EP2190218B1 - Filter bank system with specific stop-band attenuation for a hearing device - Google Patents

Description

The present invention relates to a filter bank system for a hearing apparatus having an analysis filter bank for decomposing an input signal into subband signals, processing means for manipulating at least one of the subband signals and a synthesis filter bank for composing the manipulated subband signal with at least one other of the subband signals. Moreover, the present invention relates to a hearing aid with such a filter bank system. The term "hearing device" is understood to mean here any sound-emitting device that can be worn in or on the ear or on the head, in particular a hearing device, a headset, headphones and the like.

Hearing aids are portable hearing aids that are used to care for the hearing impaired. In order to meet the numerous individual needs, different types of hearing aids such as behind-the-ear hearing aids (BTE), hearing aid with external receiver (RIC: receiver in the canal) and in-the-ear hearing aids (IDO), e.g. Concha hearing aids or canal hearing aids (ITE, CIC). The hearing aids listed by way of example are worn on the outer ear or in the ear canal. In addition, bone conduction hearing aids, implantable or vibrotactile hearing aids are also available on the market. The stimulation of the damaged hearing takes place either mechanically or electrically.

Hearing aids have in principle as essential components an input transducer, an amplifier and an output transducer. The input transducer is usually a sound receiver, z. As a microphone, and / or an electromagnetic receiver, for. B. an induction coil. The output transducer is usually used as an electroacoustic transducer, z. As miniature speaker, or as an electromechanical transducer, z. B. bone conduction, realized. The amplifier is usually integrated in a signal processing unit. This basic structure is in FIG. 1 shown using the example of a behind-the-ear hearing aid. In a hearing aid housing 1 for carrying behind the ear, one or more microphones 2 for receiving the sound from the environment are installed. A signal processing unit 3, which is also integrated in the hearing aid housing 1, processes the microphone signals and amplifies them. The output signal of the signal processing unit 3 is transmitted to a loudspeaker or earpiece 4, which outputs an acoustic signal. The sound is optionally transmitted via a sound tube, which is fixed with an earmold in the ear canal, to the eardrum of the device carrier. The power supply of the hearing device and in particular the signal processing unit 3 is effected by a likewise integrated into the hearing aid housing 1 battery. 5

Sound signals which are recorded by one or more microphones of a hearing device or another hearing device are usually decomposed into subband signals for further processing. For this one usually uses an analysis-synthesis filter bank system. This has one or more frequency-selective digital analysis filter banks (AFB), with which the sound signal is decomposed into K> 1 subband signals. Subband-specific signal manipulation then takes place, in particular amplification or attenuation of the subband signals. Subsequently, a re-synthesis of the manipulated subband signals is performed by means of one or more digital synthesis filter banks (SFB). The filter bank system is usually oversampling and has an oversampling factor U ≥ 1.

High-quality filter banks in hearing aids are subject to certain requirements. For example, in the bottommost bands, a channel width of at least about 250 Hz is needed. Otherwise, the band gap should be based on the Bark scale. A finer resolution, for example in the broader bands according to the Bark scale, is the application but not contrary. Furthermore, a channel number of at least 22 is desirable. Disturbance due to aliasing and imaging should be below about 40-60 dB, depending on the application (hearing impairment of the patient). Due to the intensive subband processing (especially the high gain required to compensate for hearing loss) in hearing aids, conventional methods for erasing aliasing and imaging are not effective. The filter banks are therefore fundamentally "non-critical" to scan. Furthermore, the group delay (in each case for AFB and SFB) should be well below 5 ms and the group delay distortions should not exceed a certain range. Especially for high frequencies, the group delay is to be kept as low as possible, which represents a significant limiting factor for the filter bank.

Specifically, the AFB and the SFB should be designed such that the signal / interference distance at the output of the filter bank system with arbitrary manipulation (amplification / attenuation) of the subband signals changed only within predeterminable limits. This also includes the case of the complete independence of the signal / interference distance at the output of the filter bank system from the manipulation of the subband signals.

In addition, the AFB and the SFB should be designed so that for a at the output of the filter bank system in response to the manipulation of the subband signals permissible variation of the signal / interference distance of the circuit complexity (corresponds to the filter orders) and / or the group delay (total delay ) of the filter bank system is reduced over the prior art.

The document 698 33 749 T2 describes a filter bank analysis structure for a hearing aid in which a buffer is connected to a multiplication unit so that the buffer values can be multiplied by a window function representing a prototype low-pass filter of a fast Fourier transform procedure. In this In connexion, it is taught to those skilled in the art that it is necessary to design this function correctly to give the filter bands a desired bandpass and stopband response and thereby reduce the audible adulteration disturbance.

For example, in previous approaches to solving these problems, filter banks AFB and SFB have been designed the same (in particular, equal magnitude specification of minimum phase AFB and maximum phase SFB). In the case of the frequency-independent stop attenuation, the oversampling factor was not considered in detail. Likewise, the signal manipulation was not considered in the system specification of the filter bank system. This resulted in large fluctuations in the signal quality (signal / interference distance) as a function of the respective manipulation (amplification) of the subband signals.

The object of the present invention is thus to propose a filter bank system with reduced computational effort, with which nevertheless a high signal quality, in particular a specific signal / interference distance, can be achieved.

According to the invention, this object is achieved by a filter bank system for a hearing device with an analysis filter bank (AFB) for decomposing an input signal into subband signals, a processing device for manipulating at least one of the subband signals and a synthesis filter bank (SFB) for assembling the manipulated subband signal with at least one further of the subband signals, wherein the stopband attenuation of at least one of the transfer functions of the analysis filter bank (AFB) is composed of a separately configurable, frequency independent analysis baseline damping portion and a separately configurable, frequency dependent analyte attenuation portion, and / or the stopband attenuation of at least one of the synthesis filter bank (SFB) transfer functions a separately configurable, frequency independent synthesis baseline damping portion, a separately configurable, frequency dependent one and manipulating dependent first synthesis attenuation part and a separately configurable frequency dependent second synthesis attenuation part.

Advantageously, it is thus possible to exploit certain effects in the stopband attenuation and to keep the damping high only where it is necessary. As a result, the filter order can be significantly reduced under certain circumstances.

Preferably, all the attenuation components are configured on the basis of the signal / interference distance at the output of the filter bank system. Thus, the expert is given a simple criterion in the design of the filter bank systems at hand.

In a specific embodiment, the analysis baseline attenuation fraction depends on the downsampling factor of the analysis filter bank. Thus, the basic attenuation can be set to a minimum value.

In addition, a masking effect of a human ear can be taken into account in the frequency-dependent analysis-attenuation component and / or synthesis-attenuation component. This makes use of the fact that, in the case of sound perception, two parts located close to one another in a spectrally obscure manner may be completely or partially obscured. Noise components which are obscured by other components would thus no longer have to be fully attenuated.

In a further embodiment, the frequency-dependent analysis-attenuation component may be periodically modified, the periodicity being determined by the oversampling factor of the analysis filter bank (AFB) and the maximum attenuation reduction by the passbands (transmission behavior) of the AFB and SFB. In the same way, the frequency-dependent synthesis damping component can be periodically modified, the periodicity also being determined by the upward-scaling factor or integration factor of the synthesis filter bank and the maximum attenuation reduction by the oversampling factor. The periodicity of the attenuation components takes into account the fact that the artefacts also occur periodically through aliasing and imaging.

Likewise, the baseline synthesis fraction of synthesis may depend on the synthesis filter bank up-down factor.

Furthermore, it is of particular advantage if the frequency-dependent first synthesis damping component depends on a gain of the processing device. Thus, the attenuation of the interference components can just be chosen so that only then strongly attenuated if they are also high due to high amplification of the useful signal. This, too, can generally or temporarily reduce the filter effort.

Furthermore, it is favorable if the frequency-dependent second synthesis damping component is periodically modified, the periodicity is determined by the oversampling factor of the synthesis filter bank (SFB) and the maximum attenuation reduction by the transmission range (transmission behavior) of the SFB.

It has already been indicated that the above-described filter bank system can be used particularly advantageously in a hearing aid.

FIG 1

eine Prinzipskizze eines Hörgeräts gemäß dem Stand der Technik;

FIG 2

ein Blockschaltbild eines Filterbanksystems;

FIG 3

ein Beispiel für die Spezifikation eines AFB-Prototypfilters und

FIG 4

ein Beispiel für die Spezifikation eines SFB-Prototypfilters.

The present invention will now be explained in more detail with reference to the accompanying drawings, in which:

FIG. 1

a schematic diagram of a hearing aid according to the prior art;

FIG. 2

a block diagram of a filter bank system;

FIG. 3

an example of the specification of an AFB prototype filter and

FIG. 4

an example of the specification of a SFB prototype filter.

The embodiments described in more detail below represent preferred embodiments of the present invention.

In FIG. 2 a filter bank system is shown schematically, as it is used for example in a device. An input signal e (eg, speech signal) is supplied to an analysis filter bank AFB. This decomposes the input signal e into 32 subbands. This is followed by a subband-specific manipulation M1, M2, M3,..., M32 of the individual signals. After Manipulation, the individual subband signals are synthesized in an adjusted with respect to the amount of frequency response to the analysis filter bank AFB synthesis filter bank SFB again to an output signal a.

Both in the analysis filter bank AFB and the synthesis filter bank SFB, the individual filter functions are implemented by so-called prototype filters, from which the individual filters are derived, for example, by a complex-modulating transformation kernel in the filter bank. These prototype filters have, for example, low-pass characteristics. In the following, only the blocking area of the prototype filter is considered.

According to the core idea according to the invention, the notch attenuation specification of the transfer functions of the AFB is composed of a frequency-independent basic attenuation a ^AFB and frequency-dependent attenuation components. The blocking attenuation specification of the transfer functions of the SFB is composed of a frequency-independent basic attenuation a ^SFB and first and second frequency-dependent attenuation components, wherein the first frequency-dependent attenuation component additionally depends on the signal manipulation between the two filter banks. In particular, the AFB and the SFB should be specified such that by using the oversampling of the subband signals by a factor U> 1 (U = oversampling factor) the circuit complexity, ie the filter order and / or the total group delay (delay) of the filter bank system is reduced. In addition, as an option, mutual masking effects of frequency closely adjacent signal components are to be used in order to reduce the circuit complexity or the filter order and / or the total group delay of the filter bank system. Furthermore, the deterioration of the signal / interference distance at the output of the filter bank system should be approximately the same for any manipulation (amplification / attenuation of the subband signals by aliasing contributions of the AFB or by imaging contributions of the SFB.

The configuration of a concrete filter bank system will now be described on the basis of FIG. 3 and 4 explained in more detail. As an objective criterion for the dimensioning of the filters, the signal / interference distance SNR is generally used at the output of the filter bank system.

In the configuration of the analysis filter bank AFB a frequency independent attenuation a basic ^AFB is first determined by the desired SNR ^AFB. This signal / interference distance SNR ^AFB results from aliasing in the AFB. The basic attenuation a ^{AFB is} dependent on the decimation factor M (= down-turn factor) of the AFB, which determines the number of aliasing units that are present at all. The basic attenuation in the stopband is now supplemented by a frequency-dependent additional stop attenuation. A first part 10 of the frequency-dependent attenuation component of the AFB specification results from the fact that masking effects of the human ear are utilized. This reduces the stopband attenuation near the filter passband. In the example of FIG. 3 this second part 10 is reduced by 10 dB in the vicinity of the passband and then ramps up from the eighth subband to its final value.

A second part 11 of the frequency-dependent damping component of the AFB specification results from the fact that the interference components generated in the AFB by aliasing are evaluated differently in the SFB. The second part 11 is periodic so that the total stopband attenuation over the frequency is periodically modified. Here, the number of periods depends on the oversampling factor U, and the depth of the allowable attenuation reduction is determined by the product of the passbands of the prototype filters in the AFB and SFB.

The total stop attenuation 12 results from the sum of all attenuation components including the fundamental attenuation, based on a logarithmic measure (decibel). In FIG. 3 the entire frequency-dependent stopband attenuation 12 is shown. Their absolute level results from the basic attenuation a ^AFB (in FIG. 3 not shown), which is added in the logarithmic measure to the frequency-dependent stopband attenuation. The frequency-dependent stop attenuation thus increases over the 32 subbands selected here according to the ramp of the first part 10.

The basic attenuation a ^{SFB of} the synthesis filter bank is likewise defined by the desired signal / interference distance SNRS ^FB . It results from imaging in the SFB. Again, the fundamental attenuation a ^{SFB is} dependent on the interpolation factor M of the SFB, which is equal to the decimation factor M of the AFB.

A first frequency-dependent attenuation component 13 increases the basic attenuation a ^{SFB of} the attenuation attenuation specification by the frequency-dependent amplification (attenuation = negative amplification in dB) with which the subband signals between the filter banks are manipulated. This gain-dependent component is important because the imaging components in the SFB are also enhanced.

A first part 14 of the second frequency-dependent damping component of the SFB specification (cf. FIG. 4 ) results, as in the AFB, in exploiting masking effects of the human ear. Thus, the specification of the stop band attenuation of the transfer function of the SFB filters in the vicinity of the filter pass band is reduced exactly as in the AFB. This reduction can also contribute to a reduction of the filter order.

Furthermore, a second part 15 of the second frequency-dependent damping component of the SFB specification results from the fact that U • (M-1) spectral components (images of the SFB) of different strengths come to lie in each channel. In one channel, there are M-1 main components (central region of a spectral distribution of signal and alias power) and KM-1 differentially attenuated secondary components (outer regions of the spectral distribution of signal and alias power). The specification of the stopband attenuation is thus modified periodically over the frequency, the number of the periods is dependent on the oversampling factor U and the depth of the permissible attenuation reduction is determined by the passband of the prototype filter (in the SFB).

The total stop attenuation 16 of the prototype filter for the SFB again results from the sum of all attenuation components. Here, too, the fundamental attenuation a ^{SFB determines} the absolute position of the stopband attenuation by adding it for the specification of the prototype filter to the frequency-dependent attenuation parts in the logarithmic sense.

The frequency-dependent reduction of the stopband attenuation 12 and 16 in the optimized specifications (see. FIG. 3 and 4 ) is now used to reduce, for example, the circuit complexity and the filter order. Alternatively, in addition or implicitly (in the case of a fixed relationship between filter order and group delay given a given type of filter), the frequency-dependent reduction of the stopband attenuation can also be used to reduce the group delay. Optionally, the frequency-independent basic attenuation of the filter banks can be increased if the stopband attenuation can be reduced in a frequency-dependent manner. These advantages can be used individually or in combination.

Claims (9)

1. Filter bank system for a hearing aid having

   - an analysis filter bank (AFB) for splitting an input signal (E) into sub-band signals,

   - a processing device for manipulating (M1 to M32) at least one of the sub-band signals and

   - a synthesis filter bank (SFB) for combining a manipulated sub-band signal with at least one further one of the sub-band signals,
   characterized in that

   - the stop-band attenuation (12) of at least one of the transfer functions of the analysis filter bank (AFB) is composed of

   ○ a separately configurable, frequency-independent analysis pass-band attenuation component and

   ○ a separately configurable, frequency-dependent analysis attenuation component (10, 11), and/or

   - the stop-band attenuation (16) of at least one of the transfer functions of the synthesis filter bank (SFB) is composed of

   ○ a separately configurable, frequency-dependent synthesis pass-band attenuation component,

   ○ a separately configurable, frequency-dependent and manipulation-dependent first synthesis attenuation component (13), and

   ○ a separately configurable, frequency-dependent second synthesis attenuation component (14, 15).
2. Filter bank system according to Claim 1, wherein all attenuation components are configured by means of the signal/noise ratio at the output of the filter bank system.
3. Filter bank system according to Claim 1 or 2, wherein the analysis pass-band attenuation component depends on the down-sampling factor of the analysis filter bank (AFB).
4. Filter bank system according to one of the preceding claims, wherein a masking effect of human hearing is taken into consideration in the frequency-dependent analysis attenuation component (10, 11) and/or second synthesis attenuation component (14, 15).
5. Filter bank system according to one of the preceding claims, wherein the frequency-dependent analysis attenuation component (10, 11) is periodically modified, the periodicity being determined by the oversampling factor of the analysis filter bank (AFB) and the maximum reduction in attenuation being determined by the pass-bands of the analysis filter bank (SFB).
6. Filter bank system according to one of the preceding claims, wherein the synthesis pass-band attenuation component depends on the down sampling factor of the synthesis filter bank (SFB).
7. Filter bank system according to one of the preceding claims, wherein the frequency-dependent first synthesis attenuation component (13) depends on an amplification of the processing device.
8. Filter bank system according to one of the preceding claims, wherein the frequency-dependent second synthesis attenuation component (14, 15) is modified periodically, the periodicity being determined by the oversampling factor of the synthesis filter bank (SFB) and the maximum reduction in attenuation being determined by the pass-band (transfer characteristic) of the SFB.
9. Hearing device with a filter bank system according to one of the preceding claims.

Images