CN101924977B - Method of adjusting signal processing parameter for first and second hearing aid and hearing aid constituting signal processor - Google Patents

Abstract

The present invention relates to a method of adjusting a signal processing parameter for a first and a second hearing aid forming part of a binaural hearing aid system to be worn by a user, the binaural hearing aid system comprising a user specific model representing a desired asymmetry between the first ear and the second ear of the user, the method comprising the steps of detecting a request for a processing parameter change at the first hearing aid, adjusting the signal processing parameter in the first hearing aid in response to the request, adjusting a processing parameter for the second hearing aid in response to the request and based on the user specific model, and wherein the method further comprises the step of adapting the model with respect to the desired asymmetry during use of the binaural hearing aid system.

Description

Adjust the method for signal processing parameter of the first hearing aids and the second hearing aids and the hearing aids that comprises signal processor

Technical field

The present invention relates to a kind of method of adjusting hearing aids processing parameter, relate in particular to the ears hearing assistance system with the first hearing aids and the second hearing aids, each comprises microphone the first hearing aids and the second hearing aids, for responding the A/D converter that the voice signal receiving at acoustic environment microphone separately provides digital input signals, be applicable to according to predetermined signal processing algorithm process digital input signals to generate the processor of the output signal of processing, D/A converter, and for the voice signal of processing being separately converted to the output transducer of voice output signal.

Background technology

Hearing aids typically comprises digital signal processor (DSP), and the sound receiving for the treatment of hearing aids is to make up user's hearing loss.Commonly known in the art, the processing procedure of DSP is controlled by signal processing algorithm, and this algorithm has the various parameters of processing for adjusting the signal of actual execution.In multichannel hearing aids, the gain of each passage is exactly the example of this class parameter.

The flexibility of DSP can be used to provide a plurality of parameter sets of a plurality of different algorithms and/or special algorithm.For example, can be noise suppressed polyalgorithm is provided, be i.e. the decay of unwanted signal and the amplification of desired signal.Desired signal is normally talked or music, and unwanted signal can be confused noise, music (when talking as desired signal), the traffic noise in background conversation, restaurant etc.

Different algorithms or parameter set provide the tonequality of comfortable and clear reproduction under alternative sounds environment such as talk, delphic talk, restaurant's confused noise, music, traffic noise.The audio signal obtaining from alternative sounds environment can have extremely different features, for example average and maximum sound pressure level (SPL) and/or frequency content.In hearing aids, the acoustic environment of every type can be associated with specific program, and wherein, the algorithm parameter of signal processing algorithm specific is arranged on the treated sound with optimum signal quality is provided under specific sound environment.This type of parameter set can typically comprise the parameter relevant with the corner frequency (conner frequency) of bandwidth gain, frequency selectivity filtering algorithm or slope and control such as automatic gain and control the knee point (knee-point) of (AGC) algorithm and the parameter of compression ratio.

Hearing aids can provide some different programs, and each program is applicable to specific sound environment category and/or particular user preferences.Signal processing parameter can determine joining during listening joining of shop (dispenser ' s office) to listen session (fitting session) at first, and by activate hearing aids nonvolatile storage respective algorithms and algorithm parameter and/or respective algorithms and algorithm parameter are sent to nonvolatile storage and are programmed in this device.

When listening barrier user to have on left and right hearing aids, conventionally wish that two hearing aidss can operate according to synchronous to a certain extent mode.Problem is: need the synchronous of much degree, need the synchronous of which kind of type, and user needs the synchronous of which kind of type under which kind of environment.A complicated problem is, owing to may depending on environment for the symmetric preference of ears hearing assistance system, may change during use, or be difficult to simply to join and to listen process and pre-determine based on laboratory, be therefore difficult to join just pre-determine after listening session desired synchronously.

The recent research of disclosed " the online personalization of auditory prosthesis " by name on the EURASIP magazine of processing about audio frequency, talk and music, volume in 2008, article numbers 183456, the 14 pages, 2008 years.Doi:10.1155/2008/183456, author Alexander Ypma, Job Geurts, Serkan

, Erik van der Werf and Bert de Vries, wherein require one group of 10 people listen the noise reduction parameters of barrier user individual two devices time, show the asymmetry of some participant's preferences in ears hearing assistance system.

In order to configure ears hearing assistance system, user need to adjust respectively two hearing aidss in left and right at present.For user, this and bilateral user interactions hearing assistance system is considered to a burden.

Left and right hearing aids can for example communicate with one another by the wireless connections between hearing aids.Utilize this structure, can be by use the comprehensive knowledge about symmetry and asymmetry left and right preference with the synchronous hearing aids of asymmetry mode, that is, benefit from easily synchronously but allow asymmetry simultaneously.

In addition, the model of asymmetry hearing loss and/or preference can be used to predict that asymmetry parameter changes.Further, in ears hearing assistance system, user can be used to infer the adjustment to another hearing aids to the adjustment of one of them hearing aids, or even only based on part (left device or right device) input, upgrade the setting of ears hearing assistance system.

Summary of the invention

A first aspect of the present invention provides a kind of method of signal processing parameter of the first and second hearing aidss of adjusting a part that forms the ears hearing assistance system that user wears, this ears hearing assistance system comprise represent the first ear of user and the second ear represent between user's special purpose model of asymmetry of expectation, the method comprises the steps:

The request that detection changes processing parameter in the first hearing aids,

In response to this request, be adjusted at the signal processing parameter in the first hearing aids,

In response to this request and based on user's special purpose model, adjust the processing parameter of the second hearing aids.

The method also can further be included in adapt to the step of (adapt) model during ears hearing assistance system with respect to desired asymmetry.

The step of adaptive model can be carried out after the second request of adjusting the processing parameter of the second hearing aids and changing in response to the parameter of processing the second hearing aids.For example, the new adjustment of the dissatisfied processing parameter for the second hearing aids of user and manually adjust the second hearing aids by the second request.Preferably, for manual adjustment, if be only to carry out in the short predetermined time interval after adjusting the step of processing parameter of the second hearing aids at it, be just merged in model, second ask after the processing parameter of adjustment the second hearing aids, in predetermined time interval, to carry out.

The step of adaptive model can comprise the asymmetry behavior of revising observed acoustic environment.For example, the environmental data of the step of adaptive model based on statistical disposition.For example, the specific setting of the processing parameter of the first and second hearing aids can represent the specific balance (compromise) of the type that depends on acoustic environment, user is conventionally in this acoustic environment moderate time, indicated as the environmental data of statistical disposition.

Detecting step can comprise for example by the tracer signal of hardware interrupts or other signal transmission forms or the request of parameter variation.

When for example passing through certain control, such as regulating wheel (as, volume wheel), the actuator of the remote equipment of button, toggle switch or control hearing aids, while operating one of hearing aids, the method according to this invention is synchronizeed another hearing aids with the first hearing aids, but preferably and not simply identical adjustment is copied to another hearing aids.The method according to this invention guarantees the difference in the preference of two ears and hearing loss to take into account.The measurement of the derivative of this model based on by for example audiogram or some other picture PAT.PTA is pure tone average, for example 500,1000, and the pure tone average of 2000Hz.

In ears hearing assistance system, the role of the first and second hearing aidss can be exchanged and be played the part of by left hearing aids and right hearing aids.

According to the model using in the method for first aspect present invention, it can be frequency dependence model.This be not while immobilizing may be favourable at whole frequency spectrum or on given frequency interval when hearing loss.

Be appreciated that hearing loss (hearing loss) word can be interpreted as referring to the first ear hearing loss and/or the second ear hearing loss.In other embodiment, hearing loss one word can be interpreted as referring to the difference of the hearing loss between the first ear and the second ear, and the data that also may comprise the other types of the asymmetry that for example may reflect any expectation.

In the method according to the invention, the request that changes processing parameter detected.Request may come from or the combination of event in several events, including, but not limited to the operation of an adjusting runner on one of hearing aids, press the button on one of hearing aids, to the operation of Long-distance Control of controlling hearing aids one or both of or communicating by letter with hearing aids one or both of, the equipment of monitoring of environmental sound or circuit or any combination wherein.

For example, ears hearing assistance system can be automatically by user's sound environment classification to one in a plurality of relevant or typical daily sound environment classification, such as the clamour in talk, unclean talk, restaurant, music, traffic noise etc.

Can be by the classification results obtaining for hearing aids separately automatically to select the signal processing feature of hearing aids, for example, for considered environment automatically switches to optimal algorithm.This hearing aids can be for other hearing aid user keeps optimum tonequality and/or speech intelligibility in various acoustic environments.

For example, may between comprehensive and shotgun microphone initialize program, switch, the level of being not only background noise of dependence, and also have other characteristics of signals of this background noise.Hearing aids user and another person, there is talk under background noise in the situation that identification the type of background noise is classified is useful.When noise is traffic noise, can select comprehensive operation, make user clearly hear approaching vehicle, and irrelevant with its arrival direction.On the other hand, if background noise is classified as unclean noise, select orientation to listen to program so that user listens to target talk by improved signal to noise ratio (snr) during talking.

The sound property of user's two ears may be obviously different, and the different situation of user's determined acoustic environment of two ears will occur, and this can cause carrying out different sound signal processing for every ear of user.

In the first hearing aids, process request and adjust relevant parameter or a plurality of parameter.Model or rule based on asking and represent by use hearing loss and/or the second ear preference, calculate, predict or determine the corresponding adjustment of the second hearing aids.Then correspondingly adjust the processing parameter of the second hearing aids.

The method according to this invention is utilized the existing knowledge about the hearing loss of every ear, and the existing knowledge of other hearing or psychophysics, and carries out the synchronous environmental information of adjusting in asymmetrical mode.

An advantage of the invention is that the request based on processing parameter is changed and further represent the particular model of the first ear hearing loss of wearer by use can be adjusted the signal processing parameter in the first hearing aids.This allow to use represents first ear and the second ear model of hearing loss or the hearing aid processing parameter of rule adjustment the first hearing aids in these two.When synchronization steering parameter level, advantage of the present invention is, in the both sides of hearing assistance system, retraining identical guiding parameter, still can be counted as asymmetry synchronous.This is because the asymmetry of left and right hearing aid parameter can be caused by the different acoustic fields of two ears.Guiding parameter (steering parameter) is for managing the parameter of the calculating of hearing aid processing parameter, from the environment descriptor such as sound property or sound classification output.Guiding parameter can also be acoustic environment to be associated with to the parameter of hearing aids processing parameter.These are not fixed as particular value.The mode that can also may reflect user preferences according to the value of the hearing aid parameter in specific environment in addition, is revised guiding parameter.

And user only needs to operate in hearing aids, yet adjust two hearing aidss to be suitable for the mode of left and right ear hearing loss.

As mentioned above, the request that processing parameter changes can be derived from the initialization operation of wearer to actuator, or can generate in response to the variation of characteristics of signals.Hearing aids can comprise the circuit that detects ambient sound environment, such as existing acoustic environment condition, such as noise situation, for example, due to sound of the wind noise or be derived from and around talk or other ambient noise source.

Processing parameter can be audio volume level in certain embodiments, but also can use other parameters, such as balance parameters, sound classification parameter, noise reduction parameters, noise reduction, compression ratio, time constant, classifier modules (classifier module) parameter, wave beam (directional process) parameter, feedback inhibition parameter, dynamic range compression parameter etc.In addition, can control or change super parameter.Super parameter (hyperparameter) itself is not hearing aids processing parameter.It is the parameter of the work of management processing algorithm, and it is typically fixed as particular value.

Special benefits of the present invention is that the request that can respond processing parameter variation carrys out adaptive model.If user or wearer be in specific environment situation, and correspondingly adjust hearing aids, model or rule can be adjusted in response to this change request or revise so.Expectation reduces the number of times that wearer need to adjust hearing aids, may increase thus the satisfaction of wearer to hearing aids.

Advantageously, the method according to this invention provides following possibility, and the request that processing parameter changes can comprise the information of the one or more processing parameters about changing and the parameter that represents change amount.This request can comprise the quantity that the information that will change about which or which parameter and described parameter change, for example, and the increase of volume or the amount of minimizing.

In one embodiment of the present of invention, the first hearing aids can be main equipment, and the second hearing aids can be from equipment.This allow user first, master hearing aid place makes a change separately, this variation will transmit or have influence on second, from hearing aids.May be also that two hearing aidss all can be taken on main equipment further, but be not simultaneously, mean that two equipment all can receive change request and transmit subsequently or apply this to change to another one equipment.

According to a favourable embodiment of the present invention, model can comprise two independent guiding vectors, its each relevant to the first ear of user and the hearing loss of the second ear respectively, wherein guiding vector is coupled by the probabilistic model that represents associating binaural system.

According in another favourable embodiment of the method for first aspect present invention, asymmetrical global level also depends on the difference between the microphone record of the first and second hearing aidss.

According to the present invention, the model of representative of consumer hearing loss can comprise hearing loss the first ear of user and/or the measurement of the second ear or that estimate.This hearing loss in two ears is favourable when not identical.

In further advantageous embodiment of the present invention, the request that processing parameter changes can be derived from user to the initialization operation of actuator or produce in response to the variation of characteristics of signals.For example request can be derived from other interactive mode of volume roller or user's operation.

At present preferably, according to the method for first aspect present invention, be not to carry out joining to listen under situation.Join listen situation conventionally by technical staff for example in laboratory or clinical execution.The method according to this invention, preferably when wearer uses under such as any situation in work, the large organization of assembling such as the Leisure Environment of having a dinner in restaurant etc. and everybody all other men.

The method preferably by wearer, in any noisy environment, use hearing aids in realize, if hearing impaired does not wear hearing aids and will feel under the weather in noisy environment.

Request can be based on parameter vector.The first and second hearing aids models can be to share or common parameter or parameter set or vector.

A second aspect of the present invention relates to the hearing aids that comprises signal processor, wherein, this hearing aids is applicable in use form the part of ears hearing assistance system and in use from another hearing aids, receives information, another hearing aids is also applicable to form the part of ears hearing assistance system, and the request of processing parameter variation and user's special purpose model of expression the first ear of user and the asymmetry of the expectation between the second ear that wherein signal processor is configured to based in another hearing aids are adjusted the signal processing parameter in this hearing aids.

Signal processor can further be configured to response request and adjust user's special purpose model.

According to the hearing aids of second aspect, can further be configured or be applicable to carry out mentioned relevant to any step of the method for first aspect present invention.

This model can be placed in to the first hearing aids or be placed in the second hearing aids.In optional embodiment, this model is placed in to the 3rd equipment, such as Long-distance Control, such as can wearable device or personal hand-held equipment of PDA, personal digital assistant, movement/cell phone etc.

In an embodiment of the present invention, model can be shared as follows between the first and second hearing aidss, the some parts of this model is placed in the first hearing aids, and some parts is placed in the second hearing aids.For example, in one embodiment, by in this model about the part of the ear hearing loss by the first hearing aids compensation is placed in the first hearing aids, and by this model about the part of the hearing loss of the ear by the second hearing aids compensation is placed in the second hearing aids.

In another embodiment, these parts of model can be overlapping, in certain embodiments can be completely overlapping, the first and second hearing aidss the two under extremely overlapping situation, can there is identical model.

Accompanying drawing explanation

Exemplary embodiment is below with reference to accompanying drawings described this invention in more detail, wherein

Fig. 1 schematically illustrates the sketch of ears hearing assistance system.

Fig. 2 is the schematic diagram of flow chart of the step of diagram first embodiment of the invention.

Fig. 3 is the optional diagram of first embodiment of the invention.

Fig. 4 is the schematic diagram of the first embodiment of the modification of the method according to this invention.

Fig. 5 schematically illustrates the second embodiment of the present invention.

Fig. 6 illustrates identical with the structure shown in Fig. 1 in fact structure.

Fig. 7 illustrates embodiments of the invention, and wherein any one in two hearing aidss all can be taken on the role of main equipment.

Fig. 8 A, 8B, and 8C is the emulation schematic diagram of second embodiment of the invention.

Fig. 9 is the schematic diagram of third embodiment of the invention.

Figure 10 is the schematic diagram of the revision of the 3rd embodiment.

Figure 11 is the schematic diagram of fourth embodiment of the invention.

Figure 12 is the schematic diagram of the 6th embodiment.

Figure 13 and 14 is the schematic diagram of people's hearing loss.

Embodiment

Referring now to accompanying drawing, be described in more detail hereinafter the present invention, exemplary embodiment of the present invention shown in the drawings.Yet the present invention may be with multi-form enforcement, and should not be limited to embodiment described here.More properly, provide these embodiment so that the disclosure is comprehensively with complete, and pass on fully scope of invention to those skilled in the art.Accompanying drawing is exemplary, and simplifies for clarity, only illustrates and understands details required in this invention, omits other details simultaneously.In the text, for identical or appropriate section, use identical Reference numeral.

Fig. 1 illustrates the simplified block diagram of binaural hearing aid 2.Binaural hearing aid 2 comprises two independently hearing aidss 4 and 6, and they are applicable to or are configured to communicate each other.Hearing aids 4,, each of 6 have input pickup 8,, 10, for example microphone and/or inductance coil (not shown), to provide electronic input signal.Hearing aids 4,, 6 also comprise audio signal processor such as compressor compresses device 12,14, volume controls 16,18 and such as the output transducer 20,22 of receiver.Binaural hearing aid 2 in Fig. 1 illustrates by principal and subordinate's structure, wherein, depends on the indicated hearing user loss model of processing module 24, in the volume of adjusting master hearing aid 4, controls after 16, automatically adjusts the volume of the second hearing aids 6.In this example, the adjustment of the hearing aids processing parameter of master hearing aid is the adjustment of volume, yet should be appreciated that, can be the hearing aids processing parameter of any type, and the adjustment of a kind of processing parameter of master hearing aid 4 be not the adjustment that must be accompanied by the identical type hearing aid parameter (yet having volume adjustment from hearing aids 6 this example) from hearing aids 6 yet.In addition, should be appreciated that, the adjustment of processing parameter in master hearing aid (being volume in this example) can trigger automatically, for example, and the automatic change of master hearing aid Program.The automatic change of this program can be by the change triggers of the ambient sound environment of for example binaural hearing aid 2.Model processing modules 24 can be merged in any one in hearing aids 4 and 6.Should be appreciated that in this embodiment, is optional from the volume control 18 of hearing aids 6.

Fig. 2 illustrates the schematic diagram of flow chart of the step of first embodiment of the invention.

The method according to this invention relates in ears hearing assistance system and adapting to, adjusts or change signal parameter.Ears hearing assistance system comprises two hearing aidss, one of them left ear for wearer or user and another auris dextra for wearer or user.Binaural hearing aid in this specification is called to the first and second hearing aidss.Under different situations, the role that left hearing aids and right hearing aids can be taken on the first and second hearing aidss.A quilt in hearing aids operates or receives the request that changes processing parameter, and this hearing aids is called as the first hearing aids, and another is synchronous by asymmetry pattern quilt.This another hearing aids is called as the second hearing aids so.

Receive the request (step 26) that changes processing parameter.This request comprises the indication that will change which processing parameter.In a particular embodiment, this request can comprise the indication of several parameters.Except the sign to parameter, this request also can comprise the indication of the variable quantity of parameter.

The request that changes processing parameter can be generated by one of several equipment or unit.Conventionally one or two in binaural hearing aid comprises volume roller.This volume roller can produce the request that changes processing parameter.This request indication that processing parameter should reformed amount of can enclosing.

The method is further included in and in the first hearing aids, adjusts signal processing parameter (step 28).In one embodiment of the invention, directly in the first hearing aids, change or revise processing parameter, not considering the hearing loss of the first ear.

The method also comprises that the request and the particular model 32 that based on processing parameter, change determine that the processing parameter of the second hearing aidss changes (step 30), wherein the hearing loss of this model representation user the second ear and/or according to user preference separately the preferred asymmetry in the first ear and the second ear.

Therefore suppose that when user need to change the processing parameter of first ear at the first ear to the perception of sound based on user and operational example is as volume roller at the first ear, it is favourable can imagining this.

The model of the request that in one embodiment, the parameter of the method according to this invention based on to the first ear changes and the hearing loss of the second ear provides automatic change or the adaptation of the processing parameter of the second ear.The model of the request that in certain embodiments of the invention, the parameter of the method based on to the first ear changes and the hearing loss of the second ear provides automatic change or the adaptation of the same treatment parameter of the second ear.The model of the second ear is preferably frequency dependence model.

The example of asymmetry hearing loss comprises different loudness perception, the regenerations that difference is measured or L-R(wherein L-R represent L-R) hyperakusis, it causes one or more different threshold levels, different the most comfortable level (MCL level), different uncomfortable level (UCL level) or can select or measure the mapping of L-R level joining during listening.

And asymmetry SNR loss can affect L-R mapping curve, for example, with respect to comfortable or clarity preference.Seem to be difficult to prediction and hint test during listening or measure joining.

The method is also included within the second hearing aids and changes or to adapt to one or more signal processing parameters.Can in arbitrary hearing aids, carry out the calculating that the signal processing parameter of the first and/or second hearing aids changes or determine.In some embodiment of binaural hearing aid, two hearing aidss all comprise signal processing unit.Can be in a hearing aids signalization processing parameter, be then sent to another hearing aids.One of example is the ears hearing assistance system that wherein wireless connections by for example bluetooth or other applicable agreements of two hearing aidss communicate.Alternatively, two hearing aidss can be connected by electric conductor.

Fig. 3 illustrates the embodiment of ears hearing assistance system, and it is synchronous that wherein this system is used the asymmetry of left and right hearing aid parameter.

In a preferred embodiment, the model between two of ears hearing assistance system hearing aidss or transmitting function can provide the non-linear or asymmetry transmitting function of the change producing to another hearing aids from a hearing aids.

Preferably, if user only controls the first hearing aids, the second hearing aids can be synchronizeed in asymmetry mode with the first hearing aids.For most of listening situation, for user, this is completely acceptable.

For example, if user operates the volume adjusting roller of a hearing aids in ears hearing assistance system, and left ear and auris dextra have different audibility ranges, the volume change of the volume change of the second hearing aids and the first hearing aids may be different, thereby cause increase or the minimizing of the loudness aspect that two ears perceive equally.In this case, the embodiment of system described herein allows the operation based on carrying out on volume roller and represents that the model of difference of user's audibility range adjusts the second hearing aids automatically.Like this, user does not need to adjust respectively each of two hearing aidss.

In certain embodiments, configurable system is calculated in the first ear because of the magnitude of volume adjustment with respect to total change in gain of the audibility range of the first ear, then in the second ear, issue (issue) change in gain, this change in gain has identical magnitude with respect to the audibility range of the second ear.

In Fig. 3 to Figure 12, subscript L and R represent respectively left and right.In Fig. 3, left and right sound import represents with x, and by hearing aids, HA processes, the sound y that HA output was processed.

The sound y of output is input to left ear and auris dextra E, is converted into left and right auditory nerve signal n combined in brain, in brain, it is observed, integrated and estimates.The ears of the left and right sound based on this was processed are integrated and are estimated, the user d that can make decision adjusts left and/or right hearing aids.

This will cause adjusting, and this is adjusted to form and proofreaies and correct r, is published in some way hearing aids.

The mapping that study module L proofreaies and correct r by rule learning the application of regulation from user.Only, in the situation that r is proofreaied and correct or adjusts in a device issue in ears hearing assistance system, rule is at the device of adjusting with at given ears effectiveness (utility) model uother device in calculate optimum hearing processing parameter θ.In the simplest situation, such utility models transmit the left and right ear hearing loss HL about patient to model or rule _land HL _rinformation.Generally speaking, utility models can comprise sense of hearing summary (auditory profile) α, and the information that it comprises a relevant left side and/or right hearing loss can also comprise other parameters that reflect hearing user loss, sound preference and/or life style aspect.Utility models also can comprise effectiveness parameter ω.Study module can comprise parameter beta, and parameter beta is managed from being adjusted to the mapping of parameter.In this first embodiment, the calculating of left and right processing parameter in regulation management study module, is indicated by the arrow from rule to study module.The selection of fixing mapping f (.) arranges to represent by some of parameter beta, and parameter beta is by regulation management.In other embodiments, mapping can be unfixed, and is variable.

The behavior of this example can carry out modeling with renewal equation:

$\left[\begin{array}{c}{\mathrm{\θ}}_t^L\\ {\mathrm{\θ}}_t^R\end{array}\right]=\left[\begin{array}{c}{\mathrm{\β}}_t^L\\ f({\mathrm{\β}}_t^R{r}_t^L;{\mathrm{HL}}_L,{\mathrm{HL}}_R)\end{array}\right]+\left[\begin{array}{c}{r}_t^L\\ r_t^R\end{array}\right]---\left(1\right)$

Wherein output

with

that left and right hearing aids is being agreed to parameter (row) vector of k constantly,

with

the previous value of left and right hearing aid parameter vector, and

it is the user's updating vector at moment k to left hearing aids.In addition,

be certain (may for the nonlinear) scaling that is applied to the left hearing aid user updating vector of auris dextra, and considered the hearing loss of ears.In practice, hearing aid parameter vector is one dimension typically, but can provide the suitable user's updating vector having more than one dimension as user time, Multidimensional Parametric Vectors also can be synchronous by asymmetry ground.

In this embodiment, timestamp t is defined as the ongoing time, for example, when the multiple of sampling period 1/Fs, measure, and wherein Fs is the sample frequency of digital deaf-aid processor.

Agree to that (consent time) k is also defined as timestamp t constantly _k, at t _kthe user of place provides the explicit agreement to specific adjusted.In order to affect the acoustic processing function of hearing aids, user operates and controls function (roller, button, Long-distance Control or some other user control interface).The moment (and it is not changed at specific time quantum) that user sends user's control is known as and agrees to instantaneous (consent moment).Agree to that instantaneous k refers to send k that (and make its constant) control constantly.In certain embodiments, when execution user is synchronous to the asymmetry of control adjustment, system is configured to agreeing to instantaneous work.In the agreement moment of equation (1), the left and right hearing aid parameter vector of k is applied in hearing assistance system, as at the instantaneous k of current agreement and the next new processing parameter of agreeing to any moment between instantaneous k+1, upgrades

with

be used as t _kand t _k+1between timestamp place

with

.Similarly rule is used to the guiding parameter agreeing to the renewal at place to be constantly transformed into the random time stamp during the online processing of sound import.In one embodiment, user can select non-linear zoom function:

$f(r_k^L;{\mathrm{HL}}_L,{\mathrm{HL}}_R)=\mathrm{scalebacl}(\mathrm{scale}(r_k^L;{\mathrm{HL}}_L);{\mathrm{HL}}_R)$

Wherein, scale (.) function adjustment at left hearing aids according to left ear hearing loss convergent-divergent, and scaleback (.) function is used and should " the convergent-divergent adjustment of perception ground " calculate adjustment according to auris dextra hearing loss.Thereby right hearing aid parameter is synchronizeed with left hearing aids, but use the left hearing aid correction of revising, allow the asymmetry between hearing aids.Further, by a correction that is only published to left hearing aids, for proofreading and correct two hearing aidss, it avoids operating two controls, and this is considered to be user favourable.

Replace the equation that enforcement or embodiment can utilize renewal:

$\left[\begin{array}{c}{\mathrm{\θ}}_k^L\\ {\mathrm{\θ}}_k^R\end{array}\right]=\left[\begin{array}{c}{\mathrm{\θ}}_{k-1}^L\\ {\mathrm{\θ}}_{k-1}^R\end{array}\right]+\left[\begin{array}{c}{r}_k^L\\ f(r_k^L;{\mathrm{HL}}_L,{\mathrm{HL}}_R)\end{array}\right]---\left(2\right)$

Non-linear zoom again applies left hearing aid correction so that the variation perceiving in left hearing aids is similar to the variation perceiving in right hearing aids.Yet except the hearing loss of two ears, this function is considered the preceding value of right hearing aid parameter vector equally now.The further user correction of right hearing aids

normally 0, but allow if necessary user in right hearing aids, to carry out extra fine setting.In certain embodiments, further user is proofreaied and correct can be by model learning or absorption, and therefore the hearing loss of an ear of this model representation improves following adjustment based on this model.

Notice in above-mentioned example, that left hearing aids is played the part of is the role of the first hearing aids, but role can exchange.For example in other embodiments, right hearing aids can be played the part of the role of the first hearing aids.

In other embodiments, also consider that the optimum that the difference of expressing parameter adjustment is controlled and the hearing loss of the adjustment from the first ear and two ears is calculated in another ear revises the different models that change.

Flow chart shown in Fig. 2 and Fig. 3 relate to above embodiment.

Fig. 4 is the schematic diagram of the first embodiment of the method according to this invention modification.Fig. 4 comprises the similar step in Fig. 2, similar Reference numeral numbering for similar step.

Except the step in Fig. 2, in Fig. 4, illustrated method comprises frame 36.It is used to indicate the hearing loss model that uses the first ear when carrying out or calculating for the processing parameter of the first hearing aids or the adjustment of a plurality of processing parameters.

The second embodiment of the present invention provides uses asymmetry user feedback and the synchronous left and right of asymmetry sound characteristic guiding parameter.In Fig. 5, illustrate the second embodiment.

Can be by introducing left and right hearing aids sound characteristic (OK) vector

with

expand the synchronous thought of asymmetry.These vectors are by a series of weight coefficients by sharing between two hearing aidss or guiding parameter beta _t, the parameter of two hearing aidss that lead, for example, use mapping:

$\left[\begin{array}{c}{\mathrm{\θ}}_t^L\\ {\mathrm{\θ}}_t^R\end{array}\right]=\left[\begin{array}{c}{s}_t^L\\ s_t^R\end{array}\right]\mathrm{\β}+\left[\begin{array}{c}{r}_t^L\\ r_t^R\end{array}\right]---\left(3\right)$

The Representation Equation left and right (scalar) hearing aids processing parameter of this system is used shares guiding vector φ along with acoustic environment is (by left and right sound characteristic vector

with

represent) and change.

Further, user is adjusted

with

add environmental orientation part to φ and

φ.In this embodiment, we will consider scalar auxiliary parameter vector

with

, but it is not limited to this embodiment this thought is applied to one-dimensional case, because in an alternate embodiment of the invention, can use equally the asymmetry of Multidimensional Parametric Vectors synchronous.

Notice our not designated user adjustment

with

how along with the time changes.For example,, as learning procedure

result, based on agreeing to the adjustment of k to left hearing aids constantly, we can be adjusted at this timestamp τ and be kept to

applied Learning step now.Otherwise we can make to adjust constant (therefore revise the sole mode of adjusting and pass through exactly user interactions).

One-component in each sound property vector can be arranged to 1, the deviation that is independent of environment is provided thus.Allow user to operate any one in hearing aids, or they the two, this will cause or left ear user correction

, or auris dextra user proofreaies and correct

, or the combination of left and right ear user correction.Share guiding vector β such as the predefines such as prior art, user preference that can utilize relevant hearing loss.

In addition, on-line study method can be designed to merge user's correction and upgrade common weight vectors.In this article, during term " online " refers to and uses auditory prosthesis, this is contrary with off-line, is joining during listening joining of shop to listen session etc.Therefore, at the horizontal synchronization ears hearing assistance system of guiding parameter, but when the feature difference between ears and/or the user between ears proofread and correct difference, the actual hearing aid parameter that comes from this guiding may difference between ears.More specifically, propose hearing aid parameter to use linear Gauss model, it is also referred to as " output model ":

$\left[\begin{array}{c}{\mathrm{\θ}}_t^L\\ {\mathrm{\θ}}_t^R\end{array}\right]=\left[\begin{array}{c}{S}_t^L\\ S_t^R\end{array}\right]{\mathrm{\β}}_t+\left[\begin{array}{c}{\mathrm{\ϵ}}_t^L\\ {\mathrm{\ϵ}}_t^R\end{array}\right]---\left(4\right)$

Wherein, with

for thering is respectively variance

with 0 average Gaussian noise source, it is illustrated in the noise of agreeing to that k user adjusts constantly.Notice in model term φ _kfor stochastic variable, it represents current guiding vector, and it is used to estimate/be updated in the shared guiding vector φ of application in hearing aids processing.

Our modeling asymmetry alignment error and intrinsic user and noise source inconsistency, described noise source is the Gaussian random variable with different averages and covariance matrix.Further,

and φ _kwhile being, become stochastic variable, wherein, we will

as scalar and by φ _kas vector.As previously mentioned, can make and comprise multidimensional according to alternative embodiment of the present invention expansion.

The instantaneous k of ears of explicit agreement refers to now at timestamp t _kspecific " window lock in time " that place starts.Here user sends the control of hearing aids one or both of is revised to hearing aid parameter, then makes the controlling value of sending not add change in the specific time period.During this ears are agreed to instantaneous (hereinafter to be referred as " agreeing to constantly "), the hearing aid parameter value of expectation is known at least partly, and sound characteristic always can be retrieved from two hearing aidss of hearing assistance system.For modeling changes user preference, for example we suppose the differentiation of parameter, i.e. " state model ", for example, can be modeled as:

φ _k＝φ _k-1+ξ _k （5）

ξ wherein _kfor thering is covariance matrix Γ _kzero-mean Gaussian noise, for example, at state (guiding) variable φ _kdifferentiation in represent uncertain.By the Posterior Mean of computing mode variable, for example use more new formula of Kalman filtering, we can agree to constantly upgrade guiding parameter at each.Also can use other suitable formula.For example the special feature of this model is the renewal that utilizes recurrence least square or the poor acquisition of normalized minimum mean square.When, in lock in time, in window, carried out the timing of two hearing aidss, observed all sidedly " ears output vector " ${\underset{\‾}{\mathrm{\θ}}}_k=\left[\begin{array}{c}{\mathrm{\θ}}_k^L\\ {\mathrm{\θ}}_k^R\end{array}\right]$ " ears sound characteristic vector " ${\underset{\‾}{S}}_k=\left[\begin{array}{c}{S}_k^L\\ S_k^R\end{array}\right],$ And can use the more new formula of standard.Under Bayesian frame for example, we can derive as follows:

We define ears noise vector ${\underset{\&OverBar;}{\mathrm{\&epsiv;}}}_k=\left[\begin{array}{c}{\mathrm{\&epsiv;}}_k^L\\ {\mathrm{\&epsiv;}}_k^R\end{array}\right],$ It is according to having zero-mean and covariance matrix ${\mathrm{\&Sigma;}}_k=\left[\begin{array}{cc}{\mathrm{\&Sigma;}}_k^{\mathrm{<figure-callout\; id="0"\; label="L"\; filenames="H2009102468937E00061.png,H2009102468937E00062.png"\; state="\{\{state\}\}">L</figure-callout>}}& 0\\ 0& {\mathrm{\&Sigma;}}_k^R\end{array}\right]$ Normal distribution distribute.When observing all sidedly two output vectors in hearing aids and sound characteristic, output model equation (4) can be rewritten as:

${\underset{\&OverBar;}{\mathrm{\&theta;}}}_k={\underset{\&OverBar;}{S}}_k{\mathrm{\&phi;}}_k+{\mathrm{\&epsiv;}}_k---\left(6\right)$

Itself and state model equation (5) filter renewal equation in conjunction with the Kalman who produces below:

${\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}={\mathrm{\&Sigma;}}_{k-1}^{\mathrm{\&phi;}}+{\mathrm{\&Gamma;}}_k$

$K_k={\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}{\underset{\&OverBar;}{s}}_k^T{(\underset{\&OverBar;}{s}{\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}{\underset{\&OverBar;}{s}}_k^T+{\mathrm{\&Sigma;}}_k)}^{-1}$

${\widehat{\mathrm{\&phi;}}}_k={\widehat{\mathrm{\&phi;}}}_{k-1}+K_k({\underset{\&OverBar;}{\mathrm{\&theta;}}}_k-{\underset{\&OverBar;}{s}}_k{\widehat{\mathrm{\&phi;}}}_{k-1})$

${\mathrm{\&Sigma;}}_k^{\mathrm{\&phi;}}=(I-K_k{\underset{\&OverBar;}{s}}_k){\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}$

Wherein, we carry out the recursive estimation of the posterior probability of (sharing) ears guiding vector effectively,

$p\left({\mathrm{\&phi;}}_k\right|{\underset{\&OverBar;}{\mathrm{\&theta;}}}_1,...,{\underset{\&OverBar;}{\mathrm{\&theta;}}}_k)=N({\widehat{\mathrm{\&phi;}}}_k,{\mathrm{\&Sigma;}}_k^{\mathrm{\&phi;}})$

Under N (μ, Σ), we represent normal distribution with average μ and covariance matrix Σ.

When only presenting one of correction, output vector is just only partly observed, and does not observe the input corresponding to the parameter of the expectation of another hearing aids.We can be by integrating out such " half-proof " of hidden parts study of output vector.Renewal equation is followed Kalman and is filtered renewal equation, but when we have half-proof, we can integrate on the hidden parts of output vector, obtain slightly different renewal equations.For example, when we only observe user for ears hearing assistance system in the action of right device

time, we will only use right device users action to carry out

posteriority recursive estimation:

${\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}={\mathrm{\&Sigma;}}_{k-1}^{\mathrm{\&phi;}}+{\mathrm{\&Gamma;}}_k$

$K_k={\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}{s}_k^{\mathrm{RT}}{({s_k^R\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}{s}_k^{\mathrm{RT}}+{\mathrm{\&Sigma;}}_k^R)}^{-1}$

${\widehat{\mathrm{\&phi;}}}_k={\widehat{\mathrm{\&phi;}}}_{k-1}+K_k({\mathrm{\&theta;}}_k^R-s_k^R{\widehat{\mathrm{\&phi;}}}_{k-1})$

${\mathrm{\&Sigma;}}_k^{\mathrm{\&phi;}}=(I-K_k{s}_k^R){\mathrm{\&Sigma;}}_{k|k-1}^{\mathrm{\&phi;}}$

During user action on only observing left instrument, keep identical equation, but replace subscript R with subscript L.We use

the sound characteristic vector that is illustrated in right device place is allowing the transposition of k constantly,

transposition.

Because for left ear and auris dextra user action, we have different variance terms

with

, therefore the on-line tracing of these terms is caused in left ear and auris dextra user action for conforming different estimations.Based on previous expection (for example, when main body is left-handed person, he may experience inconsistency still less in his left side action) left ear and the conforming asymmetry of auris dextra, can be placed into for example as initial value with

in asymmetry.

The special feature of this model is to utilize recurrent least square method or poor the obtained renewal of normalization minimum mean-square, and it can be implemented based on the disclosure by those skilled in the art.

From above, can notice, can be based on estimating in the posteriority recurrence of determining any one parameter that leads that allows instantaneous left ear adjustment, auris dextra adjustment or associating left and right ear adjustment.Therefore, left ear device and the auris dextra device of our synchronous ears hearing assistance system in shared guiding parameter level, but the consistency of the asymmetry of asymmetry in allowing to adjust or adjustment.

The flow chart of another embodiment shown in Fig. 5.

Except described in Fig. 2, the adjustment of the left and right ear that likely noise adjustment is considered as hearing assistance system to combine, by considering the noise in left ear and/or auris dextra adjustment, applies it to two hearing aidss.In addition, study and oriented module L learn and apply the mapping from sound characteristic vector s to hearing aid parameter θ.The sound characteristic of particular type is identification feature, and in this case, for hearing aid parameter vector, parameter learning and guiding are trained the also value of personalized application effectively.Based on monophony environmental knowledge, by the feature extraction unit FE extraction environment sound characteristic of each hearing aids.At ears feature extraction unit FE _lRmiddle use binaural environment knowledge, these features are combined and adapt to each hearing aids, obtain subsequently " ears optimization " monophony characteristic vector σ.The example of relevant sound characteristic has: input RMS value, talk possibility, signal noise ratio, by the signal noise ratio of the frequency band importance function weighting of talking, environment classifier output etc.

In hearing assistance system and access customer adjustment, in Fig. 5, be illustrated as two arrows adjusting r that comprise of self-adjusting frame AD.Use ears utility models

, by adjusting Σ at left ear and auris dextra user ₀in previous inconsistent estimation, initial asymmetry is input to system.Because it is previous information rather than ongoing information flow, the arrow from utility models to study module is shown in broken lines.Yet by parameter learning and oriented module L, the mapping of processing parameter is adjusted in these initial estimation impacts, parameter learning and oriented module L are common (synchronous) guiding vector β sharing.

Below relate to the emulation to second embodiment of the invention, and at Fig. 8 A, shown in 8B and 8C.

In emulation, digitlization a piece of music, processes and plays to human user by artificial hearing aids.The model of the guiding coefficient based on expectation, suppose that human user accesses identical sound characteristic with artificial hearing aids, if exceed worry (annoyance) threshold value of corresponding ear, user's correction to left hearing aids, right hearing aids or two hearing aidss by issue.

The worried threshold value of pre-defined each ear, and can differ from one another for each ear.Based on the difference between guiding coefficient desired and that consciousness arrives in arbitrary ear, determine current worry amount.Further, the amount of user's inconsistency is added the desirable noise of proofreading and correct to when they are published, can be different for each ear, so the asymmetrical flexibility ratio of emulation (dexterity).Finally, the sound characteristic value in each ear can (extremely) difference, therefore at two ear emulation different acoustic fields, produces different left ears and auris dextra characteristic value.

Fig. 8 A, 8B and the study of 8C indicative icon are from the common guiding coefficient of asymmetry user input and asymmetry sound characteristic.

Now with reference to each the width figure in Fig. 8 A, 8B and 8C, this simulation result is discussed, Fig. 8 A, 8B and 8C have as the line number of indicated they in Fig. 8 A-8C, the row with Reference numeral 42 is first subgraphs, and the row with Reference numeral 52 is last width subgraphs.In all row, transverse axis represents sample number, in other words: the time.

Each sample is corresponding to the sample that plays to the music signal of human user.During playing, (jointly) of expectation guiding parameter a _tit is scalar.In order to simplify, suppose to be one-dimensional characteristic vector for each hearing aids.In Fig. 8 A, parameter changes according to line 54.Can find out estimated value β _t(in the title of subgraph, being called as θ) be the parameter a of " tracking " expectation very well _tvalue, only need a small amount of renewal.

Each is drawn cycle 56A-56J and represents a step of updating, and at a _teach conversion after, each conversion by minority almost the minority shown in overlapping cycle upgrade and be enough to adapt to new desired value.

In the second row 44, the noise during user proofreaies and correct is along with the time changes, and each ear has very big difference, correction noise or inconsistency that higher value representation is higher, and solid line 58 is left ears, dotted line 60 is auris dextras.At middle two rows 46,48, show the worried threshold value of two ears, the threshold value that higher value representation is higher.

When playing music, we have a lower worried threshold value from left ear, have the less worried section departing from the guiding parameter value of expectation and start.The worried threshold value of auris dextra is very high, thus the user of right hearing aids proofread and correct by only in the situation that guiding parameter very large depart from or change just and issue.Then worried threshold value is contrary at second segment, so compared with the correction to left hearing aids, to more easily issue of the correction of right hearing aids, be all low at the 3rd section to two ears, be all high, and finally again equate with first paragraph at the 4th section to two ears.

We can see which user shown in the first row proofreaies and correct generation tracking behavior now.The guiding parameter a of expectation _tthe first conversion according to some users that issue in left hearing aids, proofread and correct to learn, near time samples 130, be shown the small leak 62 of row 50, its expression is published to one group of noise of left hearing aids and proofreaies and correct.During near time samples sample 130, right hearing aids issue not to be proofreaied and correct, this can find out, during these time samples, at null value place, be flat in the figure that auris dextra user proofreaies and correct.

On the other hand, from being published to the user of right hearing aids, proofread and correct near the conversion of time samples 1300 of following the tracks of row 52.Memory is low to the worried threshold value for auris dextra in this part now, so compared with the correction to left hearing aids, will more easily issue the correction of right hearing aids.Near conversion time samples 1800 is also the same.

During the 3rd section, near conversion time samples 2400 user in two hearing aidss proofread and correct to follow the tracks of.Ensuing three conversion are very large, so that they all exceed the threshold value of two ears, and also issue correction in two ears.

Finally, more small conversion only causes worry at left ear in the 5th section, and the correction based on left ear user completes tracking.

From this figure, do not see the asymmetry between the feature on hearing aids, namely identical characteristic extraction procedure is applied to the music signal of two hearing aidss, but the distortion and be taken as subsequently the characteristic value of right hearing aids in the situation that of considerable noise of the characteristic value in left hearing aids.

In emulation from the above description, can very clearly find out, can utilize from thering is all or part of evidence of different nonconforming left ears and auris dextra user correction and utilizing the different characteristic value in two ears to follow the tracks of common guiding parameter vector.Therefore, in hearing aids, user feedback can by asymmetry be published in hearing aids, and the symmetry of hearing aid parameter guiding will depend on the symmetry of sound field in ear.Further, it depends on the symmetry of extracted sound characteristic.Since hearing aids is shared common guiding vector, similarly sound field produces similar guiding hearing aid parameter vector, and vice versa.

Learning process can be processed all and/or half-proof, since can follow the tracks of user's nonuniformity in each hearing aids, and user's nonuniformity of the step-length of learning rules and estimation is inversely proportional to, feedback from " more consistent ear " will recently be followed the tracks of and make larger contribution from being fed back to of " more noisy ear ", and it is preferred.Therefore, above-described embodiment is the synchronous real asymmetry mechanism of hearing aids.

The 3rd embodiment is below described, in guiding parameter with

level rather than in hearing aid parameter level with

level is used synchronous thought.The 3rd embodiment shown in Fig. 9.

Yet, in the 3rd embodiment, synchronously will in the super parameter level in guiding parameter, occur, make to allow equally asymmetry guiding parameter.In other words, can carry out synchronously controlling those parameters of the distribution of left and right guiding parameter, rather than synchronization steering parameter itself.

Left and right guiding parameter is coupled by common probabilistic model, and it comprises left and right ear hearing loss, but may be also user preference function.

Basic principle is that, if considering that " natural asymmetry " is afterwards by synchronously in whole hearing assistance system, user is provided as preferred by perception hearing aid parameter.This will partly depend on the asymmetry of hearing loss, but is equally also subject to joining as the asymmetry of hearing aids the consideration of listening, and described asymmetric joining listens the output Geng center of permission left and right hearing aids to process (cerebral cortex).

Therefore,, by using asymmetry and asymmetry in hearing loss to join the inspiration of listening method, this embodiment provides a kind of method of knowledge of the previous investigation on asymmetric distribution of using guiding parameter.Do not have extra user to proofread and correct, this previous distribution will be arranged ears guiding; In addition, possible asymmetry ground, is used Bayesian frame in guiding parameter, user to be proofreaied and correct and is used for upgrading conventional probabilistic model, obtains the parameter that leads

with

the posteriority mean value of online updating.

More specifically, suppose to decompose as follows output model:

$\left[\begin{array}{c}{\mathrm{\&theta;}}_t^L\\ {\mathrm{\&theta;}}_t^R\end{array}\right]=\left[\begin{array}{c}{s}_{\mathrm{<figure-callout\; id="0"\; label="t\; L"\; filenames="H2009102468937E00061.png,H2009102468937E00062.png"\; state="\{\{state\}\}">t</figure-callout>}}^{L}0\\ 0s_t^R\end{array}\right]\left[\begin{array}{c}{\mathrm{\&beta;}}_t^L\\ {\mathrm{\&beta;}}_t^R\end{array}\right]+\left[\begin{array}{c}{\mathrm{\&epsiv;}}_t^L\\ {\mathrm{\&epsiv;}}_t^R\end{array}\right]---\left(6\right)$

Wherein, sound characteristic vector can comprise " constant " characteristic component, illustrates that left avertence is poor and/or right avertence is poor, and hearing aid parameter

with

, and guiding parameter with

or stochastic variable.Be modeled as the nonconforming left and right of user output noise

with

be modeled as again the Gaussian random variable may with different averages and covariance matrix.

The left and right branch of this model is coupled by implement ears asymmetry model in the parameter that for example leads.For example, we can propose the parameter that leads, and it is regarded as stochastic variable again, depend on conditionally " previous asymmetric knowledge " in the hearing aids of left and right, by following distribution, are represented:

$p\left(\left[\begin{array}{c}{\mathrm{\&phi;}}_k^L\\ {\mathrm{\&phi;}}_k^R\end{array}\right]\right|\underset{\&OverBar;}{U}(\mathrm{\&omega;},\mathrm{\&alpha;}))$

With " ears utility function "

represent previous asymmetry knowledge, it can be incorporated to by left and right effectiveness parameter ω and/or by some asymmetry of the model representation of some preferred asymmetry (user preference model) joins and listens method, and the model of these some preferred asymmetry (user preference model) is represented by " user's asymmetry parameter " α.Notice that left and right ear hearing loss is by a part that is user's asymmetry parameter.

Use Bayes's technology for example can calculate the most probable of given such ears asymmetry model or maximum posteriority guiding parameter, and about " observation " α of hearing user loss, life pattern, deep layer hearing condition etc.Further, when (may be asymmetry) user adjustment is applied to ears hearing assistance system, Bayes's technology allows to upgrade previous ears asymmetry model, and the Posterior Mean of the modification of guiding parameter is used in thread environment guiding.

Notice by use common utility models in two hearing aidss of ears hearing assistance system left and right guiding parameter

with

can not move freely, but be similar to a certain extent in soft mode, limit.As the example of a restriction, even can be to carry out directly (hard) constraint in the difference allowing in the guiding parameter of left and right." restriction " ears utility models will be encouraged more synchronization steering parameter more at most, and vice versa.Study action occurs as being applied to the adjustment result of one or more hearing aidss.As the result of these adjustment, pass through to upgrade (study action) in utility models and/or by the restriction of adaptation left and right ear guiding parameter, will cause upgrading left and right ear guiding parameter and the parameter in two hearing aidss.

The flow chart of above-described embodiment has been shown in Fig. 9.It is compared difference with Fig. 5 and is, is solid arrow between from utility models to study module.These arrows represent the information flow of ongoing current (left ear and auris dextra) effectiveness about experienced sound y.Another difference is exactly that the solid arrow that the expression user who sends from AD unit adjusts the ongoing stream of r offers ears utility models rather than study module now.Can find out, study module is now based on left and right ear effectiveness rather than left and right ear adjustment and be updated.

For example, if to the r that makes adjustment in hearing aids, the observation based on new can be upgraded preferred asymmetry quantity in ears utility models.According to the value of utility u upgrading, left and right ear guiding parameter is also modified.

The present invention comprises the distortion of the 3rd embodiment equally, and in the 3rd embodiment, effectiveness u is used the certain methods of restriction left and right ear guiding parameter and is combined, that is, ears parameter model, it is conversely by vectorial ξ parametrization.Now shown in Figure 10 and illustrate the flow chart of the revision of the 3rd embodiment.

Shown in Fig. 9, we directly limit left and right ear guiding parameter by ears parameter model now.The character of restriction (allow suitable asymmetry or may Complete Synchronization guiding parameter) is pointed to the solid arrow of ears parameter model from ears utility models at (modification) effectiveness u() impact be modified.In addition, due to ears parameter, restriction can affect two study module L, and it is represented by two-way (dotted line) arrow from study module to ears parameter model.

The fourth embodiment of the present invention has been described MS master-slave configuration.

Fig. 6 is illustrated in identical with the structure shown in Fig. 1 in essence structure.Yet, in this embodiment, because the variation of signal processing parameter in the first hearing aids causes the variation of the second hearing aid signal processing parameter after the automatic renewal of signal processing parameter in the second hearing aids, Renewal model 24.

As previously mentioned, hearing aids 4 is master hearing aids, and hearing aids 6 is from hearing aids.As previously mentioned, after the adjustment of volume control 16, be accompanied by according to the adjustment of the volume of the hearing aids 6 of model 24.Yet if dissatisfied this of user adjusted, and the volume by subsequently controls 18 adjustment it is proofreaied and correct, the positive indication that the adjustment of advising with model 24 so there are differences can be used for Renewal model 24.It is illustrated by dotted arrow 38.Preferably, if be performed in its shorter predetermined time interval after the adjustment of volume control 16, the adjustment of volume control 18 is only merged in model 24, because otherwise it probably can not control 16 first adjust and cause by volume, but more may be caused by the variation of acoustic environment.

Fig. 7 indicative icon wherein any in two hearing aidss in ears hearing assistance system all can be used as the structure of master hearing aid.

Ambient sound environment can affect renewal or the modification of illustrated model in Fig. 6 and Fig. 7.Ears hearing assistance system can detect the ambient sound environment of user in which kind of type at any given time.If noise situation for example detected, user wishes that change can be by the signal processing parameter of ambient sound environmental impact.In response to the change of ambient sound environment, model and/or signal processing parameter can be changed automatically.

In each example, user or wearer change signal processing parameter, can adapt to or revise the model for each ear.Its dotted arrow 38 and 40 diagrams in Fig. 7.

The 5th embodiment relates to the switching between the different synchronistic models outside first to fourth embodiment.

Except the feature of embodiment mono-to four discussion above, embodiment also can comprise discrete " synchronistic model " variable, and it controls " total amount of asymmetry " in ears hearing assistance system.

As example, " height " value of synchronistic model variable makes it similar restricted guidance parameter, and " centre " or " low " value will allow more deviations, and finally " closing " by the adjustment between can not synchronous two ears.When receiving calls (wherein ears hearing assistance system should for example be operated in asynchronous mode), the latter can be for example useful.The value of synchronous mode variable can be inputted by user (for example, by pressing the button), but also can be by on-line tracing.Above-mentioned study mechanism should be expanded subsequently to merge discrete mode and switch variable, and this can for example be used for tracing mode variable simultaneously by employing and filter to obtain with the switch card Germania of guiding variable.In Figure 12, synchronous mode switches as the asymmetry pattern that acts on " ears optimization " monophony characteristic vector σ and switches variable S.Yet, be also noted that user can directly affect pattern and switch (for example using button or Long-distance Control).Omitted arrow from ears integral unit to mode switch element to strengthen the readability of this figure.

In optional example, the value of switching variable S is set to " little ", its can by according to some range measurements, allow left and right guiding parameter only a small amount of difference realize.The amount allowing does not also rely on ears value of utility μ.

The sixth embodiment of the present invention comprises the present invention first to all features of the 5th embodiment, and it is synchronous further to comprise the asymmetry of any first parameter vector.Unit's parameter is the arbitrary parameter that impact is used to process the hearing aid parameter of sound.For example, how " study initiative " parameter is carried out the study of controlling guiding parameter in the hearing aids of left and right.This is the example of first parameter, rather than has previously planted sector of breakdown.Based on the conforming ongoing estimation in operation control wheel to user, it can be tracked.For example, the knowledge of tracked initiative, certified useful use left hearing aids when following the tracks of the initiative of right hearing aids in.

The 6th embodiment comprises any or all feature of the first to the 5th embodiment that relates to the parameter that leads.Yet, should catch any first parameter of the function of determining any hearing aids processing module.This can be the switching variable of determining the symmetry quantity of otoacoustic emission field, left and right, is used in the hearing aids of left and right to adapt to processing.Further, also should comprise non-guiding situation, by personalized, fixing but revisable, the first parameter not changing along with environment also should fall into this embodiment.

As frequency function, Figure 13 illustrates respectively people's the right side (R) ear and the chart of a left side (L) ear hearing loss.In this chart, show Hearing Threshold T (R) and the T (L) of given frequency f _ 0.For given frequency f _ 0, the loudness that left ear and auris dextra perceive is as the function of the sound import stress level (SPL) in two charts of Figure 14 and illustrate.

Refer now to the chart in Figure 13 and Figure 14, clearly, in order to obtain identical perception of sound loudness at place, frequency f _ 0, compare with auris dextra, left ear needs higher input SPL.In order to allow people in the left ear loudness identical with auris dextra perception, need in model 24, be incorporated to individual hearing loss model.

It is below hearing aid parameter

with

the non-exhaustive list of example, it can use the synchronous method of asymmetry according to the present invention and by synchronous.Applicable parameter list comprises: left and right grader output, volume, noise reduction parameters, beam forming parameter, feedback inhibition parameter etc.Certainly several in these parameters also can be simultaneously by synchronous.

Can be by the mode combination of any expectation according to the above-mentioned feature of the embodiment of the inventive method.

In one embodiment, can expect that synchronous feedback suppresses.Here we imagine left and right hearing aids each comprise and determining the feedback inhibition parameter of the feedback inhibition being employed.For example, in strong cycle situation, to switch the form of variable, such as occur the existence of pure tone two sound fields, and be zero if not this situation.Two periodic characteristic extraction procedure FE _land FE _rcan be applied to left and right hearing aids the two (as shown in Figure 2), and combining unit FE _lRcan be to two hearing aids output switching variablees, for the ears cycle be 1, otherwise be 0.The amount that each hearing aids can be estimated the ears cycle with it is subsequently to determine that the periodic sound of one of hearing aids is due to internal feedback or due to external tone input.

In another embodiment, hearing assistance system can provide a kind of method that detects near the phone of hearing aids.This detection can be carried out in the following way, and for example, by modeling and the typical feedback path of detection, it causes owing to holding phone near ear, or allows user manually specify " phone location " by some interface via hearing aids.When being 1 for left hearing aids phone detection variable, it can be regarded as feature extraction unit FE _loutput, and when being 0 for right hearing aids phone detection variable, the synchronous mode in hearing assistance system can be temporarily switched to specific " ear has telephony mode ".

Conceptive, it means that hearing assistance system will be switched to asymmetry pattern, and parameter wherein leads

arrange drive highly amplify, high feedback reduces and high directional pattern, and

low amplification, the comprehensive pattern of driving is set.When " ear has phone " pattern finishes, hearing assistance system turns back to " acquiescence asymmetry " pattern subsequently.

In the 3rd embodiment, it is contemplated that the synchro system that study is controlled, wherein the study of each ear is controlled at guiding parameter level by synchronous.For example, user needs left hearing aids study volume to control to arrange, and it is by the personalization coefficient that leads

determine guiding coefficient

with the setting for right LVC

identical.Output vector when hearing assistance system $\left[\begin{array}{c}{\mathrm{\&theta;}}_k^L\\ {\mathrm{\&theta;}}_k^R\end{array}\right]$ While comprising respectively left and right volume, it is carried out by the second embodiment.Therefore,, when the sound field of left and right hearing aids is different, user only experiences gain difference.Consequent acoustic processing may more reflect user's preferred process.In addition, one of volume roller of operation hearing assistance system, by causing the study of two guiding parameters in this system, therefore also causes the volume adjustment in (not operation) hearing aids.

In another embodiment, as the initial extreme case of ears soft handover directivity, the initial asymmetry that can consider the directivity parameter in two hearing aidss is joined and is listened.Here, one (for example left hearing aids) in hearing aids allows to switch, and right hearing aids in this example does not allow to switch, but it will remain on omnidirectional mode state always.This with by some left directional switching thresholds, guiding parameter

be set to some reasonable values, and by another hearing aid parameter

threshold value setting be infinite conceptive be of equal value.Then, thus user can personalizedly represent that the left and right ear guiding parameter of threshold value adjust initial, the complete asymmetrical setting of hearing assistance system by handling.Therefore, user can be in directivity switching behavior self-defined asymmetry, and make it depend on acoustic environment.In addition, the initial selected of " ear making well " obtains directed input, has low switching threshold, and " the bad ear making " obtains comprehensive input, be infinite switching threshold, it can be revised by user, and for example interested sound source is only from the bad ear making over there under specific circumstances.

Claims (23)

1. a method of adjusting the signal processing parameter of the first hearing aids and the second hearing aids, the part of the ears hearing assistance system that described the first hearing aids and the second hearing aids formation user wear, described ears hearing assistance system comprises user's special purpose model, described user's special purpose model is illustrated in the first ear of this user and the asymmetry of the expectation between the second ear, and the method comprises the steps:

Detection is in the request that processing parameter is changed of described the first hearing aids,

In response to this request, adjust the signal processing parameter in described the first hearing aids,

In response to this request and based on described user's special purpose model, adjust the processing parameter of described the second hearing aids, and

Wherein, described method further comprises step:

After the step of processing parameter of adjusting described the second hearing aids, by described ears hearing assistance system, the asymmetry with respect to described expectation during using this ears hearing assistance system adapts to described user's special purpose model.

2. the method for claim 1, wherein after the processing parameter of adjusting described the second hearing aids and in response to the second request that processing parameter is changed at described the second hearing aids, carry out the step that adapts to described user's special purpose model.

3. method as claimed in claim 2 wherein, after adjusting the processing parameter of described the second hearing aids, is carried out described the second request in predetermined time interval.

4. the step that the method for claim 1, wherein adapts to described user's special purpose model is the environmental data based on statistical disposition.

5. the method as described in arbitrary claim above, wherein, represents that described user's special purpose model of the asymmetry of described expectation comprises the first ear of described user and/or the hearing loss of the second ear measurement and/or estimation.

6. the method for claim 1, wherein described user's special purpose model is frequency dependence hearing loss model.

7. the method for claim 1, wherein described user's special purpose model provides the non-linear transmitting function of the change producing to another hearing aids from a hearing aids.

8. among one that the method for claim 1, wherein described user's special purpose model is placed in the first hearing aids, the second hearing aids neutralizes other equipment.

9. the method for claim 1, wherein, described user's special purpose model can be shared as follows between the first and second hearing aidss: the part of the described user's special purpose model with respect to the first hearing aids is placed in to the first hearing aids, and the part of the described user's special purpose model with respect to the second hearing aids is placed in to the second hearing aids.

10. method as claimed in claim 9, wherein, by in described user's special purpose model about the part of the ear hearing loss by the first hearing aids compensation is placed in to the first hearing aids, and by described user's special purpose model about the part of the hearing loss of the ear by the second hearing aids compensation is placed in to the second hearing aids.

11. methods as claimed in claim 10, wherein, the described each several part of described user's special purpose model can be overlapping.

12. the method for claim 1, wherein described user's special purpose model be both placed in the first hearing aids and be also placed in the second hearing aids.

13. the method for claim 1, wherein described processing parameter be the parameter of audio volume level, reducing noise, compression ratio, time constant, classifier modules or any combination wherein.

14. the method for claim 1, wherein, the described request that processing parameter is changed comes from event or the composition of matter of selecting from subordinate's group: the operation of an adjusting runner on one of hearing aids, the push-botton operation on one of hearing aids, the operation to the Long-distance Control of controlling hearing aids one or both of or communicating by letter with hearing aids one or both of.

15. the method for claim 1, wherein the described request that processing parameter is changed come from equipment or the circuit of monitoring of environmental sound.

16. methods as claimed in claim 15, wherein, the equipment of described monitoring of environmental sound or circuit comprise the circuit that detects ambient sound environment.

17. methods as claimed in claim 15, wherein, the equipment of described monitoring of environmental sound or circuit can be automatically by user's sound environment classification to one in a plurality of relevant or typical daily sound environment classification.

18. the method for claim 1, wherein the described request that processing parameter is changed comprise: about the information of one or more processing parameters that will change and the parameter and/or the value that are used for representing the amount that changes, or the parameter value that will change to.

19. the method for claim 1, wherein described the first hearing aids be that main equipment and described the second hearing aids are from equipment.

20. the method for claim 1, wherein, described user's special purpose model comprises two independently guiding vectors, and each is associated with the first ear of described user and the hearing loss of the second ear respectively, and described guiding vector is coupled by the probabilistic model that represents the binaural system of combination.

21. the method for claim 1, wherein asymmetrical global level further depend on the difference between the microphone record of described the first hearing aids and described the second hearing aids.

22. 1 kinds of hearing aidss that comprise signal processor, wherein, described hearing aids is applicable in use form the part of ears hearing assistance system, and be applicable to receive information from another hearing aids, described another hearing aids is also applicable in use form the part of described ears hearing assistance system, wherein user's special purpose model is placed in described hearing aids or in described another hearing aids, and wherein said user's special purpose model is illustrated in the first ear of this user and the asymmetry of the expectation between the second ear, request and described user's special purpose model that wherein said signal processor is configured to based on processing parameter in described another hearing aids is changed are adjusted the signal processing parameter in described hearing aids, and wherein said signal processor is further configured to the asymmetry with respect to described expectation during using this ears hearing assistance system and adapts to described user's special purpose model.

23. as the hearing aids of claim 22, and wherein, described signal processor is configured to carry out defined any method step in any one of claim 2-21.

Images