A HEARING AID COMPRISING AN OPEN LOOP GAIN ESTIMATOR

Abstract

A hearing aid comprises a forward path comprising an input transducer an electric input signals representing a sound comprising target signal components and background noise, a hearing aid processor for providing a processed signal in dependence of said at least one electric input signal and for providing a processed output signal in dependence thereof, and an output transducer for providing stimuli perceivable as sound to the user in dependence of said processed signal. The forward path provides a frequency dependent intended forward path transfer function. The hearing aid further comprises a feedback path estimator configured to provide a current frequency dependent estimate of a feedback path transfer function of a feedback path from the output transducer to the input transducer, and a current feedback path estimate in dependence of the current estimate of the feedback path transfer function and of the processed signal, and a combination unit in the forward path configured to subtract the current feedback path estimate from a signal of the forward path to provide a feedback corrected signal. The hearing aid may further comprise a noise estimator configured to provide a current frequency dependent noise estimate representing a background noise level in the at least one electric input signal, an open loop transfer function estimator configured to provide a frequency dependent estimate of a current open loop transfer function in dependence of the intended forward path transfer function and the current estimate of the feedback transfer function, and a confidence level estimator configured to provide a current frequency dependent estimate of a confidence level of the current estimate of the feedback transfer function in dependence of a current estimate of open loop gain and optionally the current noise estimate. The hearing aid may be configured to control processing in the hearing aid in a frequency band k in dependence of said current estimate of the open loop transfer function and/or the current estimate of the feedback path transfer function, if the current estimate of the confidence level fulfils a confidence criterion in said frequency band k. A method of operating a hearing aid is further disclosed. The invention may e.g. be used to assess a risk of acoustic feedback in a hearing aid.

Claims

1. A hearing aid adapted to be worn by a user, or for being partially or fully implanted in the head of the user, comprising a forward path comprising at least one input transducer for converting a sound to corresponding at least one electric input signals representing said sound, the sound comprising target signal components and background noise, a hearing aid processor for providing a processed signal (U) in dependence of said at least one electric input signal, and an output transducer for providing stimuli perceivable as sound to the user in dependence of said processed signal, wherein the forward path is configured to provide a frequency dependent intended forward path transfer function (F(k)) from said input transducer to said output transducer in dependence of said at least one electric input signal and of a hearing impairment of said user, k being a frequency band index, and wherein the hearing aid further comprises a feedback control system comprising a feedback path estimator configured to provide a current frequency dependent closed loop estimate (Ĥ(k)) of a feedback path transfer function (H) of a feedback path from said output transducer to said input transducer in a closed loop configuration of the forward path, and a current feedback signal estimate ({circumflex over (V)}) in dependence of said current closed loop estimate (Ĥ(k)) of the feedback path transfer function and of said processed signal (U), a combination unit in the forward path configured to subtract said current feedback path estimate ({circumflex over (V)}) from a signal of the forward path to provide a feedback corrected signal (E), an open loop transfer function estimator configured to provide a frequency dependent estimate ({circumflex over (L)}(k)) of a current open loop transfer function in dependence of said intended forward path transfer function (F(k)) and said current closed loop estimate (Ĥ(k)) of the feedback transfer function, and a confidence level estimator configured to provide a current frequency dependent estimate of a confidence level ({circumflex over (P)}(k)) of said current closed loop estimate (Ĥ) of the feedback transfer function in dependence of the estimate of the current open loop transfer function ({circumflex over (L)}(k)), and wherein the hearing aid is configured to control processing in the hearing aid in a frequency band k in dependence of said estimate ({circumflex over (L)}(k)) of the current open loop transfer function, if said current estimate ({circumflex over (P)}(k)) of a confidence level fulfils a confidence criterion in said frequency band k.

2. A hearing aid according to claim 1 wherein the estimate ({circumflex over (L)}(k)) of the current open loop transfer function is approximated by its magnitude, open loop gain (|{circumflex over (L)}(k)|).

3. A hearing aid according to claim 1 wherein said confidence criterion comprises that said current estimate ({circumflex over (P)}(k)) of a confidence level is above a threshold level (P.sub.th) in said frequency band k.

4. A hearing aid according to claim 1 wherein the confidence level P.sub.th is frequency dependent (P.sub.th(k)).

5. A hearing aid according to claim 1 comprising a noise estimator configured to provide a current frequency dependent noise estimate ({circumflex over (N)}(k)) representing a background noise level in the at least one electric input signal, and wherein said confidence level estimator is configured to provide said current frequency dependent estimate of a confidence level ({circumflex over (P)}(k)) of said current closed loop estimate (Ĥ) of the feedback transfer function in dependence of said estimate of the current open loop transfer function ({circumflex over (L)}(k)) and said current noise estimate ({circumflex over (N)}(k)).

6. A hearing aid according to claim 1 wherein the confidence level estimator (CLE) is configured to provide that said current frequency dependent estimate of the confidence level ({circumflex over (P)}(k)) of said current closed loop estimate (Ĥ) of the feedback transfer function is further determined in dependence of one or more of a) the intended forward path transfer function (F(k)), b) a style of the hearing aid, and c) a characteristic of the at least one electric input signal.

7. A hearing aid according to claim 6 wherein the characteristics of the at least one electric input signal comprises, or is dominated by, one or more of white noise, coloured noise, speech, and music.

8. A hearing aid according to claim 1 wherein said confidence criterion in said frequency band k is fulfilled if said estimate of current open loop gain (|{circumflex over (L)}(k)|) is above a minimum value and below a maximum value.

9. A hearing aid according to claim 8 wherein said confidence criterion in said frequency band k is fulfilled if a current noise estimate ({circumflex over (N)}(k)) is in a range above a minimum value and below a maximum value.

10. A hearing aid according to claim 9 wherein said maximum value of said current noise estimate ({circumflex over (N)}(k)) is smaller than or equal to 65 dB SPL, and said minimum value of said estimate of current open loop gain (|{circumflex over (L)}(k)|) is larger than or equal to −40 dB.

11. A hearing aid according to claim 9 or wherein said maximum value of said current noise estimate ({circumflex over (N)}(k)) is smaller than or equal to 90 dB SPL, and said minimum value of said estimate of current open loop gain (|{circumflex over (L)}(k)|) is larger than or equal to 0 dB.

12. A hearing aid according to claim 1 wherein said control of processing in the hearing aid comprises assessing a risk of acoustic feedback.

13. A hearing aid according to claim 1 wherein said control of processing in the hearing aid comprises changing the frequency dependent intended forward path transfer function, e.g. to control feedback.

14. A hearing aid according to claim 1 wherein said control of processing in the hearing aid comprises changing a parameter of the feedback control system, e.g. an adaptation rate of an adaptive algorithm of the feedback control system.

15. A hearing aid according to claim 1 being constituted by or comprising an air-conduction type hearing aid, a bone-conduction type hearing aid, a cochlear implant type hearing aid, or a combination thereof.

16. A method of operating a hearing aid adapted to be worn by a user, or for being partially or fully implanted in the head of the user, the hearing aid comprising a forward path comprising at least one input transducer for converting a sound to corresponding at least one electric input signals representing said sound, the sound comprising target signal components and background noise, a hearing aid processor for providing a processed signal (U) in dependence of said at least one electric input signal, and an output transducer for providing stimuli perceivable as sound to the user in dependence of said processed signal, the method comprising providing by the forward path a frequency dependent intended forward path transfer function (F(k)) from said input transducer to said output transducer in dependence of said at least one electric input signal and of said user, k being a frequency band index, and providing a current frequency dependent closed loop estimate (Ĥ(k)) of a feedback path transfer function (H) of a feedback path from said output transducer to said input transducer in a closed loop configuration of the forward path, and providing a current feedback path estimate ({circumflex over (V)}) in dependence of said current closed loop estimate (Ĥ(k)) of the feedback path transfer function and of said processed signal (U), subtracting said current feedback path estimate ({circumflex over (V)}) from a signal of the forward path to provide a feedback corrected signal (E), providing a frequency dependent estimate ({circumflex over (L)}(k)) of a current open loop transfer function in dependence of said intended forward path transfer function (F(k)) and said current closed loop estimate (Ĥ(k)) of the feedback transfer function, and providing a current frequency dependent estimate of a confidence level ({circumflex over (P)}(k)) of said current closed loop estimate (Ĥ) of the feedback transfer function in dependence of the estimate of the current open loop transfer function ({circumflex over (L)}(k)), and controlling processing in the hearing aid in a frequency band k in dependence of said estimate ({circumflex over (L)}(k)) of the current open loop transfer function, if said current estimate ({circumflex over (P)}(k)) of a confidence level fulfils a confidence criterion in said frequency band k.

17. A method according to claim 16 wherein said confidence criterion comprises that said current estimate ({circumflex over (P)}(k′)) of a confidence level is above a threshold level (P.sub.th) in said frequency band k.

18. A method according to claim 16 comprising providing a current frequency dependent noise estimate ({circumflex over (N)}(k)) representing background noise in the at least one electric input signal, and providing said current frequency dependent estimate of a confidence level ({circumflex over (P)}(k)) of said current closed loop estimate (Ĥ) of the feedback transfer function in dependence of said estimate of the current open loop transfer function ({circumflex over (L)}(k)) and said current noise estimate ({circumflex over (N)}(k)).

19. Use of a hearing aid as claimed in claim 1.

20. A data processing system comprising a processor and program code means for causing the processor to perform the method of claim 16.

21. A non-transitory computer readable medium storing a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of claim 16.

22. A method of providing a reliable and accurate estimate of an open loop transfer function acquired in a closed loop setup of a hearing aid without using probe noise, the method comprising: Estimating a feedback path transfer function using an adaptive filter of an acoustic feedback control system in a closed loop setup; Combining the estimate of the feedback path transfer function with a known hearing aid forward path transfer function to get open loop transfer function estimate; Determining at which frequencies the estimated open loop transfer function is trustful and which can then be used for further processing in the hearing aid; Applying the determined trustworthy open loop transfer function for an application in the hearing aid.

23. A method according to claim 22 wherein said application in the hearing aid comprises assessing a risk of acoustic feedback and subsequent control of processing in the hearing aid in dependence of said assessment.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0129] The aspects of the disclosure may be best understood from the following detailed description taken in conjunction with the accompanying figures. The figures are schematic and simplified for clarity, and they just show details to improve the understanding of the claims, while other details are left out. Throughout, the same reference numerals are used for identical or corresponding parts. The individual features of each aspect may each be combined with any or all features of the other aspects. These and other aspects, features and/or technical effect will be apparent from and elucidated with reference to the illustrations described hereinafter in which:

[0130] FIG. 1 illustrates that measuring in an open loop setup is dependent on the level of background noise and less dependent on the loop gain,

[0131] FIG. 2 illustrates that measuring in a closed loop setup is dependent on the loop gain and less dependent on the level of background noise,

[0132] FIG. 3A shows a first embodiment of a block diagram for open loop feedback path estimation using a probe signal comprising one or more sine tones, and

[0133] FIG. 3B shows a first embodiment of a block diagram for open loop feedback path estimation using a probe signal, wherein the probes signal generator is controlled by frequency sub-band levels of the band-split microphone signal (y(f)),

[0134] FIG. 4A shows a first embodiment of a block diagram for closed-loop feedback path estimation using frequency shift of the processed output signal, and

[0135] FIG. 4B shows a second embodiment of a block diagram for closed-loop feedback path estimation using the addition of a probe signal to the processed output signal,

[0136] FIG. 5 shows an embodiment of hearing aid according to the present disclosure.

[0137] The figures are schematic and simplified for clarity, and they just show details which are essential to the understanding of the disclosure, while other details are left out. Throughout, the same reference signs are used for identical or corresponding parts.

[0138] Further scope of applicability of the present disclosure will become apparent from the detailed description given hereinafter. However, it should be understood that the detailed description and specific examples, while indicating preferred embodiments of the disclosure, are given by way of illustration only. Other embodiments may become apparent to those skilled in the art from the following detailed description.

DETAILED DESCRIPTION OF EMBODIMENTS

[0139] The detailed description set forth below in connection with the appended drawings is intended as a description of various configurations. The detailed description includes specific details for the purpose of providing a thorough understanding of various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. Several aspects of the apparatus and methods are described by various blocks, functional units, modules, components, circuits, steps, processes, algorithms, etc. (collectively referred to as “elements”). Depending upon particular application, design constraints or other reasons, these elements may be implemented using electronic hardware, computer program, or any combination thereof.

[0140] The electronic hardware may include micro-electronic-mechanical systems (MEMS), integrated circuits (e.g. application specific), microprocessors, microcontrollers, digital signal processors (DSPs), field programmable gate arrays (FPGAs), programmable logic devices (PLDs), gated logic, discrete hardware circuits, printed circuit boards (PCB) (e.g. flexible PCBs), and other suitable hardware configured to perform the various functionality described throughout this disclosure, e.g. sensors, e.g. for sensing and/or registering physical properties of the environment, the device, the user, etc. Computer program shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executables, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0141] The present application relates to the field of hearing aids. In the present disclosure, a procedure/algorithm to estimate/monitor the open loop transfer function in a hearing aid system based on its acoustic feedback cancellation system using adaptive filters is described.

[0142] The open loop transfer function estimate can be used for several applications, including control of hearing aids based on feedback risks and detection of correct placement of ear moulds, etc.

[0143] In case the open loop transfer function estimate indicates a high risk of feedback (e.g. with a view to the current forward transfer function, and/or the current feedback path estimate), some actions can be carried out, such as, [0144] To restrict the beamformer activities: beam patterns may/should be restricted in order to limit any further change of the forward path transfer function that would lead to a more critical open loop transfer function and hence even higher feedback risks. The restrictions can be slowing down and/or stopping the adaptations of beam patterns, or to enforce the beam pattern to one specific pattern that would give less feedback risks. [0145] To control the noise reduction activities: the noise reduction system, and in particular how fast the noise reduction system releases the gain when applying noise reduction, can be restricted to a lower level. [0146] To control the compression scheme: the hearing loss compression scheme can be restricted to provide less amplification and/or slower gain releasing in order to reduce feedback risks. [0147] To control the feedback control system: the feedback control system can/should be configured to react and remove the feedback as quick as possible. This can be done by increase the adaptation speed of the adaptive algorithm for feedback cancellation.

[0148] All the above-mentioned activities can be applied at selected frequencies.

[0149] To estimate the open loop transfer function (OLTF), we need to know the forward path (FPTF) and feedback path (FBPTF) transfer functions. The forward path transfer function is known in a hearing aid application, whereas the feedback transfer function is unknown. An estimate of the feedback path transfer function can be obtained, though, using an adaptive filter from a feedback cancellation system in a hearing aid application.

[0150] Combining the two provides an estimate of the open loop transfer function, e.g. OLTF=FPTF+FBPTF, in a logarithmic representation. The estimation of the unknown feedback path using adaptive filters can be done in an open loop setup, or in a closed loop setup.

[0151] The feedback path estimation in an open loop setup using a probe signal (cf. e.g. FIG. 3A, 3B) is much easier to handle than in a closed loop setup without using probe signal (cf. e.g. FIG. 4A), and it generally provides much better accuracy, as the estimation condition is more under control. Feedback estimation in a closed loop, but including an added probe signal, may be used as a compromise (cf. e.g. FIG. 4B).

[0152] On the other hand, the estimation in a closed loop without using probe signal can be more preferable in a hearing aid application, as the measurement does not disturb hearing aid users at all, whereas the estimation in open loop using probe signal needs to replace the desired hearing aid output signal with an undesired (and, from a technical point of view, preferably loud) probe signal. This becomes especially annoying and even unacceptable if conducted as a continuous measurement.

[0153] As the preferred open loop transfer function estimate acquired in the closed loop without using probe noise can be inaccurate, we would prefer to determine when the estimate provides a reliable and accurate estimate before it can be used.

[0154] In the present disclosure, a procedure, with the following steps, to obtain and apply a reliable and accurate estimate of open loop transfer function acquired in a closed loop setup of a hearing aid without using probe noise is presented: [0155] S1. Estimate feedback path transfer function using the adaptive filters from the acoustic feedback cancellation system in a closed loop setup; [0156] S2. Combine the above estimate to the known hearing aid forward transfer function to get open loop transfer function estimate; [0157] S3. Determine at which frequencies the estimated open loop transfer function is trustful and which can then be used for further processing; [0158] S4. Apply the determined trustworthy open loop transfer function for an application in the hearing aid, e.g., processing, such as assessing the risk of acoustic feedback and subsequent control of processing in the hearing aid.

[0159] This procedure is slightly similar to procedures described in US20130170660A1 and EP2613567A1. A significant difference is provided by the above step S3, however.

[0160] An important difference to these disclosures is that we here estimate an open-loop transfer function in a closed-loop setup without using a probe signal. In the disclosures of US20130170660A1 and EP2613567A1, the feedback path was estimated, and mainly in an open-loop setup by using additional probe signals.

[0161] Closed-loop measurement without using probe signal has also been mentioned as an option in the disclosures of US20130170660A1 and EP2613567A1, without details on how to make them to work in a closed-loop setup, though. This is the part that is provided in step S3 of the above procedure. Without this important step, the estimation of open loop transfer function in a closed-loop setup without probe noise is NOT ALWAYS reliable and accurate, as illustrated below.

[0162] FIG. 1 shows Measuring in an open loop setup is dependent on the level of background noise and less dependent on the loop gain. FIG. 1 below shows that when measuring feedback path in an open loop setup using probe noise, a very important factor for its accuracy is the background noise, and it is more or less independent of open loop gain. For relatively low background noise level, typically below 60-70 dB SPL, an accurate estimate of the feedback path is possible over a wide range of open loop gain (e.g. between −50 dB and +50 dB) (given a reasonable measurement time, e.g. within few seconds).

[0163] A higher/lower background noise level would make the estimate less/more accurate, and this can be compensated by smaller/bigger step size in the adaptive algorithm. If we e.g. lower the step size by a factor of 2, we can allow 6 dB louder background noise and still have the same accuracy after the convergence, but the measurement would take twice the amount of time. A theoretical study is e.g. provided in [Guo et al., 2011].

[0164] Given a measurement time of few seconds, it turns out that we can accept a background noise level of 60-70 dB SPL to achieve our desired accuracy.

[0165] However, it is different when measuring feedback path in a closed loop without probe noise. It is therefore important to understand when the estimate is reliable and hence can be trusted.

[0166] FIG. 2 illustrates that measuring in a closed loop setup is dependent on the loop gain and less dependent on the level of background noise. FIG. 2 illustrates the measurement of feedback path is indeed highly dependent on the open loop gain, and it is accurate when the open loop gain is around and above 0 dB (given a reasonable measurement time, e.g. within few seconds), and it is practically speaking independent of background noise (e.g. preferably <100 dB SPL for audibility reason).

[0167] In other words, we can trust the feedback path estimate and hence the estimated loop gain when the loop gain estimate is indeed high.

[0168] When the loop gain is high enough, the feedback signal appears to be strong compared to the background noise (which is the input signal in this case), and hence a good feedback-to-noise ratio.

[0169] The background noise level (˜65 dB) and open loop gain value (˜0 dB) for getting accurate estimation in FIG. 1 and FIG. 2 can change, depending on the duration of the estimation. However, the dependency on background noise and loop gain would not change for measuring in open loop and closed loop setup.

[0170] This is the motivation behind the qualification/selection process in the above step 3, to determine which frequency regions the loop gain estimate is indeed high, and only in these regions the feedback path/open loop transfer function estimates can be used. After that, both the critical frequency information and the open loop gain estimates in these regions can then be used for further processing.

[0171] Based on the current forward transfer function and the worst-case forward transfer function, it is possible to compute the current/worst-case open loop transfer function estimates, which can then be used to assess the risk of feedback and/or the physical fit of hearing aids, hence to control the hearing aids, e.g., [0172] Detection of obstacle close to ear/hearing aids, and hence to ensure the stability and audibility of hearing aids. [0173] Classification of bad physical fit, e.g., bad placement or “grown-out” ears, cf. e.g. US20130170660A1. [0174] All these can be combined with motion sensors to further control hearing aids.

[0175] FIG. 3A shows a first embodiment of a block diagram for open loop feedback path estimation using a probe signal comprising one or more sine tones, and

[0176] FIG. 3B shows a first embodiment of a block diagram for open loop feedback path estimation using a probe signal, wherein the probes signal generator is controlled by frequency sub-band levels of the band-split microphone signal (y(f)).

[0177] FIG. 3A, 3B show a model for open loop feedback path estimation using a sine tone, where the adaptive filter ĥ.sub.FB is estimated from signals u(n) and e(n), where n is a time index related to a sampling rate (f.sub.s) of the system (1/f.sub.s defining a time range). The loop is ‘open’ in the sense that the forward path (e.g. providing amplification in a closed loop configuration) is ‘broken up’. A (fast) feedback path estimate can e.g. be obtained by playing a number of tones (e.g. a melody) at different frequencies (open loop feedback estimation). The listening device comprises (in a special open loop mode) a tone generator (SINE in FIG. 3A, 3B) for feeding a signal comprising the tone or tones to the loudspeaker (instead of the normal output signal from the signal processing unit (HAG in FIG. 4A, 4B, not shown in FIG. 3A, 3B). The listening device is adapted to switch the output signal u(n) to the tone generator in a particular mode of the listening device (e.g. as part of a start-up procedure, or at the request of a user, e.g. via a user interface, e.g. a remote control). This is particularly relevant for verifying an appropriate mounting of an ITE-part of a listening device for a (e.g. elderly) person needing assistance in such mounting, or for a deep in the ear canal type of listening device, where a proper mounting is difficult to verify for any person. The tones may be played one at a time or a few tones simultaneously, if the tones are well separated in frequency (e.g. more than 1 kHz apart). The tones are propagated along the feedback path and enter the microphone as feedback signal v(n) (possibly mixed with a (target) signal x(n) from the environment) and arrive in the listening device as electric input signal y(n). The feedback estimation filter ĥ.sub.FB can then—by minimizing error signal e(n)—rapidly adapt to the correct feedback path estimate value for the given frequencies (represented by the probe signal). These values may then be compared to a reference estimate of the feedback path stored in a memory of the listening device. The stored reference estimate may be slowly varying (updated over time) to comply with changes in the ear canal of the user (e.g. a child's growth). Alternatively, a stored reference estimate may be fixed, in cases where no substantial changes to the dimensions of the ear canal of the user are expected (e.g. for adult (e.g. elderly) people needing help to mount their listening device(s) by a caring person). If the values deviate too much (e.g. if the feedback deviation (ΔFB=FBE.sub.REF−FBE.sub.CUR)is smaller than a predefined value, e.g. based on a sum of the deviations, Σ[ΔFB(f.sub.T)], being smaller than a sum threshold value Δ.sub.ΣFB.sub.THR, where the sum (Σ) is over the tones f.sub.T comprised in the probe signal), the ear mould is not correctly inserted, and a warning should inform a user (or observer) accordingly. The warning signal may comprise an acoustic, a visual or a mechanical (vibration) signal (or a mixture thereof) and the listening device may comprise corresponding signal generators controlled by a signal representative of the feedback deviation. The melody may be played in a loop (i.e. persist) for a certain predefined amount of time, or until it is detected that the mould of the listening device has been correctly mounted. In an embodiment, an information signal is issued after a user-initiated or after an automatically initiated measurement of current feedback based on a probe signal comprising a selected number of tones, in case it is concluded that the mould IS correctly mounted. The two embodiments shown in FIGS. 3A and 3B are nearly identical. The embodiment shown in FIG. 3B additionally comprises a level detector LD for providing a level of the input signal y(f) at different frequencies f. This is used as a control input PSC to the probe signal generator (here tone generator SINE) to adapt the level of the tones (or at least some of the tones) of the probe signal generator to the level of the input signal at the corresponding frequencies f. Alternatively or additionally, the duration of one or more tones may be adapted to the level of the input signal, e.g. by increasing the duration with increasing level. The listening device of FIG. 3B hence comprises an analysis filter bank A-FB for converting the time domain input signal y(n) to a frequency domain input signal y(f).

[0178] FIG. 4A shows a first embodiment of a block diagram for closed-loop feedback path estimation using frequency shift of the processed output signal, and

[0179] FIG. 4B shows a second embodiment of a block diagram for closed-loop feedback path estimation using the addition of a probe signal to the processed output signal.

[0180] Alternatively, the feedback estimation can be done using closed loop estimation. FIG. 4 shows a model for closed-loop feedback path estimation using frequency shift (FIG. 4A) and using the addition of a probe signal without frequency shift (FIG. 4B). In the embodiments of a listening device shown in FIG. 4A, 4B, HA-G represents the forward path gain and (in FIG. 4A) FS is a frequency shift block for applying a (preferably inaudible) frequency shift to the output signal. In FIG. 4B, PSG is a probe signal generator for providing a probe signal (see e.g. WO2009007245A1), which is added to the output signal from the processing unit HA-G to decrease correlation between input and output signal of the forward path of the listening device. A decrease in correlation may be achieved by any relevant measure, including frequency dependent delay, phase or frequency modification, etc. (in FIG. 4A, frequency shift is used). The probe signal generator PSG (including its activation) is controlled by the signal processing unit HA-G via control signal PSC. The feedback path h.sub.FB is estimated by the feedback estimation unit (adaptive filter) ĥ.sub.FB based on the frequency shifted output signal u(n) (FIG. 4A) and the output signal u(n) comprising a probe signal (FIG. 4B), respectively.

[0181] In the estimation model shown in FIG. 4A the feedback estimation relies on external sounds x(n) that are combined with the feedback signal v(n) resulting in (electric) microphone signal y(n). In the estimation model shown in FIG. 4B a (preferably inaudible) probe signal is added to the output signal (here, no frequency shift is applied when the probe signal is added; alternatively, a frequency shift may applied to the combined output signal). In either case of the closed loop estimation, external sounds x(n) are audible, but the estimation is typically slower than in the open loop estimation of FIG. 3A, 3B. An advantage of the closed loop estimation is that it can be performed during normal operation of the listening device.

[0182] FIG. 5 shows an embodiment of hearing aid according to the present disclosure. The hearing aid (HA) may be adapted to be worn by a user. The hearing aid may be partially or fully implanted in the head of the user, e.g. in case of a bone conducting hearing aid, etc. The hearing aid comprises a forward path. In the embodiment of FIG. 5, the forward path comprises an input transducer (IT) for converting a sound to a corresponding electric input signal (X) representing the sound. The input transducer (IT) may e.g. comprise a microphone (M). The sound may e.g. comprise target signal components (e.g. speech or other sound bits of the user's current interest) and background noise (e.g. non-speech components or other sound that is not or the user's current interest). In some situations, speech components and non-speech components can be seen as the background noise, for the adaptive filter estimation. The acoustic input signal (denoted ‘Acoustic input’ in FIG. 5) may comprise an external part (denoted ‘S.sub.x’ in FIG. 5) and a feedback part (denoted ‘v’ in FIG. 5). The forward path further comprises a hearing aid processor (PRO) for providing a processed signal (U) in dependence of the electric input signal (X) or a signal originating therefrom (here feedback corrected signal E) and for providing a processed output signal (U) in dependence thereof. The forward path further comprises an output transducer (OT) for providing stimuli perceivable as sound to the user in dependence of the processed signal (U). The output transducer may e.g. comprise a loudspeaker (SPK) for converting electric stimuli to acoustic vibrations in air. The output transducer may e.g. comprise a vibrator for converting electric stimuli to acoustic vibrations in skull bone and tissue. The input (IT) and output (OT) transducers may (as in FIG. 5) comprise appropriate analogue to digital converters (AD) and/or digital to analogue converters (DA) as appropriate to allow signals to be processed in the hearing aid as digital samples. The input (IT) and output (OT) transducers may (as in FIG. 5) comprise appropriate analysis and synthesis filter banks (FBA and FBS, respectively) as appropriate to allow signals to be processed in the hearing aid in the (time-)frequency domain (k, l) (e.g. as frequency sub-band signals), where k and l are frequency and time indices, respectively. Possible analysis and synthesis filter banks (FBA and FBS, respectively) may form part of a processor (e.g. a digital signal processor, of the hearing aid). The forward path is configured to provide a frequency dependent intended forward path transfer function (F) from the input transducer to the output transducer in dependence of the at least one electric input signal X (or a signal derived therefrom) and in dependence of the user (e.g. of a hearing impairment of the user, e.g. as mapped by a hearing profile, e.g. an audiogram).

[0183] The hearing aid (HA) further comprises a feedback control system comprising a feedback path estimator (AF), e.g. comprising an adaptive filter connected to signals (E, U) of the forward path. The adaptive filter may comprise an adaptive algorithm part (ALG) and a variable filter part (FIL). The algorithm part (ALG) is configured to provide a current frequency dependent estimate (Ĥ(k, l)) of a feedback path transfer function (H) of a feedback path from the output transducer to the input transducer in a closed loop configuration including the forward path. The estimate (Ĥ) of a feedback path transfer function (H) is e.g. provided as filter coefficients (Ĥ) configured to be applied to the variable filter part (FIL). The variable filter part (FIL) is configured to provide an estimate ({circumflex over (V)}) of the feedback path in dependence of the current estimate (Ĥ, or a modified version thereof) of the feedback path transfer function and of said processed signal (U). The feedback control system further comprises a combination unit (CU), e.g. an adder (‘+’) in the forward path configured to subtract the current feedback path estimate ({circumflex over (V)}) from a signal of the forward path (here electric input signal X) to provide a feedback corrected signal (E), which is fed to the hearing aid processor (PRO) and to the algorithm part of the feedback path estimator (AF). The hearing aid (HA) further comprises a background noise estimator (NLE) configured to provide a current frequency dependent background noise level estimate ({circumflex over (N)}(k, l)) representing a level (or a parameter dependent thereof, e.g. an average thereof) of at least one electric input signal (X(k, l) or in a processed version thereof. In other words, the input signal of the background noise estimator (NLE) can be taken from other places in the forward path, e.g., using feedback corrected signal (E) instead (or both, e.g. X and E), or some background noise estimates within the processor block (PRO), etc. The hearing aid (HA) further comprises an open loop gain estimator (OLGE) connected to the hearing aid processor (PRO) and to the feedback path estimator (AF). The open loop gain estimator (OLGE) is configured to provide a frequency dependent estimate {circumflex over (L)}(k, l) of a current open loop transfer function in dependence of the intended forward path transfer function (F(k, l)) and the current estimate (Ĥ(k, l)) of the feedback transfer function. The hearing aid (HA) further comprises a confidence level estimator (CLE) connected to the noise estimator (NLE) and to the open loop gain estimator (OLGE). The confidence level estimator (CLE) is configured to provide a current frequency dependent estimate of a confidence level ({circumflex over (P)}(k, l)) of the current estimate (Ĥ(k, l)) of the feedback path transfer function in dependence of the estimate (|{circumflex over (L)}(k)|) of current loop gain and the current (background) noise estimate ({circumflex over (N)}(k, l)). The confidence level estimator (CLE) may further receive an input from the hearing aid processor (PRO) (to make use of the intended forward gain (|F|, cf. dotted arrow from PRO to CLE), and possibly a knowledge of the hearing aid style, e.g. functional details of an earpiece, etc.). The confidence level estimate ({circumflex over (P)}(k, l)) may be based solely on the estimate (|{circumflex over (L)}(k)|) of current loop gain, in which case the noise estimator (NLE) can be dispensed with (at least for this purpose). The hearing aid (HA) may further comprise a controller (CONT) configured to control the hearing aid in dependence of the current estimate (Ĥ(k, l)) of the feedback transfer function and/or of the current estimate ({circumflex over (L)}(k, l)) of the open loop transfer function if said current estimate ({circumflex over (P)}(k, l)) of the confidence level fulfils a criterion, e.g. that it is above a threshold level (P.sub.th(k)). The confidence level may be frequency dependent (as indicated by dependency of k in P.sub.th(k)). The confidence level P.sub.th may, however, be the same over frequency.

[0184] The controller (CONT) may e.g. be configured to control the use of a current estimate (Ĥ(k, l)) of the feedback path transfer function in frequency bands k that fulfill the (confidence level) criterion (e.g. that the confidence level ({circumflex over (P)}(k, l′)) is above a threshold level (P.sub.th(k)) at a given point in time l′ (cf. signal FB.sub.ctr (and dotted arrow) to the feedback path estimator (AF), here specifically to the algorithm part (ALG) of the adaptive filter). The current estimate (Ĥ(k, l)) of the feedback path transfer function in frequency bands k for which the (confidence level) criterion is not fulfilled may e.g. be left unchanged (i.e. Ĥ(k, l′)=Ĥ(k, l′−1), if {circumflex over (P)}(k, l′)<P.sub.th(k)). The time index l(l′) may e.g. represent a time frame index (or a multiple thereof).

[0185] The controller (CONT) may e.g. be configured to use a current estimate (Ĥ(k, l)) in the control of other functionality of the hearing aid, e.g. via the hearing aid processor (PRO), cf. signal PR.sub.ctr (and dotted arrow) to the processor (PRO). The current estimate (Ĥ(k, l)) may e.g. be compared to a reference value H.sub.ref(k) and based thereon (and possibly on knowledge about the hearing aid style), it may be concluded that the hearing aid is not correctly mounted (e.g. indicating an extraordinary leakage of sound from the output transducer to the input transducer(s). This may e.g. result in an alarm being issued, e.g. via a user interface (e.g. implemented as ab APP of a smartphone).

[0186] The current estimate (Ĥ(k, l)) may e.g. be compared to the current gain in the forward path (F), to evaluate if the gain is at risk to be reduced or the output sound is at risk to be distorted due to feedback issues.

[0187] It is intended that the structural features of the devices described above, either in the detailed description and/or in the claims, may be combined with steps of the method, when appropriately substituted by a corresponding process.

[0188] As used, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well (i.e. to have the meaning “at least one”), unless expressly stated otherwise. It will be further understood that the terms “includes,” “comprises,” “including,” and/or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will also be understood that when an element is referred to as being “connected” or “coupled” to another element, it can be directly connected or coupled to the other element but an intervening element may also be present, unless expressly stated otherwise. Furthermore, “connected” or “coupled” as used herein may include wirelessly connected or coupled. As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items. The steps of any disclosed method is not limited to the exact order stated herein, unless expressly stated otherwise.

[0189] It should be appreciated that reference throughout this specification to “one embodiment” or “an embodiment” or “an aspect” or features included as “may” means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the disclosure. Furthermore, the particular features, structures or characteristics may be combined as suitable in one or more embodiments of the disclosure. The previous description is provided to enable any person skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects.

[0190] The claims are not intended to be limited to the aspects shown herein but are to be accorded the full scope consistent with the language of the claims, wherein reference to an element in the singular is not intended to mean “one and only one” unless specifically so stated, but rather “one or more.” Unless specifically stated otherwise, the term “some” refers to one or more.

REFERENCES

[0191] US20130170660A1 (Oticon) Jul. 4, 2013 [0192] EP2613567A1 (Oticon) Jul. 10, 2013 [0193] WO2009007245A1 (Oticon) Jan. 15, 2009 [0194] [Guo et al.; 2011] Meng Guo; Thomas Bo Elmedyb; Søren Holdt Jensen; Jesper Jensen, ‘Analysis of Acoustic Feedback/Echo Cancellation in Multiple-Microphone and Single-Loudspeaker Systems Using a Power Transfer Function Method’, IEEE Transactions on Signal Processing, Volume: 59, Issue: 12, December 2011, pp. 5774-5788. [0195] [HA-styles] https://advancedhearingcare.org/hearing-aid-styles.