g., see Equation 7), and thus the combined variation measure can be used to control the amount of decorrelation and / or the adaptation speed of the feedback cancellation system to be applied as a function of frequency, while on the other hand maintaining the higher reliability provided by the less noisy time domain variation measure (e.g., by detecting feedback path changes with fewer false alarms).

[0097] Another advantage of this combined feature is that it can be used, for example, to reduce the number of samples (m) used to average the time to determine the frequency domain time measurement. re or m im ) reduces the amount of processing resources without significantly reducing the achieved performance.

[0098] According to an alternative embodiment, the frequency domain variation measure given by Equation 3 or Equation 5 instead of Equation 7 may be used to provide a frequency-dependent variation measure, and it should be noted that the selection of a particular frequency domain variation measure is independent of the rest of the feedback cancellation system according to the present invention.

[0099] 4.1.7.2TD&PD CM

[0100] Thus, the time domain and phase domain change measurements provide different methods of detecting changes in the feedback path. While the first measurement is specifically designed to detect fundamental changes in the feedback path, slight excursions of the feedback path can still be difficult to detect. For this purpose, the phase domain change measurement can be used. Due to the different nature of the two measurements, they can be advantageously combined into a single and more reliable measurement. A simple approach is to average the two measurements. However, other more complex schemes can also be used, such as converting the two change measurements into probabilities and treating the two measurements as independent sources. In combining these measurements, one can not only minimize the estimation noise, but also balance the detection to be sensitive to both fundamental changes and slight excursions in the feedback path, both of which are behaviors known to occur when hearing aids are used.

[0101] 4.1.8 Estimation of Noise

[0102] However, since a relatively short time averaging window is required in order to provide fast detection of feedback path changes, the feedback path change measurement is often affected by estimation noise, so that a countermeasure may be activated to avoid acoustic artifacts.

[0103] In order to mitigate the estimation noise, and according to a specific embodiment, the time domain variation measurement value CM TD and frequency domain variation measurement value CM FD Can be combined into smoothly varying measurements According to one embodiment, this may be done, for example, by determining the smooth variation measure using the following formula:

[0104]

[0105] Where n represents the time step, "a" is a smoothing factor, whose value is between 0 and 1. According to a more specific embodiment, for an update rate of 1 kHz, the value of the smoothing factor "a" is in the range between 0.95 and 0.99.

[0106] According to one embodiment, the time domain and frequency domain change measures may be combined by multiplying them together. The two change measures to be combined may be smooth or unsmooth. In the former case, we obtain a combined change measure

[0107]

[0108] This combined change measure provides significantly reduced estimation noise during periods of no feedback path change, and thus significantly reduces the noise floor. On the other hand, if only one of the two change measures detects a feedback path change, the change measure sensitivity is typically compromised.

[0109] Therefore, according to another embodiment, the combined change measurement value is determined as a weighted linear combination of smoothly varying measures (however, non-smoothly varying measures may alternatively be used):

[0110]

[0111] This combined change measure provides a reduction in the estimation noise (and the noise floor) which is not as good as the previous combined change measure, but on the other hand provides a higher sensitivity to feedback path changes, and this is independent of whether both or only one of the two change measures detects feedback changes. According to one embodiment, the value of b is 0.5, or in the range between 0.3 and 0.7.

[0112] According to another embodiment, the variation measures are combined by converting both into probabilities and treating the two probabilities as independent.

[0113] As described above, having realized the benefits that can be achieved by combining, for example, time domain variation measurements with frequency domain variation measurements, it is straightforward to conclude that similar benefits can be achieved by combining phase domain variation measurements with time domain or frequency domain variation measurements, or by combining all three types of variation measurements.

[0114] 4.1.9 AFC system

[0115] According to the invention, according to one embodiment, the determination of at least one variation measure may be advantageously used to control a robust adaptive feedback cancellation (AFC) system (i.e., having low bias due to the application of some form of decorrelation). Examples of such robust AFC systems have been given above, but include systems that use at least one of probe noise, decorrelation of the output signal in the form of a frequency or phase shift, spectrum replacement, and spectrum replication. Variation measures according to the invention may also be advantageously used by an AFC system based on a prediction error method (PEM), which may apply at least one of decorrelation and adaptive speed control of an adaptive feedback cancellation filter.

[0116] 4.2 Method Implementation

[0117] Reference now Figure 3 , which highly schematically illustrates a flow chart of a method 300 of operating a hearing aid system with adaptive feedback cancellation. The following steps of the method of the invention may be performed by a hearing aid system.

[0118] In a first step 301 of the method, at least one feedback path variation measure is determined based on at least one characteristic of an adaptive feedback cancellation filter.

[0119] In step 302, at least one of controlling an adaptation speed of the adaptive feedback cancellation filter and applying decorrelation is performed based on the at least one feedback path variation measure.

[0120] 4.3 Changes

[0121] It is generally noted that even if many features of the invention are disclosed in an embodiment including other features, this does not mean that these features necessarily need to be fully combined as described above.

[0122] At the most general level, note that different change measures according to the invention are independent of the specific actions taken by the adaptive feedback system based on the change measures. However, change measures according to the invention are particularly advantageous for actions that benefit from rapid detection of changes in the feedback path.

[0123] At the most general level, it is emphasized that the concept of monitoring the adaptive feedback cancellation filter and providing on this basis (at least one) variation measure suitable for quickly detecting variations in the feedback path is independent of the type of variation measure used. In other words, the variation measure may be based on the time domain or the frequency domain, wherein the frequency domain includes the phase domain or is based on multiple phase domains.

[0124] Thus, a plurality of variation measures may generally be combined in a number of different ways to provide a (combined) variation measure having improved properties. However, this is not essential, as the advantages of the invention may also be obtained using only a single variation measure.

[0125] One such combination consists of a time domain CM and a frequency domain CM. By combining these two types of variation measurements, the frequency-dependent combined CM has the high quality detection characteristics of the time domain CM, which cannot be achieved based on time domain CM alone because it is frequency independent. It is worth noting that this beneficial effect is generally independent of the method used to combine the time domain and frequency domain CM, i.e., whether they are combined based on, for example, averaging, linear weighting, or by multiplication.

[0126] Another such combination consists of time-domain CM and phase-domain CM, which is particularly advantageous because phase-domain CM is very good at detecting slight shifts in the feedback path, while time-domain CM is generally best at detecting more fundamental changes in the feedback path.

Claims

1. A method (300) of operating a hearing aid system with adaptive feedback cancellation, comprising the following steps: -a) determining at least one feedback path variation measure based on at least one characteristic of an adaptive feedback cancellation filter; -b) based on the at least one feedback path variation measure, performing at least one of the following operations: controlling an adaptation speed of the adaptive feedback cancellation filter and applying decorrelation.

2. The method according to claim 1, wherein step a) further comprises the following steps: -a-1) determining a first feedback path variation measurement value based on the time domain characteristics of the adaptive feedback cancellation filter, -a-2) determining a second feedback path variation measurement value based on the frequency domain characteristics of the adaptive feedback cancellation filter, and -a-3) combining the first and second feedback path variation measures and thereby providing an improved feedback path variation measure based on at least two characteristics of the adaptive feedback cancellation filter.

3. The method according to claim 2, further comprising the steps of: - Using feedback path variation measures based on frequency domain characteristics of said adaptive feedback cancellation filter to determine a frequency range over which decorrelation is applied.

4. The method according to claim 1, wherein: Decorrelation is applied using at least one of: adding a probe noise signal, providing spectral replacement, providing spectral replication, and providing decorrelation of the output signal in the form of a frequency shift or a phase shift of the output signal.

5. The method of claim 1, wherein step a) comprises the step of determining a first feedback path variation measure based on a time domain characteristic of the adaptive feedback cancellation filter by performing at least one of the following operations: - The first feedback path variation measure is determined as the time average d(l) of the sign transitions of a single filter tap using the following formula: where l is the tap index, p, n, and m are, respectively, the number of computational steps for increasing the tap value, the number of computational steps for decreasing the tap value, and the total number of computational steps for determining the time average d(l) of the sign transition; and The first feedback path variation measure is determined as the time average d(N) of the sign transitions of a plurality of N filter taps using the following formula: avg : wherein N is the number of filter taps, l is the tap index, and p(l), n(l), and m(l) are, respectively, the number of computational steps for increasing the tap value, the number of computational steps for decreasing the tap value, and the total number of computational steps for determining the time average d(l) of the sign transition for a given single filter tap; and - determining the first feedback path variation measure as the maximum value of the time average d(l) of the sign changes of a plurality of individual filter taps.

6. The method of claim 1, wherein step a) comprises the step of determining a first feedback path variation measure based on a frequency domain characteristic of the adaptive feedback cancellation filter by performing the following steps: - Determine the maximum value of the time average value of the sign transformation of the real parts of the frequency domain adaptive filter coefficients of a plurality of real parts of the frequency domain adaptive filter coefficients k using the following formula: It follows that the CM of frequency k0 Max,FD,re (k0) is determined as the d of a sliding frequency window of width Δk around k0 re (k) maximum value, where d re (k) is given as follows: where k is the frequency domain filter coefficient index, and p re 、n re and m re are respectively the number of calculation steps for increasing the real part of the frequency domain filter coefficient, the number of calculation steps for decreasing said value and the time average value d for determining the sign change re (k) the total number of computational steps; or - Determine the maximum of the time average of the sign transformation of the imaginary part of the frequency domain adaptive filter coefficients using the following formula: It follows that the CM of frequency k0 Max,FD,im (k0) is determined as the d of a sliding frequency window of width Δk around k0 im (k) maximum value, where d im (k) is given as follows: where k is the frequency domain filter coefficient index, and p im 、n im and m im are respectively the number of calculation steps for increasing the imaginary part of the frequency domain filter coefficient, the number of calculation steps for decreasing said value and the time average value d for determining the sign change im (k) the total number of computational steps; or The frequency domain variation measure is determined as the maximum of the time averages of the sign changes of the real and imaginary parts of the frequency domain adaptive filter coefficients k using the following formula:

7. The method according to claim 2, wherein: The step of combining the first and second feedback path variation measurements further comprises the steps of: -Use the following formula: CombCM(n)=CM TD (n)·CM FD (n) Where CombCM(n) represents the time domain variation measurement value CM Td (n) and the frequency domain variation measurement value CM FD (n) a combination of; or -Use the formula: CombCM(n)=b·CM TD (n)+(1-b)·CM FD (n) Where b is the time domain variation measurement value CM TD (n) and the frequency domain variation measurement value CM FD The weighting parameter of the linear combination of (n).

8. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to perform the method of any one of claims 1-7.

9. A hearing aid system (200) comprising a feedback path variation measurement unit (206, 207) and a variation measurement controller (208), wherein the feedback path variation measurement unit (206, 207) is adapted to determine a feedback path variation measure value based on at least one characteristic of an adaptive feedback cancellation filter, and wherein the variation measurement controller (208) is adapted to control at least one of an adaptation speed and an application of decorrelation of the adaptive feedback cancellation filter.

10. The hearing aid system (200) according to claim 9, wherein the feedback path variation measurement (206, 207) is further adapted to: - determining a first feedback path variation measure based on a time domain characteristic of the adaptive feedback cancellation filter, - determining a second feedback path variation measure based on the frequency domain characteristics of the adaptive feedback cancellation filter, and - combining the first feedback path variation measure and the second feedback path variation measure and thereby providing an improved feedback path variation measure based on at least two characteristics of an adaptive feedback cancellation filter.

Images