JP6019098B2 - Feedback suppression - Google Patents

Description

  A novel method for performing adaptive feedback suppression in a hearing aid and a hearing aid utilizing the method are provided. According to that method, feedback suppression is performed using a slow adaptive filter that models a gradual change in the feedback path and a fast adaptive filter that models a sudden change in the feedback path.

  In the hearing aid, an acoustic signal that reaches the microphone of the hearing aid is amplified and output by a small speaker in order to restore audibility. Since the distance between the microphone and the speaker is short, it may cause feedback. Feedback is generated by repetitive amplification in which a portion of the amplified acoustic output signal propagates back to the microphone. When the feedback signal exceeds the level of the original signal at the microphone, the feedback loop becomes unstable and typically causes audible distortion or howling. One way to stop feedback is to reduce the gain.

  The risk of feedback is that it limits the maximum gain available in the hearing aid.

  It is well known to use feedback suppression in hearing aids. With feedback suppression, the feedback signal that reaches the microphone is suppressed by subtracting the feedback model signal from the microphone signal. The feedback model signal is generated by a digital feedback suppression circuit configured to model a feedback propagation path, i.e., a feedback propagation path along which the hearing aid output signal propagates back to the hearing aid input causing repetitive amplification. Provided. The transfer function of the receiver (in the field of hearing aids, the hearing aid speaker is usually referred to as the receiver) and the transfer function of the microphone are included in the feedback propagation path model.

  Typically, the digital feedback suppression circuit includes one or more digital adaptive filters for modeling the feedback path. The output of the feedback suppression circuit is subtracted from the microphone audio signal to remove the feedback signal portion of the audio signal.

  In a hearing aid with more than one microphone, for example with a directional microphone system, the hearing aid may include separate digital feedback suppression circuits for individual microphones and groups of microphones.

  WO 99/26453 provides a useful discussion of feedback suppression methods in hearing aids.

  WO 99/26453 discloses feedback suppression using two adaptive filters connected in series (see FIG. 1).

  The first filter is adapted during fitting of the hearing aid to the intended user and / or when the hearing aid is activated in the ear. This filter adapts quickly using the white noise probe signal, and then the filter coefficients are fixed. That is, it is fixed during normal operation of the hearing aid. The first filter operates as a fixed filter.

  The first filter models the portion of the hearing aid feedback path that is assumed to be approximately constant during use of the hearing aid, such as the microphone, the amplifier and receiver resonances driving the receiver, and the fundamental acoustic feedback path.

  The second filter adapts during use of the hearing aid but does not use a separate probe signal. This filter provides a quick correction to the feedback suppression circuit when the hearing aid becomes unstable, and the feedback path that occurs in everyday use, such as caused by chewing, sneezing or using a telephone handset Keep track of the disturbances.

  The series connection of fixed and adaptive filters provides a good tradeoff between speed and accuracy. A single long filter tends to be slow and / or inaccurate. Furthermore, the fixed filter is an IIR filter with relatively low processor requirements.

  In practice, however, the filter coefficient of the fixed filter is determined for each individual user when the hearing aid is fitted to the user by a dispenser or another trained person. This not only requires an additional fitting step, but also makes it impossible to capture the true invariant part of the feedback path since the feedback path measured by the dispenser already contains several variable parts. For example, the placement of the hearing aid in the ear canal is included in the invariant portion, but it can change, for example, when the hearing aid is reinserted into the ear.

  WO 99/26453 also refers to the ability to allow the first filter to slowly adapt to follow gradual changes in the hearing aid, such as component drift. However, no further explanation on how to allow the first filter to adapt slowly, ie an adaptation method for a slow adaptive filter, is not disclosed in WO 99/26453.

  According to some embodiments, a method for adapting a slow adaptive filter is proposed. According to this, the initial setting for determining the value of the filter coefficient during fitting of the hearing aid or turning on the power is avoided.

An input transducer for generating an audio signal;
A feedback suppression circuit configured to model the feedback path of the hearing aid;
A subtractor for subtracting the output signal of the feedback suppression circuit from the audio signal to form a feedback corrected audio signal;
A hearing loss processor coupled to the output of the subtractor to process the feedback corrected audio signal to perform hearing loss correction;
An output transducer, preferably a receiver, coupled to the output of the hearing loss processor to provide an audio signal based on the processed feedback;
A hearing aid comprising:
Feedback suppression circuit
A slow adaptive filter with an input coupled to the hearing loss processor and an output;
A fast adaptive filter with an input coupled to the slow adaptive filter and an output;
A hearing aid is provided.

  The output of the fast adaptive filter may constitute the output of the feedback suppression circuit.

  A transducer is a device that converts a signal in one energy form into a corresponding signal in another energy form. For example, the input transducer may include a microphone that converts an acoustic signal reaching the microphone into a corresponding analog audio signal, where the instantaneous voltage of the audio signal is continuous with the sound pressure of the acoustic signal. Changes.

  The input transducer may also include a telecoil. The telecoil converts the magnetic field in the telecoil into a corresponding analog audio signal. In this corresponding analog audio signal, the instantaneous voltage of the audio signal changes continuously with the magnetic field strength in the telecoil. Telecoil is the signal-to-noise ratio of the voice of a speaker who speaks to many people in public places such as churches, auditoriums, theaters, movie theaters, or through on-site announcement systems such as stations, airports, shopping malls, etc. Typically used to improve. The speaker's speaking voice is converted into a magnetic field using an inductive loop system (also referred to as a “hyalin group”), and the telecoil is used to magnetically receive the magnetically transmitted speech signal.

  When using a telecoil, feedback may be generated when the telecoil receives a magnetic field generated by a hearing aid (eg, generated by a receiver).

  The input transducer includes at least two spaced apart microphones and a beamformer configured to combine the microphone output signals of the at least two microphones into, for example, a directional microphone signal well known in the art; May further be included.

  The input transducer includes one or more microphones, a telecoil, and a switch for selecting, for example, an omnidirectional microphone signal, a directional microphone signal, or a telecoil signal alone or in any combination as an audio signal; May be included.

  Preferably, the output transducer may include a receiver, i.e. a small speaker. The receiver converts the analog audio signal into a corresponding acoustic signal. In this corresponding acoustic signal, the instantaneous sound pressure changes continuously according to the amplitude of the analog audio signal.

  Typically, an analog audio signal is made suitable for digital signal processing by being converted into a corresponding digital audio signal in an analog-to-digital converter. Thereby, the amplitude of the analog audio signal is represented by a binary number. Thus, a discrete time and discrete amplitude digital audio signal in the form of a sequence of digital values represents a continuous time and continuous amplitude analog audio signal.

  Throughout this disclosure, a portion of an audio signal produced by the hearing aid itself, such as produced as a result of sound, mechanical vibrations, electromagnetic fields, etc. produced by the hearing aid, is fed back into the audio signal. It is called a signal part or simply a feedback signal.

  A feedback suppression circuit is provided in the hearing aid to model the feedback path. That is, the feedback suppression circuit preferably has the same transfer function as the feedback path itself so that the output signal of the feedback suppression circuit matches the feedback signal portion of the audio signal as closely as possible.

  A subtractor is provided for subtracting the output signal of the feedback suppression circuit from the audio signal to form a feedback corrected audio signal with the feedback signal portion removed or at least reduced.

  The feedback suppression circuit includes an adaptive filter that tracks the current transfer function of the feedback path.

  The feedback suppression circuit may include a delay due to one or more electronic components corresponding to the delay of the feedback signal propagating along the hearing aid feedback path.

  The feedback suppression circuit may include at least one fixed filter configured to model a stationary portion of the hearing aid's feedback path.

  The feedback suppression circuit may include at least one slow adaptive filter and at least one fast adaptive filter configured to model the feedback path.

  The slow adaptive filter eliminates the need to make initial settings in the feedback suppression circuit during fitting to the intended user or when the hearing aid is turned on.

  In addition, the slow adaptive filter improves the performance of the feedback suppression circuit in relation to gradual changes in the feedback path, such as earwax accumulation, changes due to re-insertion of the hearing aid into the user's ear canal, and drifts in the hearing aid electronics. To do. Thus, a slow adaptive filter may track changes that occur in minutes or more slowly, while a fast adaptive filter uses smiles, chews, sneezes or telephone handsets that are performed in tens of milliseconds to seconds. You may track changes such as.

  The filter coefficient of the slow adaptive filter may be based at least in part on the difference between the output signal of the slow adaptive filter and the audio signal.

  The filter coefficient of the slow adaptive filter may be based at least in part on the difference between the output signal of the slow adaptive filter and the output signal of the fast adaptive filter.

  The filter coefficient of the slow adaptive filter may be based at least in part on the difference between the slow adaptive filter output signal and the weighted sum of the fast adaptive filter output signal and the first audio signal.

  In the following, the hearing aid components and signals first mentioned above will be denoted as “first” components and signals, respectively, and will be referred to as “second” components and signals mentioned below. Distinguish from each.

Hearing aid
A second input transducer for generating a second audio signal;
A second feedback suppression circuit configured to model a second feedback path of the hearing aid;
A second subtractor for subtracting the second output signal of the second feedback suppression circuit from the second audio signal to form a second feedback corrected audio signal;
May further include
The hearing loss processor is coupled to a second subtractor for processing the second feedback corrected audio signal to perform hearing loss correction,
The second feedback suppression circuit is
A second slow adaptive filter comprising an input coupled to the hearing loss processor or the first slow adaptive filter, and an output;
A second fast adaptive filter comprising an input coupled to the second slow adaptive filter and an output;
including.

  The output of the second fast adaptive filter may constitute the output of the second feedback suppression circuit.

  In hearing aids with multiple input transducers, such as front and rear microphones, the distance between the input transducers is usually small due to the small size of the hearing aid housing. Feedback paths to individual input transducers in close proximity to each other are expected to have a similar transfer function. Thus, one filter may be used to model one of the feedback paths for each one of the input transducers. A simpler filter, labeled “correction filter” below, is used to model the difference between the modeled feedback path and the other feedback path to each of the other input transducers. Also good. Thereby, duplication of common functions of the slow adaptive filter is substantially avoided. Differences in the feedback path can lead to the formation of subsample delays and small amplitude responses due to the small difference in physical distance between the target output transducer and the input transducer.

  As a result, the main purpose of the correction filter may be to realize the formation of an interpolation that ideally requires an anti-causal impulse response. This is because interpolation is preferably based on samples on both sides of the interpolation point. Such filters are usually difficult to implement, but for feedback suppression circuits, this is typically possible due to the total bulk delay in the feedback loop of at least two sample blocks. Some of this bulk delay can be used to provide a response slightly in time so that the correction filter has enough information to perform the desired interpolation.

  The idea of modeling the difference in the feedback path may be applied to a fast adaptive filter. Changes in the dynamic feedback path can cause sub-sample time differences in the feedback loop, suitable for modeling by interpolation, and can also result in the formation of small amplitude responses.

  A delay due to electronic components corresponding to the delay caused by signal propagation along the feedback path may be placed in the feedback suppression circuit. This simplifies the adaptive filter and also facilitates interpolation based on samples before and after interpolation.

  The delay of the feedback suppression circuit corresponding to the propagation delay along the corresponding feedback path is one common delay, preferably the shortest delay between the output transducer and one of the input transducers, and each other input from the output transducer. It may be provided in the form of an individual delay that models the additional delay to the transducer.

  The slow adaptive filter may be a FIR filter that is less complex and more stable than the IIR filter.

  The output signal of the slow filter may be scaled using a bit shifter, preferably adaptively scaled. Scaling, such as adaptive scaling, maximizes accuracy, optionally extends the coefficient range, and also allows for any slow adaptation. Without adaptive scaling, the optimal step size may not be available for all feedback paths.

  The filter coefficient of the second slow adaptive filter may be based at least in part on the difference between the output signal of the second slow adaptive filter and the second audio signal.

  The filter coefficient of the second slow adaptive filter may be based at least in part on the difference between the output signal of the second slow adaptive filter and the output signal of the second fast adaptive filter.

  The filter coefficient of the second slow adaptive filter is at least partially in a difference between the output signal of the second slow adaptive filter and the weighted sum of the output signal of the second fast adaptive filter and the second audio signal. May be based.

及び入力ベクトル

を備えたＦＩＲフィルタアーキテクチャは、次のように説明される。

A weight vector for calculating the output signal d at time n

And input vector

The FIR filter architecture with is described as follows.

で畳み込み、ｕと同様にｄをベクトル化し、単純化のために起こり得る遅延を無視することによって、以下においてキャンセル信号ｃと表示される高速適応フィルタの出力信号ｃが提供される。

This signal is converted to a fast adaptive filter.

By convolution with and vectorizing d like u and ignoring possible delays for simplification, an output signal c of the fast adaptive filter, denoted below as cancellation signal c, is provided.

The audio sample s of the input transducer is
s (n) = x (n) + f (n)
Is assumed to be a mixture of the external signal x and the feedback signal f, and after feedback cancellation,
e (n) = s (n) -c (n) = x (n) + f (n) -c (n)
And this equation provides complete cancellation performance when f (n) is equal to c (n).

及び低速フィルタ係数

の両方を適応させることが可能である。 As a rule, fast filter coefficients using a single error criterion

And slow filter coefficients

It is possible to adapt both.

  However, in the following, a more efficient approach is disclosed that more fully draws on the basic purpose differences between the slow and fast adaptive filters. That is, the slow filter desirably models the characteristics of the feedback path that is subject only to gradual changes, while the fast adaptive filter desirably models only rapid changes. Thus, different error criteria for the slow adaptive filter and the fast adaptive filter may be more appropriate.

Under normal circumstances, the cancellation signal c (n) may on average be assumed to be the best known estimate of the feedback signal. Thus, the slow adaptive filter is connected to track the signal, thereby may incorporate changes from the high speed adaptive filter to provide an error signal e _1.
e ₁ (n) = c (n) −d (n)

Alternatively, the direct approach error signal is defined as:
e ₂ (n) = s (n) −d (n)
It is the signal that will actually be the output of the feedback suppression circuit when the fast adaptive filter is fixed in its reference state.

The error signal e ₁ is less sensitive to bias because the fast adaptive filter uses an adaptive signal model, but it can lead to a local minimum that can trap the slow adaptive filter and prevent further adaptation.

The error signal e ₂ is the decorrelation signal is optimal, it is possible that more influenced the bias caused by the tone input.

Therefore, another alternative is to use a weighted sum of the above error signals.
e _m (n) = (1−β) e ₁ (n) + βe ₂ (n)
= (1- [beta]) c (n) + [beta] s (n) -d (n)
= S (n)-(1-β) e (n) -d (n)
= T (n) -d (n)
In this equation, t (n) can be regarded as a target signal defined by a weighted sum.

  β may be a fixed predetermined parameter.

An appropriate second order error criterion to be minimized for processing a block of M samples can be formulated as follows:

が与えられ、この式で、
∇ｅ_ｍ＝∇ｔ（ｎ）−∇ｄ（ｎ）
であり、この式は、係数ｗについて、項∇ｔ（ｎ）を無視する（ターゲットは、現在の内部モデルに依存するべきではない）ことによって、

に単純化することができる。その結果、勾配方向は、加重誤差信号をそれぞれのタップにおけるＦＩＲフィルタ入力信号と相互相関させることによって推定される。 Next, by using a chain rule that computes the gradient direction that minimizes J for the slow adaptive filter coefficients:

Is given by
∇e _m = ∇t (n) −∇d (n)
And this equation ignores the term ∇t (n) for the coefficient w (the target should not depend on the current internal model)

Can be simplified. As a result, the gradient direction is estimated by cross-correlating the weighted error signal with the FIR filter input signal at each tap.

  The derivation of the correction filter coefficients from the front to the rear is similar except that the cross-correlation is performed using the common slow adaptive filter output signal d (n) input to the correction filter. May be.

  For slow and fast adaptive filters, the step size is determined by the adaptive filter, such as least mean square (LMS) algorithm, normalized least mean square (NLMS) algorithm, or line search, conjugate gradient method, Hessian estimation technique, etc. It may be determined by a method well known in the technical field.

  For slow adaptive filters, however, a simple code based algorithm may suffice and the appropriate step size may be determined directly from the current filter coefficients.

  In order to minimize the complexity of adjusting the filter coefficients, only some of the coefficients, ie at least one coefficient, may be adjusted or updated for each sample block. Since only one cross-correlation is used, the computational complexity for one weight is approximately equal to the complexity of adding one FIR filter coefficient. For example, updates exceeding 4 filter coefficients per block may be undesirable, at least for slow adaptive filters.

  Once the update cycle is complete, i.e. all coefficients have been adjusted or updated, special events are scheduled to update administrative settings such as coefficient step size, model scaling, and constraints. The For optimal accuracy, the step size and scaling must be updated during normal operation of the hearing aid. This is because the amplitude of the feedback path is not known in advance. However, a reasonable estimate may be provided to accelerate initial convergence.

  A good step size for sign-based updates is defined in proportion to the amplitude response of the feedback path. Once at least a rough measure of the feedback amplitude is known, this approach provides a nearly constant accuracy for tracking changes in the feedback path, regardless of the feedback signal level.

  If the feedback path is unknown, another approach may be used immediately after turning on the hearing aid. In the initial start-up phase, a faster and even initially non-proportional step size may be used to accelerate convergence and quickly extinguish possible initial feedback such as howling. The transition time from the initial rate to the final rate can be changed and may be from a few seconds to a maximum of approximately one minute.

  Alternatively or additionally, a gradual gain increase and reading of the coefficients previously stored in the permanent memory may be performed.

  In situations where the slow adaptive filter can track misleading or non-informational signals, one or more criteria for adaptation may be used for the slow adaptive filter to prevent adaptation of the slow adaptive filter. May be added to. Thereby, the slow adaptive filter may be configured to adjust one or more of its filter coefficients only under certain conditions.

  For example, a slow adaptive filter (1) if the signal level exceeds a predetermined threshold, (2) if the (direct error) signal and the corresponding signal model should be protected against adaptation. And / or (3) the hearing aid may be configured to only adjust one or more of its filter coefficients when it is in its initial start-up phase (immediately after power-up).

  The level threshold of (1) mainly prevents adaptation to meaningless input signals such as microphone noise. This can also extend the start-up phase if the algorithm is activated in a quiet or silenced state.

  In connection with (2), a signal is considered protected against adaptation if it is less predictable (eg, a pure tone is definitely predictable). The above case is determined, for example, by comparing the signal level of the decorrelated error signal used for updating the fast adaptive filter directly with the level of the error signal itself.

  In addition or as an alternative, the error signal is the p-norm, preferably 1 norm, of the fast adaptive filter coefficient vector (representing the signal model), which is less than a predetermined threshold (a large 1 norm indicates tone input) It is considered protected.

  The hearing aid may be a multi-band hearing aid that performs hearing loss correction differently in different frequency bands, thus taking into account the frequency dependence of the intended user's hearing loss. In a multiband hearing aid, the audio signal from the input transducer is divided into two or more frequency channels or bands. Typically, audio signals are amplified differently in each frequency band. For example, a compressor may be utilized to compress the dynamic range of the audio signal according to the intended user's hearing loss. In multiband hearing aids, the compressor performs compression differently in each of the frequency bands, changing not only the compression ratio but also the time constant associated with each band. The time constant refers to the time constant of the attack and release of the compressor. The compressor attack time is the time required for the compressor to reduce gain at the beginning of loud speech. Release time is the time required for the compressor to increase the gain after a loud audio pause.

  The frequency band may be a distortion frequency band. For example, a hearing aid may have a compressor that performs dynamic range compression using digital frequency warping as disclosed in more detail in WO 03/015468. In particular, the basic operating principle of the distortion compressor is shown in FIG. 11 and the corresponding part of the description of WO 03/015468.

  For example, the feedback suppression circuit including one or more adaptive filters may be a broadband model. That is, the model can operate in substantially the entire frequency range of hearing aid operation or in the effective portion of the hearing aid frequency range without being divided into frequency band sets.

  Alternatively, the feedback suppression circuit may be divided into frequency band sets for individual modeling of feedback paths in each frequency band. In this case, the estimated value of the residual feedback signal may be individually provided in each frequency band m of the feedback suppression circuit.

  The frequency band m of the feedback suppression circuit and the frequency band k of the hearing loss correction may be the same, but preferably they are different, and preferably the number of frequency bands m of the feedback suppression circuit is the hearing loss correction. Less than the number of frequency bands.

  Throughout this disclosure, the term audio signal is used to identify any analog or digital signal that forms part of the signal path from the output of the microphone to the input of the hearing loss processor.

  The feedback suppression circuit may be implemented as one or more dedicated electronic hardware circuits, may be combined with appropriate signal processing software to form part of a signal processor, or with dedicated hardware and appropriate It may be a combination with one or more signal processors with signal processing software.

  Signal processing in the new hearing aid may be performed by dedicated hardware, may be performed by a signal processor, or may be performed by a combination of dedicated hardware and one or more signal processors.

  As used herein, the terms "processor", "signal processor", "controller", "system", etc. are either hardware, a combination of hardware and software, software, or running software It is intended to refer to the CPU related components.

  For example, a “processor”, “signal processor”, “controller”, “system”, etc. may be, but is not limited to, a process, processor, object, executable, execution thread, and / or program running on the processor. Never happen.

  By way of illustration, terms such as “processor”, “signal processor”, “controller”, “system”, etc., refer to both applications running on the processor and hardware processors. One or more “processors”, “signal processors”, “controllers”, “systems”, etc., or any combination thereof, may also exist within a process and / or execution thread. A “processor”, “signal processor”, “controller”, “system”, etc., or any combination thereof, can be stationed on one hardware processor, possibly in combination with other hardware circuits. And / or may be distributed between two or more hardware processors and possibly in combination with other hardware circuits.

  Also, the processor (or similar) may be any component or any combination of components that can perform signal processing. For example, the signal processor may be an ASIC processor, FPGA processor, general purpose processor, microprocessor, circuit component, or integrated circuit.

  The hearing aid includes a first input transducer for generating a first audio signal, a first feedback suppression circuit configured to model a first feedback path of the hearing aid, and a first feedback corrected A first subtractor for subtracting the first output signal of the first feedback suppression circuit from the first audio signal to form an audio signal; and a first subtractor for performing hearing loss correction. A hearing loss processor coupled to the first subtractor to process the feedback corrected audio signal and a hearing loss processor to provide an audio signal based on the processed first feedback corrected audio signal And a first feedback suppression circuit is coupled to the hearing loss processor. And a first slow adaptive filter with an output, a first fast adaptive filter with an input coupled to the first slow adaptive filter, and an output. The filter coefficient of the filter is based at least in part on the difference between the output signal of the first slow adaptive filter and at least one of the output signal of the first fast adaptive filter and the first audio signal.

  Optionally, the filter coefficient of the first slow adaptive filter is based on the difference between the output signal of the first slow adaptive filter and the first audio signal.

  Optionally, the filter coefficient of the first slow adaptive filter is based on the difference between the output signal of the first slow adaptive filter and the output signal of the first fast adaptive filter.

  Optionally, the filter coefficient of the first slow adaptive filter is a difference between the output signal of the first slow adaptive filter and the weighted sum of the first fast adaptive filter output signal and the first audio signal. Based on.

  Optionally, the hearing aid includes a second input transducer for generating a second audio signal, a second feedback suppression circuit configured to model the second feedback path of the hearing aid, A second subtractor for subtracting the second output signal of the second feedback suppression circuit from the second audio signal to form a second feedback corrected audio signal, the hearing loss processor Is coupled to the second subtractor to process the second feedback corrected audio signal to perform hearing loss correction, and the second feedback suppression circuit is coupled to the input of the hearing loss processor. And a second low-speed adaptive filter having an output unit, an input unit coupled to the second low-speed adaptive filter, and a first unit having an output unit And a second low-speed adaptive filter having a filter coefficient comprising: an output signal of the second low-speed adaptive filter; at least one of an output signal of the second high-speed adaptive filter and a second audio signal; Based at least in part on the difference between.

  Optionally, the hearing aid includes a second input transducer for generating a second audio signal, a second feedback suppression circuit configured to model the second feedback path of the hearing aid, A second subtractor for subtracting the second output signal of the second feedback suppression circuit from the second audio signal to form a second feedback corrected audio signal, the hearing loss processor Is coupled to the second subtractor to process the second feedback corrected audio signal to perform hearing loss correction, and the second feedback suppression circuit is coupled to the first slow adaptive filter. A second low-speed adaptive filter having an input unit and an output unit, an input unit coupled to the second low-speed adaptive filter, and an output unit A filter coefficient of the second low-speed adaptive filter, the output coefficient of the second low-speed adaptive filter, the output signal of the second high-speed adaptive filter, and the second audio signal. Based at least in part on the difference between one.

  Optionally, the filter coefficient of the second slow adaptive filter is based on the difference between the output signal of the second slow adaptive filter and the second audio signal.

  Optionally, the filter coefficient of the second slow adaptive filter is based on the difference between the output signal of the second slow adaptive filter and the output signal of the second fast adaptive filter.

  Optionally, the filter coefficient of the second slow adaptive filter is the difference between the output signal of the second slow adaptive filter and the weighted sum of the output signal of the second fast adaptive filter and the second audio signal. Based on.

  Optionally, the first slow adaptive filter is configured to adjust one or more of the filter coefficients if at least one criterion is met.

  Optionally, the at least one criterion includes a signal level of an input signal of the first feedback suppression circuit that is greater than a predetermined threshold.

  Optionally, the at least one criterion includes an autocorrelation of the error signal that is below a predetermined threshold.

  Optionally, the at least one criterion includes performing an update that constitutes a first update that is performed immediately after the hearing aid is powered on.

  Optionally, the at least one criterion includes a p-norm of the filter coefficient vector of the first fast adaptive filter that is less than the predetermined threshold.

  Other and further aspects and features will become apparent upon reading the following detailed description of the embodiments.

  In the following, the novel method and hearing aid will be described in more detail in connection with the drawings in which various examples are shown.

1 schematically shows a hearing aid with a feedback path. 1 schematically shows a prior art hearing aid with feedback suppression. 1 schematically illustrates a novel hearing aid with feedback suppression. Figure 3 schematically shows another novel hearing aid with feedback suppression. Figure 3 schematically shows yet another novel hearing aid with feedback suppression. Figure 3 schematically shows yet another novel hearing aid with feedback suppression. Figure 3 schematically shows yet another novel hearing aid with feedback suppression. Figure 3 schematically shows yet another novel hearing aid with feedback suppression. Fig. 4 schematically shows another novel hearing aid with feedback suppression having a fast adaptive filter with signal modeling circuitry. The signal modeling circuit is schematically shown in more detail. 1 schematically shows part of a novel feedback suppression circuit. Fig. 6 shows a plot of the feedback path transfer function when reinsertion is repeated. Figure 6 shows a plot of the performance of the slow filter feedback path modeling.

  The drawings illustrate the design and utility of the embodiments, in which like elements are referred to by common reference numerals. Accordingly, similar elements may not be described in detail in connection with the description of each figure. In order to better understand how the above-listed and other advantages and objectives can be obtained, a more specific description of the embodiments is provided, which are illustrated in the accompanying drawings. Note that the drawings are only intended to facilitate the description of the features. The drawings are not intended to be an exhaustive description of the claimed invention, nor are they intended to limit the scope of the claimed invention. Moreover, the illustrated features may not represent all aspects or advantages. An aspect or advantage described in connection with a particular feature is not necessarily limited to that feature, and is realized in any other feature (even if not so illustrated and not explicitly described). be able to.

  The novel hearing aid according to the claims may be embodied in different forms not shown in the accompanying drawings and should not be construed as limited to the examples set forth herein. .

  FIG. 1 schematically shows a hearing aid 10 and a feedback path 12, ie, a feedback path 12 along which the signal generated by the hearing aid 10 propagates back to the input of the hearing aid 10.

  In FIG. 1, an acoustic signal 14 is received by a microphone 16 that converts the acoustic signal 14 into an audio signal 18 that is input to a hearing loss processor 20 for hearing loss correction. In the hearing loss processor 20, the audio signal 18 is amplified according to the user's hearing loss. The hearing loss processor 20 may include, for example, a multiband compressor. The output signal 22 of the hearing loss processor 20 is converted to an acoustic output signal 24 by the receiver 26, which is acoustically directed toward the user's eardrum when the hearing aid 10 is worn in an appropriate operating position in the user's ear. Radiates a signal.

  Typically, a portion of the acoustic signal 24 from the receiver 26 propagates back to the microphone 16 as shown as feedback path 12 in FIG.

  At low gain, the feedback simply gives a harmless timbre change. However, with a large hearing aid gain, the feedback signal level at the microphone 16 can exceed the level of the original acoustic signal 14, thereby causing audible distortion and possibly howling.

  In order to overcome feedback, it is well known to provide a feedback suppression circuit in the hearing aid, as shown in FIG.

  FIG. 2 schematically shows a hearing aid 10 with a feedback suppression circuit 28. The feedback suppression circuit 28 models the feedback path 12. That is, the feedback suppression circuit attempts to generate a signal that is identical to the signal propagated along the feedback path 12. The feedback suppression circuit 28 is preferably connected to the receiver 26 and microphone 16 so that the transfer function of the feedback suppression circuit 28 is preferably equal to the sum of the transfer function of the receiver 26, the transfer function of the feedback path 12, and the transfer function of the microphone 16. It should be noted that these models are included.

  The feedback suppression circuit 28 generates an output signal 30 to the subtractor 32 to suppress or cancel the feedback signal portion of the audio signal 18 before processing in the hearing loss processor 20.

  In conventional hearing aids 10, the feedback suppression circuit 28 is typically an adaptive digital filter that adapts to changes in the feedback path 12.

  WO 99/26453 discloses feedback suppression with a series connection of two adaptive filters. The first filter 36 is applied when the hearing aid is fitted to the intended user at the dispenser office. In fitting, the filter 36 adapts quickly using the white noise probe signal and then the filter coefficients are fixed. That is, during the subsequent normal operation of the hearing aid, the first filter 36 operates as a fixed filter 36.

  The first filter 36 models portions of the hearing aid feedback path 12 that are assumed to be substantially constant during use of the hearing aid 10, such as the transfer function of the microphone 16 and the transfer function of the receiver 26, and the basic portion of the feedback path 12. Turn into.

  The second filter 38 is adapted during use of the hearing aid 10 and does not use a separate probe signal. This filter 38 provides a quick correction of the feedback suppression circuit 28 when the hearing aid 10 becomes unstable and occurs in everyday use, such as caused by chewing, sneezing or using a telephone handset. The disturbance of the feedback path 12 is tracked. Thus, the fast adaptive filter 38 can track changes that occur from tens of milliseconds to seconds.

  Except when requiring additional fitting steps, the fixed filter 36 cannot capture the true invariant part of the modeled transfer function. This is because the determined fixed filter coefficients already contain some variable parts. For example, the attachment of the hearing aid 10 in the ear canal is included in the invariant portion, but it can change, for example, when the hearing aid 10 is reinserted into the ear.

  In the following, a new hearing aid is shown that does not require an additional fitting step and also addresses the true variable part of the modeled transfer function.

  FIG. 3 shows a first example of a hearing aid 10 according to the claims. The hearing aid 10 includes an input transducer, or microphone 16a, for generating the audio signal 18a and a feedback suppression circuit 28a that models the feedback path 12a (ie, the feedback suppression circuit 28a is a signal propagated along the feedback path 12a). To generate the same signal as The feedback suppression circuit 28a is preferably connected to the receiver 26 and microphone 16a so that the transfer function of the feedback suppression circuit 28a is preferably equal to the sum of the transfer function of the receiver 26, the transfer function of the feedback path 12a, and the transfer function of the microphone 16a. Note that the model includes:

  The feedback suppression circuit 28a generates an output signal 30a to the subtractor 32a in order to suppress or cancel the feedback signal portion of the audio signal 18a before being processed by the hearing loss processor 20.

  The hearing loss processor 20 is coupled to the output of the subtractor 32a and further to the receiver 26 for processing the feedback corrected audio signal 34a to perform hearing loss correction. The receiver 26 is coupled to the output of the hearing loss processor 20 for converting the processed feedback corrected audio signal 22 into an audio signal.

  The feedback suppression circuit 28a includes an input unit coupled to the output unit of the hearing loss processor 20, a low-speed adaptive filter 36a having an output unit, an input unit coupled to the output unit of the low-speed adaptive filter 36a, and a feedback suppression circuit. And a high-speed adaptive filter 38a having an output unit constituting the output unit 28a.

  During normal operation of the illustrated hearing aid 10, the cancellation signal 30a constitutes a good guess of the feedback signal portion of the audio signal 18a in most situations. Thus, the slow adaptive filter 36a is connected to track the signal 30a, thereby incorporating updates from the fast adaptive filter 38a.

  Accordingly, the filter coefficient of the low speed adaptive filter 36a becomes an error signal 42a equal to the difference between the output signal 44a of the low speed adaptive filter 36a output by the subtractor 40a and the cancel signal 30a output by the high speed adaptive filter 38a. Based at least in part.

  The filter coefficients of the fast adaptive filter 38a are based at least in part on the error signal 34a output by the subtractor 32a.

  By providing the low speed adaptive filter 36a, the initial setting of the feedback suppression circuit 28a becomes unnecessary. Also, gradual changes in the feedback path are appropriately modeled by the slow adaptive filter 36a.

  A fixed filter (see FIG. 11) may be connected in series with the slow adaptive filter 36a and the fast adaptive filter 38a. The fixed filter provides a transfer function for the microphone 16a, a transfer function for an amplifier (not shown) that drives the receiver 26, and a receiver so that the adaptive filters 36a, 38a are only required to handle variations from the initial value. It may be configured to model the true invariant part of the feedback path 12a, such as 26 transfer functions, as well as initial values of the basic part of the feedback path 12a.

  A bulk delay (see FIG. 11) may be connected in series with the slow adaptive filter 36a and the fast adaptive filter 38a. The bulk delay may be configured to model the propagation delay of the feedback signal propagating along the feedback path, thereby freeing the adaptive filters 36a, 38a from this task.

  A barrel shifter (see FIG. 11) may be connected to the output of the slow adaptive filter 36a and / or the fast adaptive filter 38a, preferably for adaptively scaling the output signal. Scaling, such as adaptive scaling, maximizes accuracy, optionally extends the coefficient range, and also allows for any slow adaptation. Without adaptive scaling, the optimal step size may not be available for all feedback paths.

  The hearing aid 10 shown in FIG. 4 is similar to the hearing aid of FIG. 3 except for the following points. That is, the hearing aid 10 of FIG. 4 has two microphones 16a, 16b, a front microphone 16a and a rear microphone 16b, where the hearing loss processor 20 is selectable beamforming, as is well known in the hearing aid art. 3 is the same as the hearing aid of FIG. The feedback path 12a to the front microphone 16a is modeled by a first feedback suppression circuit 28a identical to the feedback circuit 28a shown in FIG. Similarly, the feedback path 12b to the rear microphone 16b is such that the input of the second slow adaptive filter 36b is coupled to the output 44a of the first slow adaptive filter 36a rather than the output 22 of the hearing loss processor 20. Except for this point, the second feedback suppression circuit 28b corresponding to the feedback circuit 28a shown in FIG. 3 is modeled.

  In the illustrated hearing aid 10, the distance between the receiver 26 and the front microphone 16a is shorter than the distance between the receiver 26 and the rear microphone 16b. If the reverse is true, i.e., the shortest distance between the receiver 26 and the rear microphone 16b, the microphone 16a is a rear microphone and the microphone 16b is a front microphone.

  Accordingly, the first low speed adaptive filter 36a is connected to the front microphone so that the series connection of the first low speed adaptive filter 36a and the second low speed adaptive filter 36b is combined to model the feedback path to the rear microphone 16b. The second low-speed adaptive filter 36b models the difference between the feedback path to the front microphone 16a and the feedback path to the rear microphone 16b. In the illustrated example, the distance between the front and rear microphones 16a, 16b is small, and each of the feedback paths 12a, 12b has a similar transfer function with sub-sample delay differences and small differences in amplitude response formation. Therefore, the second low-speed adaptive filter 36b is simpler than the first low-speed adaptive filter 36a. The second slow adaptive filter 36b performs anti-causal interpolation enabled by the bulk delay of the feedback suppression circuits 28a, 28b (see FIG. 11).

  In another example (not shown) where the individual feedback paths 12a, 12b do not have a similar transfer function, the feedback paths 12a, 12b to the front microphone 16a and the rear microphone 16b are respectively independent feedback circuits 28a, 28b, each of the feedback circuits 28a, 28b is similar to the feedback circuit 28a shown in FIG. 3, and is the input of both the first and second slow adaptive filters 36a, 36b. Is coupled to the output 22 of the hearing loss processor 20.

  A first fixed filter (see FIG. 11) may be connected in series with the first low-speed adaptive filter 36a and the first high-speed adaptive filter 38a. The first fixed filter is a transfer function of the microphone 16a, an amplifier that drives the receiver 26 so that the first low-speed and high-speed adaptive filters 36a, 38a are only required to cope with fluctuations from the initial value ( Configured to model the true invariant part of the first feedback path 12a, such as the transfer function (not shown), the transfer function of the receiver 26, and the initial value of the basic part of the first feedback path 12a. Yes.

  A second fixed filter (see FIG. 11) may be connected in series with the second low-speed adaptive filter 36b and the second high-speed adaptive filter 38b. The second fixed filter is a transfer function of the microphone 16b, an amplifier that drives the receiver 26 so that the second low-speed and high-speed adaptive filters 36b, 38b are only required to cope with fluctuations from the initial value ( Configured to model the true invariant portion of the second feedback path 12b, such as the transfer function (not shown), the transfer function of the receiver 26, and the initial value of the basic portion of the second feedback path 12b. Yes.

  Each bulk delay (see FIG. 11) is connected in series with a slow adaptive filter 36a, 36b and a fast adaptive filter 38a, 38b. Each bulk delay is configured to model the propagation delay of the respective feedback signal propagating along feedback paths 12a, 12b, thereby freeing adaptive filters 36a, 36b, 38a, 38b from this task. Yes. The bulk delay is arranged to facilitate anti-causal interpolation in the second slow adaptive filter 36b.

  Each barrel shifter (see FIG. 11) is connected to the output of the slow adaptive filters 36a, 36b to adaptively scale the respective output signals 44a, 44b. Scaling maximizes accuracy, optionally extends the coefficient range, and also allows for any slow adaptation. Without adaptive scaling, the optimal step size may not be available for all feedback paths.

  The hearing aid 10 shown in FIG. 5 is similar to the hearing aid of FIG. 3 except for the following points. That is, the difference between the filter coefficient of the low-speed adaptive filter 36a of the hearing aid 10 of FIG. 5 between the output signal 44a of the low-speed adaptive filter 36a output by the subtractor 40a and the cancel signal 30a output by the high-speed adaptive filter 38a. And is based at least in part on an error signal 42a output by the subtractor 40a equal to the difference between the output signal 44a of the slow adaptive filter 36a and the audio signal 18a. This is the same as the hearing aid of FIG.

  The error signal 42a is also called a direct approach error. The error signal 42a is a signal that will actually be the output of the feedback suppression circuit when the fast adaptive filter is fixed in its reference state. The error signal 42a is optimal for the decorrelation signal, but may be more affected by the bias caused by the tone input. On the other hand, the error signal 42a of FIG. 3 is less sensitive to bias because the fast adaptive filter uses an adaptive signal model, but it has a local minimum that traps the slow adaptive filter to prevent further adaptation. There is a possibility of connection.

  The hearing aid 10 shown in FIG. 6 is the same as the hearing aid of FIG. 4 except for the following points. That is, the filter coefficient of the first low-speed adaptive filter 36a of the hearing aid 10 of FIG. 6 is output by the first subtractor 40a, the first output signal 44a of the first low-speed adaptive filter 36a, and the first The first output signal 44a of the first slow adaptive filter 36a output by the first subtractor 40a, not equal to the difference with the first cancellation signal 30a output by the fast adaptive filter 38a. And, based on the first error signal 42a equal to the difference between the first audio signal 18a and the second slow adaptive filter 36b, the filter coefficient of the second low-speed adaptive filter 36b is The second output signal 44b of the second low-speed adaptive filter 36b output from the second subtractor 40b and the second output signal output from the second high-speed adaptive filter 38b. Between the second output signal 44b of the second slow adaptive filter 36b and the second audio signal 18b output by the second subtractor 40b, not equal to the difference between the cell signal 30b. Similar to the hearing aid of FIG. 4 except that it is based at least in part on a second error signal 42b equal to the difference.

  The hearing aid 10 shown in FIG. 7 combines the error signals 42a shown in FIGS. 3 and 5, respectively. Accordingly, the hearing aid 10 shown in FIG. 7 is the same as the hearing aid of FIG. 3 except for the following points. That is, the filter coefficient of the low speed adaptive filter 36a of the hearing aid 10 in FIG. 7 is between the output signal 44a of the low speed adaptive filter 36a output by the subtractor 40a and the cancel signal 30a output by the high speed adaptive filter 38a. Not equal to the difference, but equal to the difference between the output signal 44a of the slow adaptive filter 36a output by the subtractor 40a and the weighted sum of the cancellation signal 30a output by the audio signal 18a and the fast adaptive filter 38a. Similar to the hearing aid of FIG. 3, except that it is based at least in part on the error signal 42a.

  The hearing aid 10 shown in FIG. 8 is the same as the hearing aid of FIG. 4 or 6 except for the following points. That is, the filter coefficient of the first low-speed adaptive filter 36a of the hearing aid 10 of FIG. 8 is output by the first subtractor 40a, the first output signal 44a of the first low-speed adaptive filter 36a, and the first Based at least in part on a first error signal 42a equal to the difference between the audio signal 18a and the weighted sum of the first cancellation signal 30a output by the first fast adaptive filter 38a, The second output signal 44b of the second low-speed adaptive filter 36b, the second audio signal 18b, and the second high-speed are output by the second subtractor 40b as the filter coefficients of the second low-speed adaptive filter 36b. Based at least in part on the second error signal 42b equal to the difference between the weighted sum of the second cancellation signals 30b output by the adaptive filter 38b. And except that they are similar to the hearing aid of FIG. 4 or 6.

  FIG. 9 shows a hearing aid 10 according to the claims in which a fast adaptive filter 38 a is included in the signal modeling circuit 64. The signal modeling circuit 64 may replace the adaptive filters 38a, 38b of the hearing aid shown in FIGS.

  The fast adaptive filters 38a, 38b shown in FIGS. 3-8 operate according to a so-called “direct approach” in order to minimize the expected signal strength of the error signals 34a, 34b. “Direct approaches” are well known in the hearing aid art, and error signal minimization typically involves least mean square (LMS) algorithms, normalized least mean square (NLMS) algorithms, preferably block It is performed using a normalized least mean square (BNLMS) algorithm where the square error criterion is minimized over a block of samples.

  The direct approach is known to provide biased results when the input signal exhibits a long tail autocorrelation function. For example, in the case of a tone signal, this typically leads to a suboptimal solution. This is because the adaptive feedback model attempts to suppress external tones instead of modeling the actual feedback.

  This problem is solved using the signal modeling circuit 64 shown in FIG. 9, which includes correlation removal circuits 54, 56 that ensure stability in the presence of tone inputs.

  Correlation removal circuit 54 applies adaptive correlation removal to error signal 34a to obtain filtered error signal 58. Correlation removal circuit 56 applies adaptive correlation removal symmetrically to fast adaptive filter input 44a to obtain filtered input 60, both in algorithm block 62 to minimize the filtered error criterion. Let the cross-correlation of the signal provide an estimate of the slope. This is known to be more robust against tones or autocorrelated external signals. In the illustrated signal modeling circuit 64, the signal model used in the correlation removal filters 54 and 56 is obtained from the error signal 34a. However, a fixed correlation removal filter may alternatively be used.

  The signal modeling circuit 64 maintains a statistical model of the external signal 18a to distinguish between the correlation already present in the external signal (tone input) and the correlation between the hearing aid output and input caused by the feedback. May be further configured, thereby reducing sensitivity to tone input.

  FIG. 10 shows an embodiment of the signal modeling circuit 64 in more detail. The illustrated signal modeling circuit 64 includes adaptive correlation removal circuits 54 and 56. Adaptive correlation removal is applied to the error signal 34a to obtain a filtered error signal 58. Furthermore, adaptive correlation cancellation is applied symmetrically to the input 44a to the fast adaptive filter 38a. That is, the filter of the decorrelation circuit 56 is identical to the filter of the decorrelation circuit 54, so that cross-correlating the signals 58, 60 decorrelated in the algorithm 62 minimizes the filtered error criterion. Provide an estimate of the slope. This is known to be more robust to conditions of tones or autocorrelated external signals.

  The decorrelation filter subtracts the linear prediction of the canceled signal (which ideally matches the external signal). In a sense, it is very similar to the well-known linear predictive coding, except that the model is incrementally updated in the circuit. Standard FIR filters are used for linear prediction, and as a result, model generation (for external signals) is IIR and can be interpreted as an autoregressive model. However, it is not necessary to limit to an autoregressive model, for example an autoregressive moving average model (ARMA) can also be used, but extra care may be required to ensure stability and efficiency .

  Alternatively, a fixed correlation removal filter may be used in the signal modeling circuit 64.

  Further, adaptive nonlinear correlation removal may be applied in the signal path. Non-linear correlation removal in the signal path reduces the correlation between the hearing aid output and the external signal. The contribution to the input signal caused by the feedback remains equally correlated (because the applied nonlinearity is known), and as a result, it becomes easier to distinguish the feedback from the tone input, Therefore, the feedback model is improved.

  FIG. 11 shows a feedback suppression circuit excluding the high-speed adaptive filter. Some or all of the illustrated fixed filter 46, delays 48, 52a, 52b and barrel shifters 50a, 50b may be included in the feedback suppression circuit 28 shown in FIGS.

  The output 22 of the hearing loss processor (not shown) is input to a fixed filter 46 connected in series with a first slow adaptive filter 36a and a first fast adaptive filter (not shown). The fixed filter 46 is an amplifier that drives a transfer function of a microphone (not shown) and a receiver (not shown) so that the adaptive filter of the feedback suppression circuit is only required to cope with variations from the initial value. Model the true invariant part of the feedback path (not shown), such as the transfer function of the (not shown), the transfer function of the receiver (not shown), and the initial value of the basic part of the feedback path (not shown). Configured to be

  The bulk delays 48, 52a, 52b are connected in series with a slow adaptive filter 36a, 36b and a fast adaptive filter (not shown). Bulk delays 48, 52a, 52b model each of the propagation delays of the feedback signal propagating along each of the feedback paths (not shown), thereby releasing the adaptive filter of the feedback suppression circuit from this task. It is configured. The bulk delay is arranged to facilitate anti-causal interpolation in the second slow adaptive filter 36b.

  The barrel shifters 50a, 50b are connected to the respective outputs of the first and second slow adaptive filters 36a, 36b so as to scale the respective output signals 44a, 44b to applicable levels. Scaling maximizes accuracy, optionally extends the coefficient range, and also allows for any slow adaptation. Without adaptive scaling, the optimal step size may not be available for all feedback paths.

  FIG. 12 shows a plot of the feedback path transfer function when reinsertion is repeated as an example of the variation in the feedback path modeled by the slow adaptive filter.

  FIG. 13 shows a plot of the feedback path transfer function 80 and the model transfer function 82 learned by the slow adaptive filter after a 60 second conversation.

  While specific embodiments have been illustrated and described, it will be appreciated that the claimed invention is not intended to be limited to the preferred embodiments, and various changes and modifications may be It will be apparent to those skilled in the art that it may be practiced without departing from the spirit and scope of the claimed invention. The specification and drawings are, accordingly, to be regarded in an illustrative sense rather than a restrictive sense. The claimed invention is intended to include alternatives, modifications and equivalents.

Claims (14)

A first input transducer for generating a first audio signal;
A first feedback suppression circuit configured to model the first feedback path of the hearing aid;
A first subtractor for subtracting a first output signal of the first feedback suppression circuit from the first audio signal to form a first feedback corrected audio signal;
A hearing loss processor coupled to the first subtractor to process the first feedback corrected audio signal to perform hearing loss correction;
A receiver coupled to the hearing loss processor to provide an audio signal based on the processed first feedback corrected audio signal;
A hearing aid comprising:
The first feedback suppression circuit comprises:
A first slow adaptive filter comprising an input coupled to the hearing loss processor and an output;
A first fast adaptive filter comprising an input coupled to the first slow adaptive filter and an output;
The filter coefficient of the first slow adaptive filter is a difference between the output signal of the first slow adaptive filter and at least one of the output signal of the first fast adaptive filter and the first audio signal. A hearing aid that is based at least in part.   The hearing aid according to claim 1, wherein the filter coefficient of the first slow adaptive filter is based on a difference between the output signal of the first slow adaptive filter and the first audio signal.   The filter coefficient of the first slow adaptive filter is based on a difference between the output signal of the first slow adaptive filter and the output signal of the first fast adaptive filter. The hearing aid described.   The filter coefficient of the first slow adaptive filter is between the output signal of the first slow adaptive filter and the weighted sum of the output signal of the first fast adaptive filter and the first audio signal. The hearing aid according to claim 1, wherein the hearing aid is based on the difference. A second input transducer for generating a second audio signal;
A second feedback suppression circuit configured to model a second feedback path of the hearing aid;
A second subtractor for subtracting a second output signal of the second feedback suppression circuit from the second audio signal to form a second feedback corrected audio signal;
Further including
The hearing loss processor is coupled to the second subtractor to process the second feedback corrected audio signal to perform hearing loss correction;
The second feedback suppression circuit comprises:
A second slow adaptive filter comprising an input coupled to the hearing loss processor and an output;
A second fast adaptive filter comprising an input coupled to the second slow adaptive filter and an output;
Including
The filter coefficient of the second low-speed adaptive filter is a difference between the output signal of the second low-speed adaptive filter and at least one of the output signal of the second high-speed adaptive filter and the second audio signal. The hearing aid according to claim 1, wherein the hearing aid is based at least in part. A second input transducer for generating a second audio signal;
A second feedback suppression circuit configured to model a second feedback path of the hearing aid;
A second subtractor for subtracting a second output signal of the second feedback suppression circuit from the second audio signal to form a second feedback corrected audio signal;
A hearing aid further comprising:
The hearing loss processor is coupled to the second subtractor to process the second feedback corrected audio signal to perform hearing loss correction;
The second feedback suppression circuit comprises:
A second slow adaptive filter comprising an input coupled to the first slow adaptive filter and an output;
A second fast adaptive filter comprising an input coupled to the second slow adaptive filter and an output;
Including
The filter coefficient of the second low-speed adaptive filter is a difference between the output signal of the second low-speed adaptive filter and at least one of the output signal of the second high-speed adaptive filter and the second audio signal. The hearing aid according to claim 1, wherein the hearing aid is based at least in part.   The hearing aid according to claim 5, wherein the filter coefficient of the second slow adaptive filter is based on a difference between the output signal of the second slow adaptive filter and the second audio signal.   The filter coefficient of the second slow adaptive filter is based on a difference between the output signal of the second slow adaptive filter and the output signal of the second fast adaptive filter. The hearing aid described.   The filter coefficient of the second slow adaptive filter is between the output signal of the second slow adaptive filter and the weighted sum of the output signal of the second fast adaptive filter and the second audio signal. The hearing aid according to claim 5, wherein the hearing aid is based on the difference.   The hearing aid according to claim 1, wherein the first slow adaptive filter is configured to adjust one or more of the filter coefficients if at least one criterion is met.   The hearing aid according to claim 10, wherein the at least one criterion comprises a signal level of an input signal of the first feedback suppression circuit that is greater than a predetermined threshold.   The hearing aid according to claim 10, wherein the at least one criterion includes an autocorrelation of an error signal that is less than a predetermined threshold.   11. A hearing aid according to claim 10, wherein the at least one criterion comprises performing an update that constitutes a first update that is performed immediately after the hearing aid is powered on.   The hearing aid according to claim 10, wherein the at least one criterion includes a p-norm of a filter coefficient vector of the first fast adaptive filter that is less than a predetermined threshold.

Images