EP3373601B1 - Method for frequency distortion of an audio signal and a hearing aid carrying out this method - Google Patents

Description

The invention relates to a method for frequency distortion of an audio signal, wherein different frequency distortions are applied to different signal components of the audio signal, and a frequency-distorted signal is thereby generated.

Controlling acoustic feedback often plays a central role in the operation of acoustic systems, by means of which sound signals from the environment are reproduced in an electrically amplified manner in the broadest sense, ie also for the operation of hearing aids, for example. The acoustic feedback can occur when an output sound signal generated by the acoustic system is partially coupled into an input transducer of the acoustic system, which is provided for picking up the sound signal from the environment and for correspondingly generating an electrical input signal. In this case, signal components of the output sound signal can be electrically amplified again by the acoustic system, so that interference noises are formed in the output sound signal, which can completely superimpose possible useful signals in the sound signal of the environment until they are completely inaudible. Suppression or compensation for acoustic feedback is therefore often provided in the electrical signal path of the acoustic system. Such a compensation often takes place by means of an adaptive filter, to which the fully amplified output signal, from which the output sound signal is generated, is supplied as an input variable. A compensation signal is generated from this, which is supplied to the input signal, which has not yet been amplified, to compensate for the feedback. Adaptive filter control usually takes place via an error signal, which is formed from the difference between the input signal and the compensation signal.

For this purpose, the fully amplified output signal in the acoustic system is often subjected to frequency distortion, as a result of which the output signal is decorrelated from the input signal, so that the signal cancellation described can largely be avoided. Depending on the type of sound signal in the environment, the frequency distortion is usually only applied to a specific frequency range of the amplified signal, for which the latter is filtered at a given division frequency into a signal component to be distorted and a signal component not to be distorted.

In order to suppress the occurrence of artefacts in the output signal as far as possible, the division frequency is mostly adapted to a determined acoustic feedback. The implementation of the division frequency usually takes place via high-pass and low-pass filters, which lead to additional latency in the acoustic system.

The EP 2 244 491 B2 , whose priority application as DE 10 2009 018 812 A1 is published, calls a method for operating a hearing aid, which provides for the division of an input signal into a high-frequency and a low-frequency signal component, wherein a frequency distortion is applied to the high-frequency signal component. In this case, a limit frequency for the division into the high-frequency and low-frequency signal components is determined by analyzing the input signal in such a way that artifacts in an output signal, which is formed using the low-frequency and the frequency-distorted high-frequency signal component, are reduced as far as possible.

The US 2016 / 0 057 548 A1 describes a method for a hearing aid in which, to suppress acoustic feedback, the frequency spectrum transmitted by the hearing aid is divided into a high-frequency and a low-frequency component at a dividing frequency, with the transfer function of the feedback loop being determined in the high-frequency component of the frequency spectrum, using this Transfer function in the range above the division frequency, the behavior of a transfer function for the low-frequency component is estimated in an environment below the division frequency, and the division frequency is adjusted, if necessary, based on the result of this estimation.

The invention is therefore based on the object of specifying a method for frequency distortion of an audio signal, which is intended to minimize the latency as far as possible and, in so doing, to suppress the formation of artefacts as far as possible.

The stated object is achieved according to the invention by the method defined in claim 1 for frequency distortion of an audio signal in a signal processing process in a hearing aid, the audio signal being divided into a plurality of predetermined frequency bands by means of a filter bank, with a band limit frequency being defined by two immediately adjacent frequency bands in each case is determined, based on the audio signal first a target frequency for a boundary between two frequency ranges with different distortion of the frequencies is determined, wherein the target frequency is determined as a critical frequency, which is given by an acoustic feedback to be suppressed on the hearing aid, and based on the target frequency a first frequency band and a second frequency band lying directly above the first frequency band are determined, with that frequency band being determined as the first frequency band, the upper band limit frequency of which is formed by the band limit frequency located directly below the target frequency, and with signal components in the first frequency band having a different distortion of the Frequencies is applied to signal components in the second frequency band. As a result (ie from the two differently frequency-distorted signal components), a frequency-distorted signal is generated. Between the splitting of the audio signal into the frequency bands and the respective distortion of the frequencies, there is a frequency-band-specific amplification of the signal components in the individual frequency bands.

In particular, the frequency-distorted signal is given in the frequency domain.

An audio signal generally includes an electrical signal whose signal profile can serve as a carrier of acoustic information and which can be converted into a corresponding sound signal by a suitable output converter. The audio signal is divided into a plurality of predefined frequency bands by means of a filter bank. The individual frequency bands, in particular the band characteristics of the individual frequency bands, such as the respective center frequency and/or bandwidth, are specified here, for example, by a higher-level application, in which the audio signal is used. The higher-level application is given, for example, by a signal processing process in a hearing device. In this case, the individual frequency bands are specified in particular on the basis of the requirements for the signal processing by frequency band in the hearing device.

Two frequency bands are to be regarded as immediately adjacent in particular if there is no further characteristic frequency of another frequency band between the two characteristic frequencies, which in each case define the position of a frequency band in the frequency space. In particular, a center frequency of a frequency band or a maximum frequency of the magnitude frequency response is used as such a characteristic frequency. The band limit frequency of two immediately adjacent frequency bands should preferably be set so that information about the filter behavior of each of the two frequency bands concerned is provided in the frequency range in which the two frequency bands concerned are adjacent, i.e. in particular in a possible overlapping area. In particular, the band limit frequency is determined as that frequency for which the two immediately adjacent frequency bands have the same magnitude frequency response, or as the arithmetic or geometric mean between the characteristic frequencies determining the two immediately adjacent frequency bands,

Within the scope of the invention, a (first) target frequency, which is a desired limit between two frequency ranges, is initially determined on the basis of the audio signal with different distortion. The first and second frequency bands are then indirectly determined on the basis of this target frequency. Since the target frequency is derived from the properties of the audio signal, this target frequency only coincides exactly with one of the band limit frequencies in exceptional cases. As a rule, it is more or less distant from the nearest band cut-off frequency.

The first target frequency is determined in particular within the framework of the higher-level application of the audio signal, e.g. in the case of a signal processing process in a hearing aid depending on the need for frequency-distorted signals in a specific frequency range occurring in the hearing aid. In this context, the first target frequency is preferably determined in such a way that it meets the requirements for the desired frequency distortion of the audio signal by the higher-level application particularly well, so that in particular the first target frequency for the higher-level application of the audio signal represents a critical value in terms of frequency distortion, at which suitably a change in the frequency distortion of the audio signal has to take place preferentially. In the case of a signal processing process in a hearing aid as a higher-level application, such a critical frequency for the frequency distortion is given, for example, by acoustic feedback on the hearing aid that is to be suppressed, which is preferably to be carried out in the smallest possible frequency range, with frequency distortion being used as part of the suppression of the acoustic feedback becomes. In this case, the critical frequency selected is, for example, the minimum frequency for which acoustic feedback must be suppressed in order to ensure an overall gain of less than one in the closed loop formed from the acoustic feedback path and the signal processing.

The first frequency band and the second frequency band lying directly above the first frequency band are preferably determined solely on the basis of the first target frequency, for example by using those directly as the first frequency band and the second frequency band lying directly above it adjacent frequency bands are selected whose band limit frequency is in particular directly below the first target frequency, i.e. that in particular there is no further band limit frequency of other frequency bands between the first target frequency and the band limit frequency below it, at which the first frequency band and the second frequency band are adjacent . In an alternative embodiment of the invention, further parameters are used in addition to the first target frequency. For example, the respective signal components in the individual frequency bands are also taken into account, and thus only those frequency bands are permitted as the first frequency band and second frequency band for whose signal components a predetermined maximum level is not exceeded. If, in this case, the first target frequency is defined as a maximum critical frequency with regard to frequency distortion as part of the higher-level application of the audio signal, the signal level is taken into account, for example, in such a way that two adjacent frequency bands with a band limit frequency below the first target frequency are determined, the signal components of which not exceed the specified maximum level.

According to the invention, between the splitting of the audio signal into the frequency bands and the frequency distortion described above, there is a frequency-band-specific amplification of the signal components carried in the individual frequency bands. This means that the signal components in the first frequency band or in the second frequency band, to which the mutually different distortion of frequencies is to be applied, are not necessarily identical to the signal components of the audio signal when divided into the individual frequency bands. In addition, further signal processing steps can follow the frequency distortion.

Preferably, the respective distortion of the frequencies is not only limited to the signal components of the first frequency band or the second frequency band, but can also extend to others, from the band limit frequency between the first frequency band and the second frequency band extend distant frequency bands with. This includes in particular the case that a specific distortion of frequencies is applied to signal components in all low frequency bands up to and including the first frequency band, and a different distortion of the frequencies than the previous distortion is applied to signal components of all high frequency bands from the second frequency band inclusive upwards. The "different frequency distortion" of the signal components in the first or second frequency band (and possibly further associated frequency bands) also includes in particular the case that the signal components in one of these two frequency bands (and possibly the associated further frequency bands) are not distorted, see above that the output frequency of this frequency band or these frequency bands corresponds to the respective input frequency.

If an audio signal is to be distorted as a function of frequency, a division into individual frequency bands that takes place as part of the higher-level application of the audio signal can now be used for said frequency distortion, so that the already existing infrastructure of the higher-level application of the audio signal can be used to implement the frequency dependency of the frequency distortion itself can be. On the one hand, this saves resources in the higher-level application and, on the other hand, saves an additional filter process that is independent for the distribution of the frequencies for frequency distortion, whereby additional latencies are avoided.

According to the invention, that frequency band is determined as the first frequency band, the upper band limit frequency of which is formed by the band limit frequency located directly below the first target frequency. The most common implementations of dividing an audio signal into a plurality of predefined frequency bands are designed in such a way that the resulting frequency bands each have an absolute value frequency response with a defined maximum and/or without local minima. For a given frequency band, in particular the range between the two band limit frequencies to the respective immediately adjacent frequency bands is specified as the range of the frequency band in which usually construction-related the magnitude frequency response has its maximum and/or the magnitude frequency response is greater than beyond one of the band limit frequencies. This area is now specifically identified as the core area of the frequency band. The proposed determination of the first frequency band as the frequency band whose upper band limit frequency is formed by the band limit frequency located directly below the first target frequency means that for the configuration of the frequency bands described, the first target frequency is in the core range of the second frequency band.

If the first target frequency is determined as a minimum frequency within the framework of the higher-level application of the audio signal, for which a certain type of frequency distortion is desired, the selection mentioned and the associated classification of the first target frequency in the core range of the second frequency band can be Frequency distortion of the second frequency band can be advantageously achieved in the context of the invention that this desired minimum property of the first target frequency is taken into account in any case.

Expediently, at a point in time after the determination of the first and second frequency band based on the audio signal, a different third frequency band is determined instead of the first frequency band. A different distortion of the frequencies is applied to signal components in this third frequency band than to signal components in a frequency band directly adjacent to (in particular directly above) the third frequency band.

In an expedient embodiment of the invention, a second target frequency is initially determined on the basis of the audio signal instead of the first target frequency. The third frequency band is then indirectly determined on the basis of this target frequency. The second target frequency also generally does not coincide with one of the band limit frequencies, but is regularly spaced more or less from the next band limit frequency.

The determination of the third frequency band (and possibly also the determination of the second target frequency) is carried out here in particular by an ongoing, periodic or event-controlled update within the framework of the higher-level application for the audio signal. The distortion of frequencies of signal components in the third frequency band or the immediately adjacent frequency band takes place in particular analogously to the above-described form of distortion of the frequencies of signal components in the first or second frequency band. In other words, the boundary between two frequency ranges different in terms of frequency distortion is shifted depending on the audio signal by switching frequency bands between different types of frequency distortion.

In particular, a distortion of the frequencies initially adjusted as described above with regard to the signal components in the first frequency band and in the second frequency band can be achieved simply by shifting the application range towards the third frequency band and the frequency band lying directly above the third frequency band. The adaptation of the distortion of frequencies to the second target frequency, which is assigned to a different frequency band and thus a different band limit frequency than the first target frequency in the manner mentioned, makes it possible to react to changed requirements for the frequency distortion of the audio signal in the higher-level application, i.e for example, to changes in a feedback to be suppressed during signal processing in a hearing aid.

It is preferably checked here whether the second target frequency is directly above the upper limit frequency of a further frequency band that is different from the first frequency band, with the further frequency band being determined as the third frequency band as a function of this check, and with different distortion being applied to the signal components in the third frequency band of the frequencies is applied than to the signal components of the frequency band immediately above the third frequency band. As a result, the second target frequency is assigned to the third frequency band in such a way that the core area of the frequency band lying directly above the third frequency band second target frequency includes. This is particularly advantageous when the second target frequency is determined based on the audio signal according to the requirements of the higher-level application as a minimum frequency for a desired frequency distortion. The classification of the second target frequency in the core area of the frequency band immediately above the third frequency band and the corresponding application of the desired frequency distortion at least to said frequency band and, if necessary, to other frequency bands above and exclusively the third frequency band then takes this minimum property of the second target frequency into account.

In an advantageous embodiment of the invention, the distortion of frequencies is given in each case by a shift by an amount that is constant over the frequency range and/or a frequency value that is modulated as a function of time. In particular, the time-dependent modulated frequency value is constant over the frequency. A frequency distortion to be applied in a different way to the signal components of the first frequency band than to the signal components of the second frequency band is then achieved in particular by a difference in the constant amount. In particular, for the signal components of one of the two frequency bands, preferably the first frequency band, the amount of the frequency shift can also be zero within the scope of the invention, so that the relevant frequencies are effectively not shifted.

In the frequency domain, the frequency distortion is correlated with a time-dependent phase modification of the frequency-distorted signal component. In concrete terms, the signal component carried in each of the frequency bands concerned is multiplied in particular by a complex-valued pointer e ^iΔt , as a result of which the frequency distortion is achieved. The variable Δ characterizes the strength of the frequency distortion for the respective frequency band. The quantity t designates the time. If Δ is the same for several frequency bands, this amounts to a constant frequency shift of these frequency bands. A change in the frequency distortion to be applied to the signal component in a frequency band is preferably always carried out in such a way that this change in the frequency distortion does not change the phase of the frequency-distorted signal component or only changes it in jumps (ie changes in leaps and bounds) to an extent below a threshold. In a particularly expedient embodiment of the invention, the change in the frequency distortion is undertaken only at a zero crossing or in a predetermined vicinity of a zero crossing of the phase modification correlated with the distortion. The change in frequency distortion thus occurs only when the phase modification phasor e ^iΔt described above is on or near the real axis of the complex plane (ie for Δ·t ≈ 0,2π,4π,... and e ^iΔt ≈ 1).

In this way, audible artifacts (e.g., "pops") in the frequency-distorted signal as the frequency distortion is changed are advantageously avoided.

In particular, the phase modification of the relevant signal components is checked for a change in a frequency distortion to be applied to the signal components in a frequency band, with a change in the frequency distortion only being permitted at or in the vicinity of the zero crossing of the phase modification. A change in a distortion to be applied to the signal components in a frequency band includes in particular a change such that as a result of an update of the first target frequency towards a second target frequency for signal components of frequency bands whose core range is at least partially between the first target frequency and the second target frequency, the applied distortion of frequencies changes.

In this case, the change can also consist in a complete activation or deactivation of a frequency distortion for one or more frequency bands. Switching off the frequency distortion is expressed numerically in that the vector e ^iΔt presenting the frequency distortion changes to a phase modification term of the value 1. As has been recognized, this transition would then lead to audible artefacts if the pointer e ^iΔt had a value significantly different from 1 at the time of switching off. In order to avoid such artifacts, switching off the frequency distortion in the advantageous embodiment of the Invention only permitted at times when the amount of the product term Δ·t representing the phase modification falls below a predetermined limit value of, for example, π/8 or even π/16.

In a further advantageous embodiment of the invention, the first frequency band is additionally filtered with a low-pass filter and/or the second frequency band is additionally filtered with a high-pass filter. The respective filtering takes place here in particular at the band limit frequency between the first frequency band and the second frequency band. As a result, the overlap between the first frequency band and the second frequency band can be reduced. For signal components from the area of the overlap between the first frequency band and the second frequency band, the respective different distortion of frequencies of signal components of the first frequency band and the second frequency band leads to a superimposition of in a subsequent synthesis and inverse transformation of the frequency-distorted signal from the frequency domain into the time domain two differently frequency distorted contributions of the same signal component. This can lead to audible artefacts and/or beats. A reduction in the overlap precisely in the area in which the signal components are each exposed to different frequency distortions of the individual frequency bands means that such double contributions with different frequency distortions, which originally come from the same signal component, can be significantly suppressed. In this case, the low-pass filter is preferably applied only to the first frequency band and/or the high-pass filter is only applied to the second frequency band. As a result, the additional latency that occurs as a result of the low-pass filter and/or the high-pass filter can be limited to a small frequency range.

The band limit frequency between the first frequency band and the second frequency band is preferably shifted from the value specified by the division of the frequency bands towards the first target frequency by means of the filter characteristic of the low-pass filter and/or by means of the filter characteristic of the high-pass filter. For this purpose, the high-pass filter preferably has a steeper edge on as the low pass filter. As a result, the range in which a desired frequency distortion is applied, and which is achieved by distorting frequencies of signal components in individual frequency bands, is changed by shifting the band limit frequency between the first frequency band and the second frequency band and the associated changes Amount frequency response of the frequency bands in question can be better adapted to a frequency shift desired or required within the framework of the higher-level application of the audio signal, as limited by the first target frequency.

Advantageously, the distortion of frequencies is only applied to signal parts of frequency bands on one side of the band cutoff frequency between the first frequency band and the second frequency band. On the one hand, this is particularly easy to implement in terms of signal processing technology. On the other hand, in many applications it is important to apply frequency distortion to as small a range of the audio signal as possible, with boundary conditions prescribing a minimum range for frequency distortion of the audio signal. In this case, the distortion of frequencies is only applied to signal components of those frequency bands in which frequency distortion is considered desirable or necessary.

An embodiment of the invention is also a method for suppressing acoustic feedback in a hearing aid, with an input converter of the hearing aid generating an input signal from a sound signal from the environment, with an intermediate signal being generated on the basis of the input signal, which is subjected to signal processing with a filter bank for dividing the frequency band by frequency intermediate signal is supplied, with an output signal being generated from a frequency-distorted signal, which output signal is converted into an output sound signal by an output converter of the hearing device, with the frequency-distorted signal being used to suppress acoustic feedback in the hearing device that occurs as a result of the output sound signal being coupled into the input converter, and with to the intermediate signal according to the invention as described above Method for frequency distortion is applied, and thereby the frequency-distorted signal is generated.

An input converter generally includes an acousto-electric converter, which is set up to convert the sound signal from the environment into a corresponding electrical or electromagnetic signal, ie a microphone, for example. An output transducer generally includes an electro-acoustic transducer which is set up to generate an output sound signal from an electrical and/or electromagnetic signal, ie for example a loudspeaker or a sound generator for bone sound conduction. In this context, signal processing is to be understood in particular as a processing of the input signal or of a signal derived from the input signal, ie in particular a frequency band-dependent amplification and/or noise suppression.

Generating the intermediate signal based on the input signal means in particular that the signal processing receives a signal that is directly dependent on the input signal, for example the input signal that has been corrected by a compensation signal to compensate for acoustic feedback. The method for frequency distortion can then be applied to the intermediate signal in particular in such a way that the intermediate signal is divided into individual predefined frequency bands at the filter bank of the signal processing unit, and after a frequency band-dependent processing of the signal components in the individual frequency bands by the signal processing, the different distortion of frequencies the processed signal components in the first frequency band or in the second frequency band is applied in order to generate the frequency-distorted signal. From this, the output signal is then generated, among other things, by synthesizing the individual frequency band components. The feedback can then be suppressed by an adaptive filter based on the frequency-distorted signal, ie in particular also by the output signal as a reference variable of the adaptive filter, via a corresponding compensation signal.

The advantages specified for the method for frequency distortion of an audio signal and its developments can be transferred analogously to the method for suppressing acoustic feedback in a hearing device.

The invention also relates to a hearing aid, comprising an input converter for generating an input signal from a sound signal from the environment, and a signal processing unit with a filter bank for dividing an audio signal derived from the input signal using the input signal and a control unit which is set up to implement the method described above to distort an audio signal. The advantages specified for the method and for its further development can be transferred to the hearing aid. In particular, the signal processing unit and filter bank are parts of the control unit. In this case the audio signal is an intermediate signal in the control unit.

FIG. 1

in einem Blockdiagramm ein Verfahren zur Unterdrückung einer akustischen Rückkopplung in einem Hörgerät,

FIG. 2

in einem Blockdiagramm ein Verfahren zur Frequenzverzerrung nach FIG. 1,

FIG. 3

den Frequenzgang einer Filterbank im Verfahren nach FIG. 2, und

FIG. 4

den durch Hochpass- und Tiefpassfilter angepassten Frequenzgang zweier benachbarter Frequenzbänder in der Filterbank nach FIG. 3.

An exemplary embodiment of the invention is explained in more detail below with reference to a drawing. Here each show schematically:

FIG. 1

in a block diagram a method for suppressing acoustic feedback in a hearing device,

FIG. 2

in a block diagram a method for frequency distortion FIG. 1 ,

FIG. 3

the frequency response of a filter bank in the process FIG. 2 , and

FIG. 4

the frequency response of two adjacent frequency bands in the filter bank, adjusted by high-pass and low-pass filters FIG. 3 .

Corresponding parts and sizes are provided with the same reference symbols in all figures.

In figure 1 a method 1 for suppressing acoustic feedback g in an acoustic system is shown schematically in a block diagram. In the present case, the acoustic system is given by a hearing aid 2. The hearing aid 2 comprises an input converter 4, which generates an input signal 8 from a sound signal 6 of the environment, and in the present case is given by a microphone. A compensation signal 10 is subtracted from the input signal 8 and is generated in an electrical feedback loop 12 in a manner to be described below. The intermediate signal 14 resulting from the input signal 8 and the compensation signal 10 is fed to a signal processing unit 16 in which the signal processing processes specific to the hearing aid 2 are carried out (in particular a frequency band-dependent amplification of the intermediate signal 14). For this purpose, the signal processing 16 includes a filter bank 18, on which the intermediate signal is divided into individual frequency bands, which are then processed in a user-specific manner. The signal processor 16 now outputs a processed signal 20 resolved by frequency band, to which a frequency distortion 22 is applied in a method that is yet to be described. The frequency-distorted signal 24 in the time-frequency domain resulting from the frequency distortion 22 is now converted in a synthesis filter bank 26 into a broadband output signal 28 in the time domain, which in turn is converted into an output sound signal 32 by an output converter 30. In the present case, the output converter 30 is provided by a loudspeaker.

On the other hand, the output signal 28 is branched off into the electrical feedback loop 12 and fed there to an adaptive filter 34, which also receives the intermediate signal 14 as a further input variable as an error signal, and from this generates the compensation signal 10 for suppressing the acoustic feedback g. Due to the frequency distortion 22, the output signal 28 is decorrelated from the input signal 8 and thus also from the intermediate signal 14, so that when the error signal 14 is re-entered into the adaptive filter 34, the latter is not fully adapted to the tonal signal components of the output signal 28. This can cause artifacts to form in the output signal 28 and thus avoided in the output sound signal 32. The suppression of the acoustic feedback g by the compensation signal 10 can in particular remain limited to specific frequency ranges, ie in this case the compensation signal 10 only has significant signal components for said frequency bands, in particular for those to which the frequency distortion 22 was applied.

In FIG. 2 1 is a schematic block diagram of the sequence of a method 40 for frequency distortion 22 of the intermediate signal 14 FIG. 1 shown. In this case, the intermediate signal 14 forms the audio signal 42 , which acts as the input variable relevant to the method 40 . In a first step S1, the audio signal 42 is used to check the frequency range in which acoustic feedback g from the output transducer 30 to the input transducer 8 of the hearing aid 2 is to be suppressed, and the frequency range in which there are also tonal signal components in the audio signal 42 that are present when the feedback is suppressed in the adaptive filter 34 may lead to artifacts. The check with regard to the acoustic feedback g to be suppressed can be carried out by the adaptive filter 34, with regard to the tonality of the signal components preferably by the signal processor 16. A first target frequency tf1 is then defined as a function of the results of these checks. In this case, the target frequency tf1 is determined in particular as the minimum frequency above which a frequency distortion is required for an effective suppression of the acoustic feedback.

In a next step S2, the audio signal 42 is now divided into individual frequency bands in a filter bank 18 . Step S2 can also include further sub-steps, such as frequency-band-dependent processing of the signal components 44 in the frequency bands generated, which, however, do not impair the course of the method 40 per se.

In a further step S3, a first frequency band FB1 is now determined on the basis of the first target frequency tf1. The first frequency band FB1 is given here as that frequency band whose upper band limit frequency is defined by the band limit frequency immediately below the first target frequency tf1 is formed, the upper band limit frequency being given by that frequency at which the magnitude frequency response of the first frequency band is equal to the magnitude frequency response of the frequency band immediately above the first frequency band FB1. The frequency band immediately above the first frequency band FB1 is set as the second frequency band FB2.

In a next step S4, a low-pass filter TP is placed over the first frequency band FB1 at its band limit frequency to the second frequency band FB2, and a high-pass filter HP is placed over the second frequency band FB2 at the same band limit frequency. On the one hand, this further reduces the overlap between the first frequency band FB1 and the second frequency band FB2 than is provided by the filter bank 18, and on the other hand, the band limit frequency can easily be shifted to the first target frequency tf1 to be shifted towards.

In step S5, frequency distortion 22 in the form of a frequency shift 46 by a time-constant amount Δ is now applied to the signal components 44 in all frequency bands from the second frequency band FB2 upwards, while the signal components 44 remain unchanged in all frequency bands from the first frequency band FB1 downwards , and thus the frequency-distorted signal 24 is generated. The method 40 also returns to step S1 with the specification of the first frequency band FB1, and updates the first target frequency continuously, periodically or event-controlled in order to, in the event of a significant change in the acoustic feedback g, which results in the first target frequency tf1 being outside the first Frequency band FB1 is to determine a third frequency band FB3, which takes the place of the first frequency band FB1 to continue the method 40 analogously.

In FIG. 3 the frequency response of a filter bank 18 is plotted against a frequency f. The individual frequency bands FB have a non-negligible overlap OV with the respectively adjacent frequency band, where two immediately adjacent frequency bands define a band limit frequency fL0 to fL3, which is given by that frequency at which the magnitude frequency response of the two adjacent frequency bands is the same. According to step S1 of method 40 after FIG. 2 the first target frequency tf1 is now specified, and on the basis of this the first frequency band FB1 is determined as that frequency band whose upper band limit frequency fL1 is formed by the band limit frequency immediately below the first target frequency tf1. As described above, a low-pass filter TP is placed over the first frequency band FB1 at the upper band limit frequency fL1, and a high-pass filter HP is placed over the second frequency band FB2 at the same band limit frequency fL1, which therefore limits the second frequency band FB2 downwards. This reduces the overlap OV1 between the first frequency band FB1 and the second frequency band FB2. Due to the fact that the two said filters TP, HP are only applied to one frequency band each, the latency that may be generated as a result is limited spectrally to the relevant frequency bands. In order to keep the additional latency as low as possible, only one complex-valued zero (filter order 1) is preferably inserted. The signal components of the audio signal 42 in the frequency bands above the upper band limit frequency fL1 of the first frequency band FB1, ie in the frequency bands from FB2 upwards, are then shifted by a constant amount.

Changes after a certain time, the acoustic feedback path, which the acoustic feedback g in the hearing aid 2 after FIG. 1 conditionally, the first target frequency tf1 is updated according to a second target frequency tf2 adapted to the change. It is now checked whether the second target frequency tf2 still corresponds to the band limit frequency fL1 between the first frequency band FB1 and the second frequency band FB2, i.e. whether the band limit frequency fL1 also forms the band limit frequency immediately below the second target frequency tf2. In that case, the frequency shift can continue to be applied unchanged to the signal components, preferably of all frequency bands from the second frequency band FB2 upwards (shaded area).
preferably applied to the signal components of all frequency bands from the second frequency band FB2 upwards (shaded area).

In the present case, however, this is not the case; the second target frequency is now above the band limit frequency fL3, which limits a frequency band that differs from the first frequency band at the top to the immediately adjacent frequency band. The frequency band limited from above by the band limit frequency fL3 is now defined as the third frequency band FB3, and now the frequency shift for signal components preferably of all frequency bands above and excluding the third frequency band FB3 in the manner already described, in particular using appropriate high-pass or low-pass filters on the Band cutoff frequency fL3 (crosshatched area).

In FIG. 4 is the magnitude frequency response of the first frequency band FB1 and of the second frequency band FB2 FIG. 3 plotted at the band limit frequency fL1 against a frequency f. The dotted lines each show the absolute frequency response of the frequency bands FB1, FB2, as specified by the higher-level filter bank in the range of the band limit frequency fL1. By means of a low-pass filter or a high-pass filter applied to the first frequency band or the second frequency band, the overlap OV1 can be reduced in the range of the band limit frequency fL1 (OV1′, dashed lines). If a low-pass filter and a high-pass filter with different, in particular asymmetrical, filter characteristics are used here, then in addition to the reduced overlap OV1', the band limit frequency fL1 can also be shifted slightly to an adapted band limit frequency fL1', for example in the direction of the first target frequency.

Reference List

1

Method for suppressing feedback

2

hearing aid

4

input converter

6

sound signal

8th

input signal

10

compensation signal

12

electrical feedback loop

14

intermediate signal

16

signal processing

18

filter bank

20

processed signal

22

frequency distortion

24

frequency distorted signal

26

synthesis filter bank

28

output signal

30

output converter

32

output sound signal

34

adaptive filters

40

Method of frequency distortion

42

audio signal

44

signal portion

46

frequency shift

FB

frequency band

FB1

first frequency band

FB2

second frequency band

FB3

third frequency band

fL0-fL3

band cutoff frequency

fL1'

adjusted band cutoff frequency

G

acoustic feedback

HB

high pass filter

OV

overlap

OV1

overlap

OV1'

adjusted overlap

S1-S5

process step

TP

low pass filter

Δ

constant frequency magnitude

Claims (12)

1. A method (40) for frequency distortion (22) of an audio signal (42) in a signal processing process in a hearing aid (2),

   wherein the audio signal (42) is divided into a plurality of predetermined frequency bands (FB, FB1, FB2, FB3) by means of a filter bank, wherein each two directly adjacent frequency bands (FB1, FB2, FB3) are determining a band limit frequency (fL0-fL3),

   wherein signal components (44) in the individual frequency bands (FB, FB1, FB2, FB3) are amplified in a frequency band-specific manner,

   wherein at first, a target frequency (tf1) for a boundary between two frequency ranges with different distortion (22) of the frequencies is determined on the basis of the audio signal (42), said target frequency (tf1) being determined as a critical frequency which is given by an acoustic feedback (g) to be suppressed at the hearing aid (2), and wherein a first frequency band (FB1) and a second frequency band (FB2) lying directly above the first frequency band (FB1) are determined on the basis of the target frequency (tf1),

   wherein the first frequency band (FB1) is being determined as the frequency band with its upper band limit frequency being formed by the band limit frequency (fL1) located directly below the target frequency (tf1),

   wherein to signal components (44) in the first frequency band (FB1), it is applied a different distortion (22) of the frequencies than to signal components (44) in the second frequency band (FB2), thereby generating a frequency-distorted signal (24), and

   wherein the frequency band-specific amplification of the signal components (44) in the individual frequency bands (FB, FB1, FB2, FB3) takes place between the splitting of the audio signal into the frequency bands (FB, FB1, FB2, FB3) and the respective distortion (22) of the frequencies.
2. The method (40) according to claim 1,

   wherein at an instant of time after the determination of the first frequency band (FB1), a third frequency band (FB3) different from the first frequency band (FB1) is determined from the audio signal (42), and

   wherein to signal components (44) in the third frequency band (FB3), it is applied a different distortion (22) of the frequencies than to signal components (44) in a frequency band directly adjacent to the third frequency band (FB3).
3. The method (40) according to claim 2,
   wherein, in order to determine the third frequency band (FB3), at first a second target frequency (tf2) different from the first target frequency (tf1) is determined on the basis of the audio signal (42), and wherein the third frequency band (FB3) is determined on the basis of the second target frequency (tf2).
4. The method (40) according to claim 3,

   wherein it is checked whether the second target frequency (tf2) lies directly above the upper band limit frequency (fL3) of a further frequency band (FB3) different from the first frequency band (FB1),

   wherein the further frequency band is being determined as the third frequency band (FB3) in dependence on this check, and

   wherein to the signal components (44) in the third frequency band (FB3), it is applied a different distortion (22) of the frequencies than to the signal components (44) of the frequency band lying directly above the third frequency band (FB3).
5. The method (40) according to one of the preceding claims,
   wherein respective the distortion (22) of frequencies is given by a shift (46) by an amount (Δ) which is constant over the frequency and/or a frequency value which is modulated as a value of a time-dependent function.
6. The method (40) according to one of the preceding claims,
   wherein a change in the distortion (22) of the frequencies to be applied to the signal component (44) in a frequency band (FB2) is always carried out in such a way that, as a result of the change in the frequency distortion, the phase of the frequency-distorted signal component (44) does not jump or jumps only to an extent which is below a limit value.
7. The method (40) according to any one of the preceding claims,
   wherein a change in the distortion (22) of the frequencies to be applied to the signal component (44) in a frequency band (FB2) is made only in a zero crossing or in a predetermined vicinity of a zero crossing of a phase modification of the frequency-distorted signal component (44) correlated with the distortion.
8. The method (40) according to any one of the preceding claims,

   wherein the first frequency band (FB1) is additionally filtered with a low-pass filter (TP), and/or

   wherein the second frequency band (FB2) is additionally filtered with a high-pass filter (HP).
9. The method (40) according to claim 8,
   wherein the band limit frequency (fL1) between the first frequency band (FB1) and the second frequency band (FB2) is shifted from the value predetermined by the division of the frequency bands (FB, FB1, FB2, FB3) towards the first target frequency (tf1) by means of the filter characteristic of the low-pass filter (TP) and/or by means of the filter characteristic of the high-pass filter (HP).
10. The method (40) according to any one of the preceding claims,
    wherein the distortion (22) of frequencies is applied only to signal portions (44) of frequency bands (FB2, FB3) on one side of the band limit frequency (fL1) between the first frequency band (FB1) and the second frequency band (FB2).
11. A method (1) for suppressing acoustic feedback (g) in a hearing aid (2), wherein an input transducer (4) of the hearing aid (2) generates an input signal (8) from a sound signal (6) of the environment,

    wherein an intermediate signal (14) is generated on the basis of the input signal (8), said intermediate signal (14) being fed to a signal processing unit (16) with a filter bank (18) for frequency-band splitting of the intermediate signal (14),

    wherein an output signal (28) is generated from a frequency-distorted signal (24), said output signal (28) being converted into an output sound signal (32) by an output transducer (30) of the hearing aid (2),

    wherein acoustic feedback (g) occurring in the hearing aid (2) due to coupling of the output sound signal (32) into the input transducer (4) is suppressed by means of the frequency-distorted signal (24), and

    wherein the method (40) for frequency distortion (22) according to one of the preceding claims is applied to the intermediate signal (14), thereby producing said frequency-distorted signal (24).
12. A hearing aid (2) comprising an input transducer (4) for generating an input signal (8) from an ambient sound signal (6), a signal processing unit (16) having a filter bank (18) for dividing an audio signal (42) derived from the input signal (8) on the basis of the input signal (8), and a control unit adapted to perform the method according to any one of claims 1 to 10.

Images