Hearing device with suppression of comb filtering effect

Abstract

A hearing device (10) to be worn at least partially in the ear of an end user, the hearing device comprising a primary microphone (12), an ear canal microphone (13), a comb-signal detector (14), and an all-pass filter (15) that is connected to the comb-signal detector (14) and a receiver (19). The all-pass filter (15) is configured to receive a first input signal (20) and to modify an output signal (16) for the receiver (19). The ear canal microphone (13) is configured to generate a second input signal (21) for the comb-signal detector (14). The comb-signal detector (14) is configured to receive the first input signal (20), to detect the presence of notches (24) in the second input signal (21) that are indicative of the comb filter effect (25), and to issue control parameters (22) to the all-pass filter (15).

Description

FIELD OF THE INVENTION

[0001] The invention relates to hearing devices for end user and the suppression of the comb filter effect that leads to an unnatural sound perception by the end user.BACKGROUND OF THE INVENTION

[0002] Hearing devices are nowadays dimensionally small and complex devices. Hearing devices can include a microphone, a processor, memory, and other electronical and mechanical components to form an audio signal processor plus a speaker / receiver to emit a sound towards the eardrum, eventually. Types of such hearing devices are Behind-The-Ear (BTE), Receiver-In-Canal (RIC), In-The-Ear (ITE), Completely-In-Canal (CIC), Invisible-In-The-Canal (IIC) devices and in-ear headphones. The selection of the type of hearing device that suits an end user depends on factors like the hearing loss, aesthetic preferences, lifestyle needs, and budget. The term "end user" denotes the user of the hearing device.

[0003] Hearing systems and devices with audio signal processing are well known in the art. Audio signal processing may comprise noise reduction routines for reducing or even removing, undesired sound that is not relevant to the end user, which sound is commonly referred to as noise. That signal processing directed to improving speech intelligibility and listening comfort is typically referred to as speech enhancement.

[0004] The perceptual effect of the speech enhancement depends on the acoustic coupling to the ear canal. The acoustic coupling leads to a vent effect which is directly related to a cutoff frequency, and below which the effectiveness of the hearing device decreases and below which direct sound from the environment reaches the ear drum and can mask, i.e., can dominate over the processed audio signal from the hearing device.

[0005] So-called open fittings suffer from comb filter effects since both direct sound and processed sound arrive at the ear drum of an end user, but typically in an uncoordinated way due to varying ear canal input impedances and so-called acoustic leakage. The latter term refers to a direct sound path between the earpiece and the ear canal wall.

[0006] In frequency ranges where the magnitudes of the processed sound are in the same order of the magnitudes of the direct sound and where the phase d is 180 degrees shifted away to one another, this phenomenon leads to pronounced dips in the magnitude frequency response of the sound in the residual ear canal. Those dips lead to an unnatural sound perception in that voices or music may sound hollow to the end user. This effect is referred to hereafter as comb filter effect. The closer the magnitudes of the signals from the processed signal and the direct sound are, the higher the peaks and the deeper the dips of the comb filter effect will be. The comb filter effect would not form a problem if the gain of the hearing device was more than about 40 dB, but at audio signals of lower frequencies in the range of about 100 to 200 Hz and poor sealing between the hearing device and the wall of the ear canal also comb filter effects can occur.

[0007] A vent is an acoustic channel extending between the inside of the ear canal, the residual ear canal, and the out-side and the ambient environment (i.e. the region of the pinna) and allows, amongst other effects, for a pressure equalization between those regions when the hearing device is worn. In addition, vents are well-known means to lower undesired acoustic effects like the occlusion effect and contribute to a better-balanced microclimate with respect to humidity. Although an acoustically closed coupling is preferable in some situations, an acoustically open coupling can be desired in other situations (own-voice, environmental awareness). The introduction of a vent renders a closed coupling towards an open coupling, dependent on the vent size. With an individualized vent size, tailored solutions to the end user's hearing loss become available. In the context of the present application, the term 'vent' also refers to the opening or set of openings that are present in domes. Domes are frequently used in conjunction with RIC hearing aids.

[0008] Various attempts to mitigate the problem of the comb filter effect in hearing devices are known in the art. EP3419310A1, for example, addresses this problem by an embodiment of a hearing device shown and described with reference to its figure 6 that comprises a comb suppressor. An output signal of a sound processing unit is fed to the receiver. The comb suppressor receives input signals from a primary microphone, a canal microphone, a sound processing unit in the form of a control signal and the processed signal from the sound processing unit and generates on that basis a control signal which is then added with an output signal of the sound processing unit before feeding it to a receiver. The control signal from the sound processing unit is indicative of one or more processor gains G_1, G_2, ..., G_N applied in the sound processing unit, whereas N is the number of frequency bands in the sound processing unit.

[0009] Since the set-up of the comb suppressor comprises several suppressors it is rather complex. Moreover, the comb suppressor only has a limited impact on the final output signal to be fed to the receiver.DESCRIPTION OF THE INVENTION

[0010] Therefore, it is an objective of the present invention to provide a hearing device that can mitigate the comb filter effect effectively in a less complex way and more directly.

[0011] That objective is met by a method to mitigate an undesired comb filter effect in the context of a hearing device and a hearing device carrying out such a method in an operating state.

[0012] A core idea of the present invention resides in employing an all-pass filter and a comb-signal detector to detect notches from the comb filter effect and to shift the phase of the processed sound signal relative to the direct sound signal in case of a cutoff frequency where a comb filter effect peak would occur. That way, a superposition of the direct sound and processed sound at the eardrum of a user is reduced compared to a superposition of the direct sound and the processed sound at the eardrum of a user without an all-pass filter and a comb-signal detector.

[0013] Note that the all-pass filter is only employed to shift the phase of the processed sound that is formed by the output signal for the receiver relative to the second input signal if there is a notch and only in a controllable manner. In frequency ranges where there is no notch caused by the comb filter effect, there will be no phase shifting. Since there will always be some notches even in case of tightly fitting hearing aids such as custom shells, it is recommendable to delimit the detection of the notches caused by the comb filter effect to a frequency range of about 300 to 5000 Hz. Hence, reference measurements may be required to sort out undesired notches that are not caused by the comb filter effect from notches caused by the comb filter effect. It is also possible to select only notches that exceed a predefined amplitude or occur in a certain frequency band.

[0014] Further note that the phase shifting addressed in the context of the all-pass filter is not to be confused with any phase shifting that comes as a side effect of the signal processing at the sound processing unit.

[0015] The term "notch" is used herein to denote a large change in the amplitude / magnitude of the second input signal within a comparatively small frequency range only. A large change is understood as a comparatively substantial change in the amplitude, for example minimally 5 dB. There may be a predefined threshold that helps the comb-signal detector to decide on whether a notch is just a flaw or irregularity, or a notch that results of the comb filter effect that would lead to a perceptible sound because only the latter requires a phase shifting by the all-pass filter according to the present invention.

[0016] The criteria on whether a sound signal frequency is a cutoff frequency or not depends on the ratio of the incoming sound signal from the primary microphone to the signal from the canal microphone. A cutoff frequency is determined if the highpass cutoff frequency of the processed sound leaving the sound processing unit is at the same frequency as the low-pass cutoff frequency of the direct sound.

[0017] In a basic embodiment, the method to mitigate an undesired comb filter effect in the context of a hearing device that includes the following features. Said hearing device is supposed to be at least partially inserted into the ear canal of a user and comprises a primary microphone, an ear canal microphone, a comb-signal detector, an all-pass filter and a receiver. The method comprises the following steps: a) generating a signal of an audio input by the primary microphone, b) feeding a first input signal that is at least based on the signal from the primary microphone to the all-pass filter, c) feeding the first input signal to the comb-signal detector, d) generating an output signal for the receiver by the all-pass filter, e) generating a second input signal for the comb-signal detector by the ear canal microphone, f) detecting the presence of notches in the second input signal that are indicative of the comb filter effect by the comb-signal detector, and g) issuing control parameters for the all-pass filter by the comb-signal detector if at least one notch from the comb filter effect is detected.

[0018] Those parameters issued by the comb-signal detector and sent to the all-pass filter by the comb-signal detector are configured to cause the all-pass filter to shift the phase (the phase angle displacement) of the output signal that is sent to a receiver relative to the second input signal from the ear-canal microphone. If one applies a phase shift at a given frequency, the phase shift is defined for an entire frequency range.

[0019] When worn by the user, the primary microphone is located at a lateral end of the hearing device outside the ear canal, i.e., at the opposite of the ear canal microphone and the eardrum at the hearing device.

[0020] Even in the case of a so-called closed acoustic coupling, for example with closed domes, there are cutoff frequencies that lead to an acoustically equivalent to a direct path formed by a vent or a leakage. Hence, a vent effect is present also in such acoustic coupling situations.

[0021] The term "first input signal" that is at least based on a signal from the primary microphone does not necessarily need to be the signal from the primary microphone. It can also be a processed or a pre-processed signal, for example a signal treatment by a beamformer a signal that is leaving a sound processing unit, for example a noise reduction and speech enhancement unit. The step of all-pass filtering is plugged-in sequentially after the hearing loss compensation via the sound processing unit.

[0022] The key characteristic of the all-pass filter relying in the capability of shifting of the signal phase is employed to deliberately move sound signals within a small frequency range to another frequency range where there are no notches of comb filter effect such that the creation of undesired audible sound effects (artifacts) for the user is avoided.

[0023] All-pass filters are beneficial since they do not alter the magnitude of an input signal but just delay / shift the phase of that input signal in time. Moreover, LHP poles in the left-hand complex plane ("LHP") can be cancelled / removed elegantly by corresponding RHP-zeros in the right-hand plane ("RHP") in the Laplace domain.

[0024] Hearing devices with so-called open domes typically have a significant acoustic leakage owing to their design whereas acoustic leakage in hearing devices with earmolds, also known as custom shells, which are manufactured according to an impression of the user's ear canal, is almost no issue. Acoustic leakage denotes a situation where audio input from outside the ear canal of a user reaches the eardrum in the form of direct sound.

[0025] There are several options available to detect the notches of the comb-effect in the second input signal. Below, only a few options are mentioned as representatives.

[0026] A first option resides in that frequency ranges having a qualitatively sufficient signal-to-noise ratio ("SNR") are identified within a target frequency range.

[0027] A second option resides in that the cutoff frequencies with the notches are detected by comparing the frequency response of the output signal of the all-pass filter that is fed to the receiver with the second input signal from the in-canal microphone.

[0028] A third option resides in forming the magnitude (absolute value), equating the first derivative followed by a verification that the second derivation is larger than zero.

[0029] Although using a common all-pass filter would work for the one dedicated cutoff frequency mentioned earlier, the comb filter effect leads to further cutoff frequencies (ripples) that need to be addressed, too. Hence, a standard all-pass filter will not work as the frequency range of most pronounced comb filter effects changes both with changes in the acoustic coupling (e.g. leakage versus sealing) and with changes in amplification. Hence, the all-pass filter must be dynamically controllable because a static control with a fixed parameter would only be able to mitigate the first notch of the comb filter effect but not the subsequent notches that also require some treatment. Therefore, the comb-signal detector dynamically controls the all-pass filter in a time-variable manner. To optimize the performance of the all-pass filter with respect to reducing notches, it is beneficial to know the real-ear occluded gain ("REOG") and the vent effect ("VE").

[0030] The application of such a dynamic all-pass filter in a hearing device allows to minimize / dampen the amplitude / magnitude of the ripple of the combined sound wave at the ear drum consisting of the direct sound and the processed sound substantially. As a result, the sound is perceived more natural by the end user of such a hearing device.

[0031] Best results are achievable if the comb-signal detector controls the all-pass filter such that the output signal of the all-pass filter is adjusted based on frequency-dependent measurements of the second input signal. That frequency-dependency is important because the all-pass filter only needs to become active and shift the phase in a frequency range where the cutoff frequency and the notch is. In a comb filter effect, there is not only one notch but a plurality of such notches, each of which occurring in only a comparatively narrow frequency range.

[0032] To determine the REOG, the first input signal is compared with the second input signal. In more detail, the comb-signal detector determines the real ear occluded gain based on the first input signal and at least one of the second input signal and the output signal fed to the receiver.

[0033] One reliable and easy way of establishing this resides in comparing the Fast Fourier Transforms of the first input signal and the second input signal to see on whether the original signal (i.e., the first input signal that can be the signal from the primary microphone or a processed signal based thereon) contained some signal content in the frequency range of the notches, or not. Since the signals are often transient, the latency of the comb signal detector shall be kept as low as possible.

[0034] The comb-signal detector applies a phase shift to the output signal of the all-pass filter that is configured to be sent to the receiver. Although one could theoretically choose the phase shift in case of a frequency range where a notch occurs freely, it is recommendable to select the phase shift carefully according to rules that are explained hereinafter.

[0035] A phase shift of the output signal of the all-pass filter to the phase of the second input signal is exactly plus 180 degrees or exactly minus 180 degrees is undesired since it may lead to a combined signal that was zero. This because left halve pole(s) are "cancelled" by corresponding right halve pole zero(s) such that natural frequencies of poles and zeros that are identical, it results in zeros in the right-hand half-plane in the Laplace plane. To avoid such an undesired situation, the method is such that a phase of the output signal of the all-pass filter is shifted with respect to a phase of the second input signal by a phase displacement angle ΔΦ, whereas said phase displacement angle ΔΦ is any angle other than plus 180 degrees or minus 180 degrees exactly.

[0036] The phase shift needs to be adapted constantly depending on the loudness of the sound. It is recommended to achieve that by way of a feedforward scheme.

[0037] Since it is technically difficult to establish a phase shift of a signal leaving the all-pass filter compared to the signal entering the all-pass filter and that is in need of the phase shifting precisely by a phase angle of a plus 180 degrees or precisely by minus 180 degrees, one better allows for some tolerance in the phase shift angle as that contributes substantially to a stable performance of the system. That stability is improvable if said phase displacement angle ΔΦ is any angle other than a plus 180 degrees and an added a phase margin, or a minus 180 degrees and an added phase margin as it helps suppressing variations in circuit characteristics caused during manufacture or during operation. Therefore, it is recommended to provide at least a small phase margin to stay safe off the exactly plus 180 degrees or exactly off the minus 180 degrees to ensure that a phase cancelation is technically achievable in a reliable manner. In an exemplary embodiment, the phase margin is about 30 degrees, but it can be different depending on other requirements on the precision and the stability of the process.

[0038] If it is desired to maintain the amplitude height of the signal from the primary microphone to be fed to the all-pass filter or the signal from the sound processing unit to be fed to the all-pass filter, and the amplitude height of the output signal leaving the all-pass filter, the phase shift is selected that the phase displacement angle ΔΦ is either a plus 120 degrees, or a minus 120 degrees.

[0039] Having an unchanged amplitude is particularly beneficial in embodiments where the output signal of the all-pass filter is fed back to the comb-signal detector as a feedback signal. If the amplitude of x(t) was deliberately superelevated by the hearing device by 10%, for example, because the comb filter reduced it again by 10%, the notches of the comb filter effect would not be in the same frequency ranges before the all-pass filter and after the all-pass filter.

[0040] A significant reduction of the comb filter effect leads to a better sound quality for the user wearing the hearing device. It is believed that the more the magnitude of the comb filter effect can be reduced, the better the comfort of the hearing experience for the user since the acoustic perception of the notches is smaller. However, a quantification of a maximum magnitude / amplitude height of the superposition curve that was considered as ideal differs from user to user depending on their hearing loss and personal preferences such that it cannot be expressed in numbers easily.

[0041] If one is to accept changes in the first input signal that is based on the signal from the primary microphone due to the digital signal processing in the sound processing unit compared to its output fed to the all-pass filter, a basic embodiment of the method is achievable in that first input signal for the comb-signal detector is formed by the signal from the primary microphone.

[0042] If a Fourier Transform is performed in the sound processing unit, it is possible to use that by-product of said calculation to form the first input signal for the comb-signal detector. The Fourier Transform is generated by a Fast Fourier Transform unit of the sound processing unit. Said sound processing unit is connected between the primary microphone and the all-pass filter and is configured to process the signal arriving from the primary microphone.

[0043] If one desires to arrive at an output signal that is as close to the first input signal as possible, the first input signal for the comb-signal detector is created by a sound processing unit after concluding the processing of the signal by the sound processing unit. In that case, the sound processing unit is connected to the primary microphone and the all-pass filter. Since the Fourier Transform already took place in the sound processing unit, the comb-signal detector can process the first input signal more directly since there is no signal conversion for the comb-signal detector required.

[0044] Determining the vent effect is achievable by comparing the output signal leaving the all-pass filter with the second input signal. A reliable implementation of such a method is available if the output signal generated by the all-pass filter is fed as a further input signal to the comb-signal detector.

[0045] Calculating the vent effect without the output signal of the all-pass filter but based on the second signal alone was theoretically possible but prone to be unreliable because one would not know whether a notch is caused by a missing signal due to the comb filter or whether it is caused by something else, for example the sound processing unit. Therefore, having at least the output signal of the all-pass filter or the signal from the primary microphone at the comb-signal detector contributes substantively to a reliable detection of notches caused by the comb filter effect.

[0046] An easy option to put that into practice resides in comparing the Fast Fourier Transforms of the second input signal with the output signal leaving the all-pass filter to see on whether the original signal (i.e., the first input signal that can be the signal from the primary microphone or a processed signal thereof) contained some signal content in the frequency range of the notches caused by the comb filter effect, or not.

[0047] All the advantages mentioned above are available in a hearing device that is configured such that it performs the above-mentioned methods.

[0048] In a basic embodiment, a hearing device according to the present invention that is to be worn at least partially in the ear of an end user would comprise a primary microphone, an ear canal microphone, a comb-signal detector and an all-pass filter that is connected to the comb-signal detector and a receiver. The primary microphone is configured to generate a signal of an audio input. The all-pass filter is configured to receive a first input signal that is at least based on the signal of the primary microphone and to generate an output signal for the receiver. The term "is at least based on the signal of the primary microphone" is understood such that the signal that serves as an input for the comb-signal detector can be the unprocessed signal of the primary microphone or the pre-processed signal of the primary microphone after its digital signal processing in part or in full by a sound processing unit. The term "sound processing unit" shall be understood functionally and not narrow in that the sound processing unit needs to take place in a single chip.

[0049] Further, the comb-signal detector is configured to receive the first input signal. The ear canal microphone is configured to generate a second input signal for the comb-signal detector. The comb-signal detector is configured to detect the presence of notches in the second input signal that are indicative of the comb filter effect. Moreover, the comb-signal detector is configured to issue control parameters to the all-pass filter.

[0050] Although analogue all-pass filters would theoretically be conceivable, digital all-pass filters are preferred because of their flexibility and accuracy.

[0051] A preferred subcategory of digital filters is formed by infinite impulse response filters (IIR). An IIR filter is preferred over a finite impulse response filter (FIR) since the filter bank of the FIR would get undesirably long.

[0052] The above-mentioned hearing device is a receiver-in-canal device (RIC) in one embodiment. However, other embodiments like an ITE, just to name a further example, are possible.

[0053] More aspects of the invention will be explained with reference to exemplary embodiments described hereinafter.BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Below, embodiments of the present invention are described in more detail with reference to the attached drawings. Fig. 1 shows a simplified cross section of an ear canal with an ear drum and a hearing device inserted into that ear canal; Fig. 2A shows a Bode diagram of the signal magnitude of an exemplary signal of a superposition of a processed signal of a hearing device and a direct sound signal, where the superposition has a typical comb filter pattern with a plurality of steep notches in several small frequency ranges; Fig. 2B shows a Bode diagram of the phase angle for the signal from fig. 2A; Fig. 3A shows the undesired result of a superposition of two signals that have the same phase angle with respect to the magnitude of the superposition; Fig. 3B shows the undesired result of a superposition of two signals that have a phase shift of exactly plus 180 degrees or minus 180 degrees leading such that there is no superpositioned signal outcome; Fig. 3C shows an undesired result of a superposition of two signals that have a phase shift of exactly plus 120 degrees or minus 120 degrees leading to a superposition signal having the same magnitude as each of the input signals; Fig. 4 is a first embodiment of a hearing device circuit when in use; Fig. 5 is a second embodiment of a hearing device circuit when in use; Fig. 6 is a third embodiment of a hearing device circuit when in use; Fig. 7 is a schematic diagram showing a direct sound signal, a processed sound signal as well as its superposition, whereas the superposition only considers the absolute amounts of the magnitudes of these input signals and ignores any phase shifts between these input signals; Fig. 8 is a schematic diagram showing a direct sound signal, a processed sound signal that is not led through an all-pass filter as well as the superposition thereof, whereas the superposition only considers both magnitudes, the polarity as well as the phase shift of these input signals; and Fig. 9 is a schematic diagram showing a direct sound signal (as in the diagram of fig. 8), a processed sound signal (as in the diagram of fig. 8) that was led through a frequency-dependent all-pass filter. The superposition thereof considers the magnitudes, the polarity as well as the phase shift of these input signals as done in the diagram of fig. 8. DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

[0055] The reference characters used in the drawings are explained hereinafter, where required. Identical and functionally identic items are given the same reference characters in the figures.

[0056] Fig. 1 displays a schematic cross-section of an overview about an exemplary hearing device 10 in the form of an ITE or RIC that is inserted into an ear canal 17 of the ear 11 of a user. The hearing device 10 has a primary microphone 12, a direct sound path, here displayed by a vent 7 for example and for simplicity, an ear canal microphone 13, a signal processing unit 8 and a receiver 19 for emitting an acoustic signal to the ear drum 9 of the user to compensate for the user's hearing loss.

[0057] Not shown in figure 1 but present nonetheless are a battery for powering the hearing loss compensation and details for detecting and mitigating undesired effects of a comb filter effect taking place because of the presence of a vent 7 in the hearing device 10 as they will be addressed later in more detail. Although this hearing device 10 has a dedicated vent 7 in the form of a channel extending between the inside of the ear canal 17 and the out-side and the ambient environment (i.e. the region of the pinna) and allows, amongst other effects, for a pressure equalization between those regions when the hearing device 10 is worn. Vents are well-known means to lower undesired acoustic effects like occlusion and sound artifacts and contribute to a better-balanced microclimate with respect to humidity in that they let some audio input 6 in the form of so-called direct sound 30 through the hearing device 10 propagate to the ear drum 9. Although an acoustically closed coupling is preferable in some use situations, an acoustically open coupling can be desired in other use situations (own-voice, environmental awareness). A vent renders a closed coupling towards an open coupling, dependent on the vent size. With an individualized cutoff frequency, tailored solutions to the end user's hearing loss become possible. Undesired vent-effects may also occur in case of so-called close couplings where the hearing device has no dedicated vent but where there is a gap between the ear canal and the hearing device that allows direct sound (audio input) to reach the ear drum. However, please note that the present invention does not aim primarily at remedying such situations where the sealing effect between the hearing device and the ear canal is imperfect.

[0058] Figure 2A shows a Bode diagram of the signal magnitude of an exemplary signal of a superposition of a processed signal of a hearing device and a direct sound signal, where the superposition has a typical comb filter pattern 25 with a plurality of steep notches 24 in several small frequency ranges (shown in the horizontal axis) at about 85 Hz, about 260 Hz, at about 400 Hz and many more. Figure 2B shows a Bode diagram of the phase angle for the signal from fig. 2A where the notches 24 are clearly visible in that they form sharp peaks in the amplitude / magnitude of the signal.

[0059] Next, let us explain the effect of having a phase shift of the output signal for the receiver relative to an input signal representing an audio input and its effect at the ear drum on a user in figure 3.

[0060] To start with figure 3A, the effect is shown in a simplified matter by way of harmonic and steady sine signals with the same cycle length. Input signal curve x(t) is shown in a dotted line 26 whereas the signal of the audio input entering the vent corresponds to direct sound shown in curve y(t) in a dashed line 27 but having the same amplitude as curve x(t) 26 in the diagram. Curve x(t) can be seen as representing the direct sound while curve y(t) can be seen as representing the processed sound. Since curve x(t) 26 and curve y(t) 27 have the same phase angle, there is no phase shift between curve x(t) 26 and curve y(t) 27. As a result, the resulting superposition 28 shown in a solid line has an amplitude that is exactly the sum of the height of the amplitude of curve x(t) 26 and the height of the amplitude of curve y(t) 27.

[0061] Figure 3B shows an undesired situation where the curve x(t) 26 and curve y(t) 27 have the same amplitude height and the same cycle length, but a phase shift of exactly plus 180 degrees or minus 180 degrees in the diagram. In this case, the resulting superposition 28 of the curve x(t) 26 and curve y(t) 27 shown as a solid line leads to no signal at all. Hence, a phase shift of 180 degrees cannot contribute to a satisfying solution of the present technical problem.

[0062] Different to the figures 3A and 3B shows figure 3C a diagram, whose curve x(t) 26 and curve y(t) 27 have the same cycle length and the same amplitude height. However, this time, the phase shift between the curve x(t) 26 and the curve y(t) 27 is exactly plus 120 degrees or minus 120 degrees. As a result, the superposition signal 28 has the same amplitude / magnitude as each of the signals 26, 27.

[0063] Figure 4 is a schematic diagram of a first embodiment of a hearing device circuit built-in the hearing device 10 such as shown in figure 1 when in use.

[0064] As already mentioned in the context of figure 1, the primary microphone 12 is configured to generate a signal 23 of an audio input 6. An infinite impulse response filter is provided as an all-pass filter 15 that is configured to receive a first input signal 20 that is at least based on the signal 23 of the primary microphone 12 and to generate an output signal 16 for the receiver 19.

[0065] Connected in between the primary microphone 12 and the all-pass filter 15, there is a sound processing unit 18. A comb-signal detector 14 is configured to receive the first input signal 20. The ear canal microphone 13 is configured to generate a second input signal 21 for the comb-signal detector 14. The comb-signal detector 14 is configured to detect the presence of notches 24 (see e.g., figures 2 or 8 or 9) in the second input signal 21 that is indicative of the presence of a comb filter effect 25. Moreover, the comb-signal detector 14 is configured to issue dynamic control parameters 22 for the all-pass filter 15. The comb-signal detector 14 controls the all-pass filter 15 in a time-variant manner and such that the output signal 16 of the all-pass filter 15 is adjusted based on frequency-dependent measurements of the second input signal 21. The sound processing unit 18, the all-pass filter 15, the comb-signal detector 14 and the electrical connections in between them belong to a signal processing unit 8 shown in a dotted frame. That signal processing unit 8 forms a functional, not necessarily a bodily entity.

[0066] The output signal 16 has a delay in the phase angle to the signal 23 from the primary microphone 12 of 120 degrees plus a phase margin of 20 degrees to allow a distinct, clear and precise control of the system.

[0067] As can be seen in figure 4, the receiver 19 converts the output signal 16 of the all-pass filter 15 into an acoustic signal that is directed towards the ear drum 9. That sound path is shown simplified by a dashed arrow. The ear-canal microphone 13 receives the same or almost the same acoustic signal issued by the receiver 19 and feeds it as input feedback to the comb-signal detector 14. That is shown by another dashed arrow.

[0068] The output signal 16 of the all-pass filter 15 is fed back to the comb-signal detector 14 to determine the vent effect reliably. That determination is done by comparing the output signal 16 leaving the all-pass filter 15 with the second input signal 21 from the ear canal microphone 13.

[0069] The difference of the schematic circuitry of a second embodiment shown in figure 5 compared to the one in figure 4 resides in that the first input signal 20 is not the same signal as the signal 23 of the primary microphone 12 but a processed signal that is leaving the sound processing unit 18. Said processed signal that is leaving the sound processing unit 18 is still based on the signal 23 of the primary microphone 12.

[0070] The advantage of such an embodiment compared to the one shown in figure 4 resides in that there may be already a phase shift between the signal leaving the sound processing unit 18 and the signal 23 of the primary microphone 12. If the sound processing unit has variable latencies, that information does not need to be fed to the comb-signal detector again.

[0071] The difference of the schematic circuitry of a third embodiment shown in figure 6 compared to the circuitries shown in figures 4 and 5 resides in that the first input signal 20 is formed by the Fast Fourier Transform unit 18.1 of the sound processing unit 18 since there is a FFT needed for the sound processing anyway, before the signal is led to an Inverse Fast Fourier Transform (IFFT-unit) 18.2 and from there to the all-pass filter 15 for further treatment. Using this interim signal leaving the Fast Fourier Transform unit 18.1 for forming the first input signal 20 is an elegant way to provide the required input for the comb-signal detector 14.

[0072] Since it is difficult to display the effect of the comb filter effect reduction according to the present invention graphically, figures 7 to 9 will assist the reader in grasping the substantial technical effect of the present invention better.

[0073] Note that hearing devices are known in the art that do not employ an FFT-IFFT. In a further embodiment not shown in the figures, the first input signal is extracted from the sound processing unit 18 and fed as an input to the comb-signal detector 14.

[0074] Figure 7 shows a magnitude frequency response of the sound in the residual ear canal. This is done by a schematic diagram showing a direct sound signal 30 (shown in a dotted line), a processed sound signal 31 (shown in a dashed line) as well as its superposition 32. The horizontal axis indicates the frequency in Hertz (Hz) while the vertical axis H indicates the magnitude of the frequency response in decibels (dB referenced to 1 Pa / Pa). The superposition 32 here is a non-harmonic signal compared to the superposition signal 28 shown in figure 3. To avoid any confusion with superposition signal 28, a different reference character 32 was used since the superposition 32 in figure 7 only considers the absolute amounts of the magnitudes of the input signals of the direct sound 30 and the processed sound 31 while ignoring any phase shift of the phase angle between direct sound 30 and the processed sound 31. As one can see, the resulting superposition 32 displayed as a double line in figure 7 would have an acceptable peak / cutoff frequency at about 1'000 Hz in this example. However, one must bear in mind that figure 7 only shows a simplified scenario where any phase shift of the phase angle between direct sound 30 and the processed sound 31 is disregarded.

[0075] When factoring in the phase shift of the phase angle, the amplitudes / magnitudes as well as the polarity of these input signals formed by the direct sound 30 and the processed sound 31, the resulting superposition 33 looks dramatically different to the one shown in figure 7. As one can see in figure 8, the comb filter effect leads to a first huge notch in the amplitude of the superposition 33 displayed as double line of about -30 dB magnitude in a small frequency range at about 320 Hz. The first huge notch is followed by a series of smaller ripples in curve 33 having a decreasing amplitude but a rising magnification up to about 25 dB at about 1'000 Hz before decreasing to a steady level of about 20 dB for higher frequencies. It becomes evident that at least that first notch at about 320 Hz is that essential that is will inevitably lead to an undesired audible artifact of the comb filter effect for the user of the hearing device if it had no comb filter effect suppressor such as the one promoted in the present application.

[0076] Figure 9 now schematically displays the effect of having a comb filter effect suppressor in the form of an all-pass filter as described above in this application. The superposition 33 from figure 8 with the curve shown by a double line is shown in figure 9 again to allow a direct comparison.

[0077] As one can see in figure 9, the superposition 34 shown as a solid line having a shift of the phase angle compared to the direct sound signal 30 (shown in a dotted line) and the processed sound signal 31 (shown in a dashed line) clearly indicate beyond doubt that the notch of curve 33 at about 320 Hz that led to a magnitude height of -30dB is now reduced to a peak of a mere -10 dB in a range of about 370 Hz such that a notch magnitude reduction 35 of substantially 20 dB becomes available.

[0078] Moreover, on can see that there is also a remarkable phase shift between the curve 33 and the curve 34 for subsequent notches at higher frequencies in figure 9.

[0079] In a nutshell, the way smaller notch at about 370 Hz causes a way less disturbing artifact than the bigger notch at about 320 Hz shown in the set-up without an all-pass filter shown and explained in the context of figure 8.LIST OF REFERENCE SYMBOLS

[0080] 6audio input 7a direct sound path, e.g., formed by a vent, leading to a vent effect 8signal processing unit 9ear drum 10hearing device 11ear 12primary microphone 13ear canal microphone 14comb-signal detector 15all-pass filter 16output signal of the all-pass filter 17ear canal 18sound processing unit 18.1Fast Fourier Transform unit (FFT-unit) 18.2Inverse Fast Fourier Transform (IFFT-unit) 19receiver 20first input signal 21second input signal 22control parameters for the all-pass filter 23signal from the primary microphone 24notches 25comb filter effect 26curve x(t) 27curve y(t) 28superposition curve x(t) + y(t) 30direct sound 31processed sound 32superposition of the amounts of direct sound and processed sound when ignoring the phase shift 33superposition of the amounts of direct sound and processed sound when considering the phase shift 34superposition of the amounts of direct sound and processed sound when considering the phase shift, compared with the output signal from the all-pass filter 35notch peak / magnitude reduction

Claims

1. A method to mitigate undesired comb filter effect (25) in the context of a hearing device (10) for an end user wearing that hearing device (10), wherein that hearing device (10) is supposed to be at least partially inserted into the user's ear (11) and comprises a primary microphone (12), an ear canal microphone (13), a comb-signal detector (14), an all-pass filter (15) and a receiver (19), the method comprising the steps of: generating a signal (23) of an audio input (6) by the primary microphone (12), feeding a first input signal (20) that is at least based on the signal (23) from the primary microphone (12) to the all-pass filter (15), feeding the first input signal (20) to the comb-signal detector (14), generating an output signal (16) for the receiver (19) by the all-pass filter (15), generating a second input signal (21) for the comb-signal detector (14) by the ear canal microphone (13), detecting the presence of notches (24) in the second input signal (21) that are indicative of the comb filter effect (25) by the comb-signal detector (14), and issuing control parameters (22) for the all-pass filter (15) by the comb-signal detector (14) if at least one notch from the comb filter effect is detected.

2. The method according to claim 1, characterized in that the comb-signal detector (14) dynamically controls the all-pass filter (15) in a time-variable manner.

3. The method according to claim 2, characterized in that the comb-signal detector (14) controls the all-pass filter (15) such that the output signal (16) of the all-pass filter (15) is adjusted based on frequency-dependent measurements of the second input signal (21).

4. The method according to any one of claims 1 to 3, characterized in that comb-signal detector (14) determines the real ear occluded gain based on the first input signal (20) and at least one of the second input signal (21) and the output signal (16).

5. The method according to claim 3, characterized in that a phase of the output signal (16) of the all-pass filter (15) is shifted with respect to a phase of the second input signal (21) by a phase displacement angle (ΔΦ), and whereas said phase displacement angle (ΔΦ) is any angle other than a plus 180 degrees, or a minus 180 degrees.

6. The method according to claims 5, characterized in that said phase displacement angle (ΔΦ) is any angle other than a plus 180 degrees and an added a phase margin, or a minus 180 degrees and an added phase margin.

7. The method according to any one of claims 5 to 6, characterized in that the phase displacement angle (ΔΦ) is either plus 120 degrees, or minus 120 degrees such that an amplitude height of the output signal (16) of the all-pass filter (15) remains unchanged compared to an amplitude height of an output signal of a sound processing unit (18) that is connected between the primary microphone (12) and the all-pass filter (15) and that forms an input signal for the all-pass filter (15).

8. The method according to any one of claims 1 to 7, characterized in that the first input signal (20) for the comb-signal detector (14) is formed by the signal (23) from the primary microphone (12).

9. The method according to any one of claims 1 to 7, characterized in that the first input signal (20) for the comb-signal detector (14) is generated by a Fast Fourier Transform unit (18.1) of a sound processing unit (18), wherein said sound processing unit (18) is connected between the primary microphone (12) and the all-pass filter (15), and wherein the sound processing unit (18) processes the signal (23) arriving from the primary microphone (12).

10. The method according to any one of claims 1 to 7, characterized in that the first input signal (20) for the comb-signal detector (14) is created by a sound processing unit (18) after concluding the processing of the signal (23) by the sound processing unit (18), wherein said sound processing unit (18) is connected to the primary microphone (12) and the all-pass filter (15).

11. The method according to any one of claims 1 to 10, characterized in that the output signal (16) generated by the all-pass filter (15) is fed as a further input signal to the comb-signal detector (14).

12. A hearing device (10) to be worn at least partially in the ear of an end user, the hearing device comprising a primary microphone (12), an ear canal microphone (13), a comb-signal detector (14), and an all-pass filter (15) that is connected to the comb-signal detector (14) and a receiver (19), wherein the primary microphone (12) is configured to generate a signal (23) of an audio input (6), and wherein the all-pass filter (15) is configured to receive a first input signal (20) that is at least based on the signal (23) of the primary microphone (12) and to generate an output signal (16) for the receiver (19), and wherein the comb-signal detector (14) is configured to receive the first input signal (20), and wherein the ear canal microphone (13) is configured to generate a second input signal (21) for the comb-signal detector (14), and wherein the comb-signal detector (14) is configured to detect the presence of notches (24) in the second input signal (21) that are indicative of the comb filter effect (25), and wherein the comb-signal detector (14) is configured to issue control parameters (22) to the all-pass filter (15) if at least one notch from the comb filter effect is detected.

13. The hearing device according to claim 12, characterized in that the hearing device (10) is configured for carrying out the method according to anyone of claims 1 to 11.

14. The hearing device according to claim 12, characterized in that the all-pass filter (15) is a digital filter.

15. The hearing device according to any one of claims 12 to 14, characterized in that the all-pass filter (15) is an infinite impulse response filter.

16. The hearing device according to any one of claims 12 to 15, characterized in that the hearing device (10) is a receiver-in-canal device.

Images