EP0661905A2 - Method for the fitting of hearing aids, device therefor and hearing aid - Google Patents

Abstract

A setting method for a hearing aid and a device therefor are proposed, by means of which a model for the perception of a psychoacoustic quantity, especially the loudness, is parametrised for a standard group of persons (LN) and for an individual (LI). Setting information is determined on the basis of differences in the models, especially with respect to their parametrisation, by means of which information the signal transmission is designed or adjusted at a hearing aid (HG) ex situ, or controlled in situ, respectively. <IMAGE>

Description

The present invention relates to a method according to the preamble of claim 1, a device according to that of claim 23 and a hearing aid according to claim 39.

Definitions

A psycho-acoustic perception quantity is understood to be a quantity that is formed non-linearly, by individual laws of perception, from physical-acoustic quantities, such as frequency spectrum, sound pressure level, phase position, time course, etc.

Hearing aids known to date change physical, acoustic signal quantities in such a way that an hearing-impaired individual equipped with the hearing aid hears better. The hearing aid is adapted by adjusting physical transmission variables, such as frequency-dependent amplification, level limitation, etc., until the individual is satisfied with the hearing aid within the scope of the possibilities presented.

Although it is known, for which reference is made to the attached literature references, that human acoustic perception follows complex psycho-acoustically individual evaluations, these known phenomena have so far not been used to optimize a hearing aid.

Satisfactory corrections could thus be made with previously known hearing aids, essentially only averaged over all acoustic stimulus signals occurring in practice; mutual influences of signal quantities of the acoustic stimulus signals could only be considered unsatisfactorily, if at all. Non-linear phenomena of psycho-acoustic perception, such as loudness and loudness summation, frequency and time masking, were not taken into account.

It is an object of the present invention to provide a method or a device or a hearing aid of the type mentioned above which allow an individual, damaged, psycho-acoustic perception behavior to be corrected with respect to that of the norm, including the statistical normal perception behavior of people be understood.

This is achieved in a method of the type mentioned at the outset when it is carried out according to the characterizing part of claim 1, in a device of the type mentioned above when it is implemented in accordance with the characterizing part of claim 23.

Preferred embodiments of the method according to the invention are specified in claims 2 to 22, the device according to the invention in claims 24 to 38 and the hearing aid according to the invention in claim 40.

As will become apparent, the device according to the invention can be designed as an adaptation device separately from the hearing device. However, it also includes adjustment measures on the hearing aid in order to correct the perceived size taken into account for the individual.

The device according to the invention defined in the claims, the method according to the invention and the hearing aid according to the invention, together with any other inventive aspects, are subsequently described, for example, with the aid of Figures explained.

Fig. 1

schematisch, eine Quantifizierungseinheit zur Quantifizierung einer individuell wahrgenommenen, psycho-akustischen Wahrnehmungsgrösse;

Fig. 2

schematisch, in Form eines Blockdiagrammes, ein grundsätzliches erfindungsgemässes Vorgehen;

Fig. 3

in Abhängigkeit des Schallpegels, die wahrgenommene Lautheit der Norm (N) sowie eines schwerhörigen Individuums (I) in einem kritischen Frequenzband k;

Fig. 4

in Form eines Funktionsblock-Signalflussdiagrammes, eine erste Ausführungsvariante einer erfindungsgemässen Vorrichtung, nach dem erfindungsgemässen Verfahren arbeitend, womit erfindungsgemäss Stellgrössen für die Uebertragung eines Hörgerätes ermittelt werden;

Fig. 5

anhand einer Darstellung analog zu Fig. 3, eine vereinfachte graphische Darstellung des mit der Vorrichtung gemäss Fig. 4 vorgenommenen erfindungsgemässen Vorgehens;

Fig. 6a

vereinfacht, das Vorgehen nach Fig. 5, mit in

Fig. 6b

vereinfachter Darstellung des resultierenden Verstärkungsverlaufes in einem betrachteten kritischen Frequenzband, einzustellen am Uebertragungsverhalten eines erfindungsgemässen Hörgerätes, das in

Fig. 6c

in seinem prinzipiellen Aufbau betreffs Uebertragungsstrecke dargestellt ist;

Fig. 7

eine ausgehend von der Anordnung nach Fig. 4 weiterentwickelte Anordnung, bei der das in Fig. 4 implementierte Lautheitsmodell verfeinert implementiert ist;

Fig. 8

in Analogie zu Fig. 5, graphisch vereinfacht, das Verarbeitungsvorgehen an der Vorrichtung gemäss Fig. 7;

Fig. 9

über der Frequenzachse, schematisch, kritische Frequenzbänder der Norm und beispielsweise eines Individuums (a) mit einer beispielsweise resultierenden Korrekturverstärkungsfunktion (b), schallpegel- und frequenzabhängig, für einen einem betrachteten kritischen Frequenzband entsprechenden Hörgerät-Uebertragungskanal;

Fig. 10

analog zur Darstellung der Vorrichtung nach Fig. 4, deren Weiterentwicklung zur Mitberücksichtigung beim Individuum bezüglich der Norm veränderter kritischer Frequenzbandbreiten;

Fig. 11

in Analogie zur Darstellung von Fig. 10, eine erfindungsgemässe Vorrichtung, mittels welcher "in situ" ein erfindungsgemässes Hörgerät betreffs Uebertragungsverhalten eingestellt wird;

Fig.12a) und b)

je in Form eines Funktionsblock-Signalflussdiagram-mes, die Struktur erfindungsgemässer Hörgeräte, woran die Uebertragung einer psycho-akustischen Grösse korrigierend gesteuert wird, insbesondere die Lautheitsübertragung;

Fig. 13

eine Ausführungsvariante eines erfindungsgemässen Hörgerätes, woran die Vorkehrungen der Vorrichtung nach Fig. 11 sowie diejenigen nach Fig. 12a) kombiniert am Hörgerät implementiert sind;

Fig. 14

als Beispiel ausgehend von einer erfindungsgemässen Vorrichtung nach Fig. 11, deren Weiterentwicklung zur Mitberücksichtigung des Klangempfindens eines Individuums;

Fig. 15

ausgehend von der Darstellung eines erfindungsgemässen Hörgerätes nach Fig. 12b), eine bevorzugte Realisationsform, bei der die Korrekturübertragung einer psycho-akustischen Wahrnehmungsgrösse, am bevorzugten Beispiel der Lautheit, im Frequenzbereich aufbereitet wird;

Fig. 16

ausgehend von der Darstellung eines erfindungsgemässen Hörgerätes nach Fig. 15, dessen Weiterentwicklung zur Mitberücksichtigung einer weiteren psycho-akustischen Wahrnehmungsgrösse, nämlich der Frequenzmaskierung;

Fig. 17

schematisch, das Frequenzmaskierungsverhalten der Norm und eines schwerhörenden Individuums mit daraus sich ergebendem, qualitativ dargestelltem, zu realisierendem Korrekturverhalten an einem erfindungsgemässen Hörgerät nach Fig. 16;

Fig. 18

anhand einer Frequenz/Pegelcharakteristik, das Vorgehen zur Eruierung des Frequenzmaskierungsverhaltens eines Individuums;

Fig. 19

in Form eines Funktionsblock-Signalflussdiagrammes eine Messanordnung zur Durchführung des Ermittlungsverfahrens, wie anhand von Fig. 18 erläutert;

Fig. 20

über der Zeitachse einem Individuum präsentierte Signale bei der Eruierung, wie sie anhand von Fig. 18 erläutert wurde;

Fig. 21

ausgehend von einem erfindungsgemässen Hörgerät mit der in Fig. 15 bzw. 16 dargestellten Struktur, dessen Weiterentwicklung zur Mitberücksichtigung des Zeitmaskierungsverhaltens als eine weitere psychoakustische Wahrnehmungsgrösse;

Fig. 22

das vereinfachte Blockdiagramm eines erfindungsgemässen Hörgerätes, welches wie das in Fig. 21 dargestellte als weitere psycho-akustische Wahrnehmungsgrösse das Zeitmaskierungsverhalten berücksichtigt, aber in anderer Ausführungsform;

Fig. 23

die am erfindungsgemässen Hörgerät gemäss Fig. 22 vorgesehene Zeitmaskierungs-Korrektureinheit;

Fig. 24

schematisch, das Zeitmaskierungsverhalten der Norm und eines Individuums als Beispiel zur Erläuterung daraus resultierender Korrekturmassnahmen, um mit einem erfindungsgemässen Hörgerät das Zeitmaskierungsverhalten eines Individuums auf dasjenige der Norm zu korrigieren;

Fig. 25

schematisch, über der Zeitachse, bei der Eruierung des Zeitmaskierungsverhaltens einem Individuum zu präsentierende Signale.

Show it:

Fig. 1

schematically, a quantification unit for quantifying an individually perceived, psycho-acoustic perceptual quantity;

Fig. 2

schematically, in the form of a block diagram, a basic procedure according to the invention;

Fig. 3

depending on the sound level, the perceived loudness of the norm (N) and a hearing impaired individual (I) in a critical frequency band k;

Fig. 4

in the form of a function block signal flow diagram, a first embodiment variant of a device according to the invention, operating according to the method according to the invention, with which actuating variables according to the invention are determined for the transmission of a hearing aid;

Fig. 5

based on a representation analogous to FIG. 3, a simplified graphic representation of the procedure according to the invention carried out with the device according to FIG. 4;

Fig. 6a

simplified, the procedure according to FIG. 5, with in

Fig. 6b

Simplified representation of the resulting amplification curve in a critical frequency band under consideration, to be set on the transmission behavior of a hearing device according to the invention, which in

Fig. 6c

the basic structure of the transmission link is shown;

Fig. 7

an arrangement developed further from the arrangement according to FIG. 4, in which the loudness model implemented in FIG. 4 is implemented in a refined manner;

Fig. 8

in analogy to FIG. 5, graphically simplified, the processing procedure on the device according to FIG. 7;

Fig. 9

above the frequency axis, schematically, critical frequency bands of the standard and, for example, an individual (a) with a correction gain function (b), for example, resulting from sound level and frequency, for a hearing aid transmission channel corresponding to a critical frequency band under consideration;

Fig. 10

analogous to the representation of the device according to FIG. 4, its further development to take account of the individual with regard to the norm of changed critical frequency bandwidths;

Fig. 11

in analogy to the representation of FIG. 10, a device according to the invention, by means of which a hearing device according to the invention is set "in situ" with regard to transmission behavior;

Fig.12a) and b)

each in the form of a functional block signal flow diagram, the structure of hearing aids according to the invention, to which the transmission of a psycho-acoustic variable is correctively controlled, in particular the loudness transmission;

Fig. 13

an embodiment variant of a hearing aid according to the invention, on which the precautions of the device according to FIG. 11 and those according to FIG. 12a) are implemented in combination on the hearing aid;

Fig. 14

as an example, starting from a device according to the invention according to FIG. 11, its further development for taking into account the sound sensation of an individual;

Fig. 15

based on the representation of a hearing aid according to the invention according to FIG. 12b), a preferred form of implementation in which the correction transmission of a psycho-acoustic perceptual quantity, using the preferred example of loudness, is processed in the frequency range;

Fig. 16

starting from the representation of a hearing aid according to the invention according to FIG. 15, its further development to take into account a further psycho-acoustic perceptual variable, namely frequency masking;

Fig. 17

schematically, the frequency masking behavior of the norm and of a hearing-impaired individual with the resultant, qualitatively represented, to be realized correction behavior on a hearing aid according to the invention according to FIG. 16;

Fig. 18

based on a frequency / level characteristic, the procedure for determining the frequency masking behavior of an individual;

Fig. 19

in the form of a functional block signal flow diagram, a measuring arrangement for carrying out the determination method, as explained with reference to FIG. 18;

Fig. 20

Signals presented to an individual over the time axis during the investigation, as explained with reference to FIG. 18;

Fig. 21

starting from a hearing aid according to the invention with the structure shown in FIGS. 15 and 16, its further development to take the time masking behavior into account as a further psychoacoustic perceptual variable;

Fig. 22

the simplified block diagram of a hearing aid according to the invention which, like the one shown in FIG. 21, takes into account the time masking behavior as a further psycho-acoustic perceptual variable, but in another embodiment;

Fig. 23

22 the time mask correction unit provided on the hearing aid according to the invention;

Fig. 24

schematically, the time masking behavior of the norm and an individual as an example to explain the resulting corrective measures to correct the time masking behavior of an individual to that of the standard with a hearing aid according to the invention;

Fig. 25

schematically, signals to be presented to an individual over the time axis when determining the time masking behavior.

Psycho-acoustic perception, especially loudness and its quantification

The loudness "L" is a psycho-acoustic quantity, which indicates how "loud" an individual feels a presented acoustic signal.

Loudness has its own unit of measurement; a sinusoidal signal with a frequency of 1 kHz and a sound pressure level of 40dB-SPL produces a loudness of 1 "Sone". A sine of the same frequency with a level of 50dB-SPL is perceived exactly twice as loud; the corresponding loudness is 2 sone.

In the case of natural acoustic signals, which are always broadband, the loudness does not match the physically transmitted energy of the signal. The incoming acoustic signal is evaluated psychoacoustically in individual frequency bands, the so-called critical bands. The loudness results from a band-specific signal processing and a cross-band overlay of the band-specific processing results, known under the term "loudness summation". These basics were described in detail by E. Zwicker, "Psychoacoustics", Springer-Verlag Berlin, University text, 1982.

If the loudness is now considered to be one of the most important psycho-acoustic parameters determining acoustic perception, the present invention has as its object to propose a method and devices suitable for this, with which a hearing aid to be adapted to an individual can be adjusted in such a way that the acoustic perception of the individual corresponds at least in a first approximation to that of a norm, namely the person with normal hearing.

One possibility to record the individually perceived loudness on selected acoustic signals as a further usable variable is the one shown schematically in FIG. 1, for example from O. Heller, "Auditory field audiometry using the method of category division", Psychological Contributions 26, 1985, or V. Hohmann, "Dynamic Compression for Hearing Aids, Psychoacoustic Fundamentals and Algorithms", dissertation UNI Göttingen, VDI-Verlag, series 17, No. 93, known method. An individual I is presented with an acoustic signal A which can be adjusted on a generator 1 with regard to the spectral composition and transmitted sound pressure level S. The individual I evaluates or "categorizes" the acoustic signal A currently heard according to e.g. thirteen loudness levels or categories, as shown in FIG. 1, to which levels numerical weights, for example from 0 to 12, are assigned.

With this procedure it is possible to measure the perceived individual loudness, i.e. to be quantified, but only selectively with respect to given acoustic signals, so that such measurements do not initially make it possible to draw conclusions about the individually perceived loudness that is perceived with natural, broadband signals.

If in the following the loudness is primarily considered as the quantity influencing the psycho-acoustic perception, it is because this quantity decisively determines the psycho-acoustic perception of acoustic signals. However, as will be explained further below, the procedure according to the invention can also be used to take further psychoacoustic variables into account, such as for example the variable "masking behavior in the time domain and / or in the frequency domain".

2, the basic principle in the preferred procedure according to the invention described in more detail below is shown schematically for the time being.

The norm, N, is used to determine a psycho-acoustic perception variable, in particular the loudness L _N , by means of standardized acoustic signals A _o and compared with the values of this variable, corresponding to L _{I of} an individual, with the same acoustic signals A _o . From the difference corresponding to ΔL _NI , setting data are determined which act directly on a hearing aid or on the basis of which, manually, a hearing aid is set. L _I is determined on the individual without a hearing aid or with a hearing aid that has not yet been adapted, possibly progressively adapted.

However, the loudness itself is a variable that in turn depends on several variables. On the one hand, there is a large number of measurements that must be carried out on an individual in order to receive even approximate enough information, with the intervention on the hearing aid, for all broadband signals occurring in a natural environment to be able to make the desired perception correction. On the other hand, the correlation of detected size differences with interventions in the transmission behavior of a hearing aid is not clear and extremely complex.

In a preferred manner, a reduction in the measurements to be made on the individual is now sought, and a solution is thus sought which allows the necessary interventions to be concluded relatively easily from measurement results on the individual and their comparison with standard results.

Basically, a quantifying model of the Perception size, especially loudness, used. Such a model should be able to be entered with any kind of acoustic signals; at least approximate is the corresponding size. On the other hand, the model that is valid for the individual should be identifiable with relatively few measurements. The identification should be able to be terminated when the model has been identified to a certain extent.

Such a quantifying model of a psycho-acoustic perception quantity does not have to be given by a closed mathematical expression, but can be defined by a multidimensional table, from which the perceived quantity of perception can be called up with the prevailing frequency and sound level relationships of a real acoustic signal as a variable can.

Darin bezeichnen:

k:

Laufparameter mit 1 ≦ k ≦ k_o, Numerierung der Anzahl k_o berücksichtigter kritischer Bänder;

CB_k:

spektrale Breite des betrachteten kritischen Bandes mit der Nummer k;

α_k:

Anstieg einer linearen Approximation der in Kategorien skalierten Lautheitsempfindung bei logarithmischem Auftrag des Pegels eines präsentierten sinusförmigen oder schmalbandigen akustischen Signals, dessen Frequenz circa bandmittig des betrachteten kritischen Bandes CB_k liegt;

T_k:

Hörschwelle beim erwähnten Sinussignal;

S_k:

den mittleren Schalldruckpegel eines präsentierten akustischen Signals im betrachteten kritischen Frequenzband CB_k.

Although different mathematical models for loudness are conceivable, it was recognized according to the invention that the model based on Zwicker according to A. Leijon, "Hearing Aid Gain for Loudness-Density Normalization in Cochlear Hearing Losses with Impaired Frequency Resolution", Ear and Hearing, Vol. 12, No. 4, 1990, is particularly suitable for the goals pursued here. It is said:

Denote therein:

k:

Running parameters with 1 ≦ k ≦ k _o , numbering of the number k _o considered critical bands;

CB _k :

spectral width of the considered critical band with the number k;

α _k :

Increase in a linear approximation of the loudness perception scaled in categories when the level of a presented sinusoidal or narrow-band acoustic signal is logarithmically applied, the frequency of which lies approximately in the middle of the band of the critical band CB _k under consideration;

T _k :

Hearing threshold for the mentioned sine signal;

S _k :

the average sound pressure level of a presented acoustic signal in the critical frequency band CB _{k considered} .

As can be seen from this, the band-specific, mean sound pressure levels S _k form the model variables defining a presented acoustic signal, which determine the current spectral power density distribution. The spectral width of the considered critical bands CB _k , the linear approximation of the loudness perception, α _k , and the hearing threshold T _k are parameters of the model or the mathematical simulation function according to (1).

It has now been further recognized that on this model the parameters α _k , T _k , CB _k can be determined relatively easily on the one hand by means of relatively fewer acoustic tests on individuals and that these coefficients are also relatively easy to determine Transmission variables on a hearing aid are correlated and can thus be changed for an individual by manipulation of a hearing aid.

The model parameters α _k , T _k and CB _k have been determined using the standard N, ie for people with normal hearing.

The linear approximation of loudness in categories per increase in the mean sound pressure S _k in dB in the respective critical bands CB _{N of} the standard is given in the literature, for example in E. Zwicker, "Psychoacoustics" as the same for all bands.

In FIG. 3, the curve L _{kN represents} the loudness curve of the standard as a function of the sound level S _{k of} an acoustic signal presented in a respective critical band k, recorded as explained with reference to FIG. 1. A sinusoidal signal or a narrowband noise signal is presented. As can be seen from this, the parameter α _{N represents} the slope of a linear approximation or regression line of this course L _kN at higher sound levels, ie at sound pressure levels from 40 to 120 dB SPL, where the acoustic useful signals also predominantly occur. This is also referred to below as "large signal behavior". As mentioned, in the norm this increase can be assumed to be the same, α _N , in each of the frequency bands.

gesetzt wird.3 with a view to the mathematical model according to (1) also shows that disregarding the level dependence of the gradient steepness of L _kN , ie approximating this curve with a regression line, can only lead to a model of the first approximation. The model becomes more accurate if, in every critical band, depending on the sound pressure level, the parameter values are used, ie α _N = α _N (S _k ), ie if in each band k

is set.

In contrast to the parameter α _N , the hearing threshold T _kN also differs in the norm and in a first approximation in every critical frequency band CB _kN and is not a priori identical to the 0dB sound pressure level. The typical hearing threshold curve of the standard is precisely defined by ISO R226 (1961).

Furthermore, the bandwidths of the critical bands CB _kN for the standard and their number k _o in ANSI, American National Standard Institute, American National Standard Methods for the Calculation of the Articulation Index, Draft WG S. 3.79, May 1992, V2.1 , standardized.

In summary, the preferred mathematical loudness model according to (1) is known for the standard.

As can be easily seen, large deviations can occur between the perceived loudness of individuals and that of the statistically determined norm. In particular, in the case of individuals I deviating from the norm, in particular hearing _impaired persons, a specific coefficient α _kI can be determined for each critical frequency band; _Furthermore , there are deviations from the norm with regard to the hearing threshold T _kI and the width of the critical bands CB _kI .

Leijon has described a procedure that allows the further band-specific coefficients or model parameters α _kI and CB _kI to be estimated from the hearing thresholds T _kI of individuals. However, the estimation errors are usually large when considering individual cases. Nevertheless, when identifying individual loudness models, it is possible to start with estimated parameters, for example those estimated from diagnostic information. This drastically reduces the effort and the burden on the individual.

Metrological determination of the coefficients α _kI , CB _kI  and T _kI

As already indicated, the loudness L, recorded with a category scaling according to FIG. 1, is plotted in FIG. 3 as a function of the mean sound pressure level in dB-SPL for a sinusoidal or narrow-band signal of the frequency f _k in a critical band of the number k considered . As has been mentioned further, the loudness L _{N of} the standard increases non-linearly with the signal level in the selected representation, the gradient curve is in a first approximation for normal hearing people for all critical bands with the regression line with the gradient α _N entered on the curve N in FIG. 3 reproduced in [categories per dB-SPL].

From this representation it is readily apparent that the model parameter α _N corresponds to a nonlinear amplification, the same for normal hearing people in every critical band, but to be determined for individuals with α _kI in every frequency band. The straight line with the slope α _k approximates the non-linear loudness function in band k by a regression line.

In FIG. 3, L _kI typically denotes the course of the loudness L _{I of the} hearing _impaired in a band k.

As can be seen from the comparison of the curves L _kN and L _kI , the curve of a hearing _impaired person has a larger offset to the zero point and is steeper than the curve of the norm. The larger offset corresponds to an increased hearing threshold T _kI , the phenomenon of the fundamentally steeper loudness _curve is referred to as loudness recruitment and corresponds to an increased α parameter.

It is known to fundamentally determine hearing thresholds using classic threshold audiometry. However, it is also possible, also in the sense of threshold _audiometry, to detect the hearing thresholds T _kI on individuals with an arrangement according to FIG. 1 by threshold _detection between inaudible and audible. However, larger errors must be accepted around the threshold. In the following it is assumed that the respective hearing thresholds T _{kI have} already been recorded and known, simply by audiometry.

Referring to the remaining model parameters according to (1), the width of the respective critical bands CB _kI , it can be stated that the presence of several such bands only becomes effective when psycho-acoustic processing of broadband audio signals, i.e. broadband signals, the spectrum of which is at least two adjacent critical bands. In hearing impaired people, a widening of the critical bands is typically noticeable, whereby primarily the loudness summation is impaired even after (1).

Various measurement methods have been described for determining the bandwidth of the critical bands. In this regard, reference can be made to BR Glasberg & BCJ Moor, "Derivation of the auditory filter shapes from notched-noise data", Hearing Research, 47, 1990; P. Bonding et al., "Estimation of the Critical Bandwidth from Loudness Summation Data ", Scandinavian Audiolog, Vol. 7, No. 2, 1978; V. Hohmann, Dynamic Compression for Hearing Aids, Psychoacoustic Fundamentals and Algorithms", dissertation UNI Göttingen, VDI-Verlag, Row 17, No. 93. The measurement of the loudness summation with specific broadband signals according to the last-mentioned literature reference, both for normal and hard of hearing people, is well suited for experimental measurement of the respective bandwidths of the critical bands.

• die individuellen α_kI-Parameter sich aus den Regressionsgeraden gemäss Fig. 1 ermitteln lassen,
• die individuellen Hörschwellen T_kI sich durch Schwellenaudiometrie bestimmen lassen,
• die individuellen Bandbreiten CB_kI der kritischen Bänder sich, wie in obgenannter Literatur angegeben, bestimmen lassen, wobei
• diese Grössen für die Norm, d.h. für die Normalhörenden, bekannt und normiert sind.

It can thus be stated that:

• the individual α _kI parameters can be determined from the regression lines according to FIG. 1,
• the individual hearing thresholds T _kI can be determined by threshold _audiometry ,
• the individual bandwidths CB _{kI of} the critical bands can be determined as indicated in the above-mentioned literature, where
• these values are known and standardized for the norm, ie for the normal hearing.

However, the individual recording of the loudness or scaling _curves L _kI according to FIG. 3 for the subsequent determination of the model parameters α _kI and possibly T _kI and the known procedure for determining the width of the critical bands CB _{kI are} so time-consuming that they, except in the frame scientific investigations, an individual present to clarify his perceptual behavior can hardly be expected.

A preferred procedure is therefore to be explained with reference to FIG. 4.

It is based on the knowledge that when using standard acoustic narrow-band signals A _o , which are essentially centered in the critical frequency bands CB _N , the model parameters CB _kI, which are still unknown to the individual, can be equated to the known CB _kN without intolerable errors.

_Furthermore , it is assumed that the hearing thresholds T _{kI of} the individual I were determined in a different measurement environment by means of classic threshold audiometry, since an individual to be clarified with regard to hearing behavior is only subjected to such an examination in the vast majority of cases. It can be seen that are primarily call in the T _{kI kI} α and for identification of the individual loudness model, its individual parameterization.

4, individual I, as shown, for example, via headphones, electrically or by means of an electrical-acoustic transducer, is supplied with narrow-band norm-acoustic norm signals A _ok lying in the frequency bands CB _Nk . For example, via an input unit 5 according to FIG. 1, the individual I evaluates and quantifies the perceived loudness, L _S (A _ok ).

Corresponding to the channel or band membership of the signal A _ok , the associated standard bandwidth CB _kN and the parameter α _N are provided on the output side via a selection unit 7 from a standard storage unit 9. The electrical signal S _e (A _ok ) corresponding to the sound pressure level of the signal A _ok is fed together with the associated bandwidth CB _{kN to} a computing unit 11 which, according to the preferred one mathematical loudness model according to (1), a loudness value L '(A _ok ) is calculated, namely from S _e , CB _kN , α _N and, as mentioned previously, the hearing threshold _value T _kI stored in a storage unit 13 and _{predetermined} .

5, it is to be shown which loudness L 'the computing unit 11 calculates on the basis of these predetermined parameters. Based on the use of the hearing threshold T _{kI of} the individual and the parameter α _{N of} the standard, a loudness value L 'is determined on the computing unit 11 at the given sound level, corresponding to S _{e of} the signal A _ok , as it corresponds to a scaling function N', which is determined by the Regression line with α _N and the hearing threshold T _{kI is defined} in a first approximation.

4, on the output side of the computing unit 11, this loudness value L 'is compared at a comparison unit 15 with the loudness value L _I by the input unit 5. The difference .DELTA. (L ', L _I ) appearing on the output side of the comparison unit 15 acts on an incrementing unit 17. The output of the incrementing unit 17 is superimposed on a superposition unit 19 with the α _N parameter supplied to the computing unit 11 by the storage unit 9 with the correct sign. The incrementing unit 17 thus increments the signal corresponding to α _N by increments Δα according to the number of increments n until the difference appearing on the output side of the comparison unit 15 reaches or falls below a predeterminable minimum dimension.

5, this means that α _N on the curve N 'is changed until the loudness value L' calculated on the unit 11 matches the loudness value L _I to the extent required. Based on the course N ', the computing unit 11 thus has the regression line of the individual scaling curve I found.

an den Ausgang der Messanordnung ausgegeben; es gilt:

${\text{α' = α}}_{\text{kI}}\text{.}$

Damit ist mit geforderter Genauigkeit entsprechend Δr im betrachteten kritischen Frequenzband k der Parameter α_kI des Indivuums gefunden.The output signal of the comparison unit 15 in FIG. 4 is compared on a comparator unit 21 with an adjustable signal Δr in accordance with a predeterminable, maximum error - as an abort criterion. When the difference signal Δ (L ', L _I ) on the output side of the comparison unit 15 reaches the value Δr, as shown schematically, opening the switch Q₁ and closing the switch Q₂ on the one hand aborts the incrementation of α, and on the other hand the α- Value accordingly

${\text{α '= α}}_{\text{N}}\text{+ nΔα}$

output to the output of the measuring arrangement; the following applies:

${\text{α '= α}}_{\text{kI}}\text{.}$

The parameter α _{kI of} the individual is thus found with the required accuracy corresponding to Δr in the critical frequency band k considered.

By defining the termination criterion Δr in such a way that the α _{kI identification} meets practical accuracy requirements, the process is optimally short or only as long as necessary.

In Fig. 6a, analogous to Fig. 5, the scaling function N of the norm and I of a hearing impaired individual is shown again. At a given sound pressure level S _kx , an amplification G _x must therefore be provided on the hearing device so that the individual perceives the loudness L _x with the hearing device as the norm N. 6a shows, depending on various, for example, entered sound pressure levels S _kx , several gain values G _{x to be} provided on the hearing aid are entered.

FIG. 6b shows the gain curve resulting from the considerations of FIG. 6a as a function of S _k , as can be realized on a transmission channel on the hearing aid corresponding to the critical frequency band k, as shown in FIG. 6c. The non-linear gain curve G _k (S _k ) shown in FIG. 6b is determined heuristically and schematically from the parameters T _kI and α _kI or the differences T _kN -T _kI and nΔα as determined with reference to FIGS. 4 and 5.

The described procedure is optimally repeated in every critical frequency band k. For each critical frequency band and when approximating with a regression line, only one norm-acoustic signal has to be presented to the individual; if necessary, others can be used to check the regression lines found.

From the considerations, in particular with respect to FIGS. 4 to 6, it is now readily apparent that the proposed method can be expanded to an arbitrarily precise approximation by simple extension. An increase in the accuracy achieved with a hearing aid, with which an individual has the same loudness perception as the norm, can be achieved with reference to FIG. 5 by basically approximating the scaling curves piece by piece by means of several regression lines in the sense of a regression polygon.

The procedure described with reference to FIGS. 4 to 6 is essentially based on the respective individual or standard scaling curve N or I as a first approximation only through a pair of regression lines, namely for low sound pressure levels and for high sound pressure levels, approximate.

This also corresponds to the approximation, with which the simulation model according to (1) takes into account the respective scaling curves in the critical frequency bands.

The model according to (1) which is preferably used becomes arbitrarily more precise (1 *) by using α _k (S _k ) instead of the level-independent parameters α _k . In (1), α _{k is} replaced by α _k (S _k ).

This procedure, which is expanded on the basis of the explanations relating to FIGS. 4 to 6, will be explained with reference to FIGS. 7 and 8.

In FIG. 7, the function blocks that act analogously to the function blocks of FIG. 4 are provided with the same position numbers.

8 shows the scaling curve N of the norm and of an individual I in analogy to FIG. 5. In contrast to the approximation of FIG. 5, the scaling curve N is sound pressure level-dependent slope parameter α _N (S _k) is approximated, ie by a polygon of support values S _k of the curve N. This sound pressure level dependent parameter α _N (S _k) are assumed to be known by they can be easily determined from the known scaling curves N of the standard at the given support values S _kx .

In analogy to the considerations of FIG. 5, the arrangement according to FIG. 7, taking into account the individual hearing threshold T _kI for the time being, assumes that the curve N 'offset by the individual hearing threshold _value T _kI , on which the sound pressure level-dependent values continue to be formed Standard parameters α _N (S _k ) apply. The latter are changed until the curve N 'conforms to the scaling curve I of the individual with the required accuracy. At least as many level values S _{kx are} to be evaluated on the individual as the desired number of approximation tangents used for the approximation indicates.

From the respective necessary changes in the parameters α _N (S _k ), which are now dependent on the sound pressure level, the more precise course of the sound pressure level-dependent amplifications to be set on the hearing device on a channel-specific basis is determined with reference to FIG. 6b.

7, in addition to the bandwidths of the critical frequency bands CB _kN , a set of sound pressure level-dependent slope parameters α _N (S _k ) is stored in the memory unit 9. The individual I is again presented with normacoustic, narrow-band signals lying in the respective critical bands, but, in contrast to the procedure according to FIG. 4, per critical frequency band at different sound pressure levels S _kx .

The individual loudness evaluations for these standard acoustic signals of different sound pressure levels are preferably stored in a buffer unit 6. With reference to FIG. 8, these stored loudness perception values hold the scaling curve I of the individual with reference values.

The storage unit 9 supplies the bandwidth CB _{kN associated} with the critical frequency band under consideration and the set of α parameters dependent on sound pressure level to the computing unit 11, in addition to the previously determined, individual, band-specific hearing threshold T _kI .

As already explained with reference to FIG. 4 and shown here only in a simplified manner, the frequency of the norm acoustic signal determines the critical frequency band k under consideration, and the values relevant for this are retrieved from the memory unit 9 accordingly. The sequence F of the following sound pressure level _values S _kx is preferably further stored in a memory device 10. As soon as the individual loudness perception values are recorded and stored in the storage unit 6, the sequence of the stored sound pressure level _values S _kx is also _fed from the storage unit 10 to the computing unit 11, with which the latter, according to FIG. 8, calculates the scaling _curve N 'from the hearing threshold _value T _kI , the bandwidth CB _kN and the sound pressure level-dependent steepness values α _N (S _kx ), and thus determined which loudness values would be expected according to curve N 'in FIG. 8 at the sound pressure levels S _{kx used} .

8, all sound pressure level-dependent difference values Δ are now determined on the comparison unit 15 and, if necessary, different incremental adjustment of the sound pressure level-dependent standard parameters α _N (S _kx ) by the incrementing unit 17 and on the superimposition unit 19, as indicated by Δ ' α is shown, the sound pressure level-dependent coefficients are changed and thus the course of the calculated curve N 'until a sufficient approximation of curve N' to curve I is achieved.

To this end, the difference appearing on the output side of the comparison unit 15, here in the sense of a difference in sound pressure level dependent between the curves S and the changed curve N 'according to FIG. 8, is assessed with regard to falling below a predetermined maximum range - as a termination criterion - and as soon as the deviations mentioned indicate a TARGET 4, on the one hand the optimization or incrementation process is interrupted; on the other hand, the sound pressure level-dependent α parameters pending at the computing unit 11 are output which correspond to the tangent slope values on the individual scaling _curve I, _i.e. α _kI (S _kx ) or the Δ'α _kI (S _kx ).

From these sound pressure level-dependent values, in analogy to FIGS. 6b and 6c, the nonlinear amplification function assigned to the specific critical frequency band on the hearing aid is determined and adjusted there.

It was shown how the required, non-linear amplification of the hearing aid transmission in a channel, which corresponds to the critical frequency band under consideration, is determined with any accuracy and used to set this channel.

In a first approximation, it was assumed that the width of the respective critical frequency band is irrelevant for the individual perception of a narrowband signal, which, however, is only approximate, as can be seen from (1).

However, the width of the critical bands CB _k becomes relevant for the loudness perception of the individual if the presented normacoustic signals have spectra that lie in two or more critical frequency bands, because loudness summation according to (1) or (1 *) then occurs .

So far it has been found that deviations of the band-specific parameters α and T of an individual from the norm can be compensated for by setting the non-linear level-dependent amplification on the channels of a hearing aid assigned to the critical frequency bands. As mentioned, the width of the critical frequency bands differs individually, in particular for hearing impaired people, from that of the norm, the critical frequency bands of hearing impaired are usually wider than the corresponding of the norm.

A simple measurement method for the position or the limits of the critical frequency bands is described by P. Bonding et al., "Estimation of the Critical Bandwidth from Loudness Summation Data", Scandinavian Audiolog, Vol. 7, No. 2, 1978. For this purpose, the range of presented standard acoustic test signals is continuously increased, and an individual scales, as described, the perceived loudness. The mean sound pressure level is kept constant. Where the individual perceives a noticeable increase in loudness, the boundary lies between two critical frequency bands, because loudness summation then occurs.

It is therefore essential to determine the width of the critical frequency bands CB _{k I} for individual loudness perception correction based on broadband acoustic signals, ie when loudness summation occurs. From the knowledge of the frequency band limits deviating from the norm, the nonlinear amplification G of FIG. 6b in the respective hearing aid channels assigned to the critical bands is changed, depending on the frequency, in particular in frequency ranges which do not belong to the same critical band on the individual as in the norm are.

This is to be explained in a simplified and heuristic manner with reference to FIGS. 9a and 9b.

In FIG. 9a, frequency bands CB _k and CB _{k + 1} , for example critical frequencies for the standard N, are drawn in over the frequency axis f. Below, in the same representation for an individual I, are the correspondingly broadened, corresponding bands registered.

The nonlinear reinforcements found so far have been determined channel-specifically or band-specifically with reference to the critical bandwidths of the standard. Taking into account the critical bandwidths of the individual, it can be seen from FIG. 9a that, for example, the hatched area Δf in the individual falls within the broadened critical band k, while in the norm it falls within the band k + 1. However, this means that, with the previous reference to the critical bandwidths of the standard, signals e.g. in the hatched frequency range .DELTA.f, the gain must be corrected on the individual.

If, according to FIG. 9b, signals that are transmitted on a hearing aid channel that corresponds to the critical frequency band k of the standard are amplified with the nonlinear level-dependent gain function G _k (S _k ) previously explained with reference to FIG. 6b, then Signals in the superimposition range Δf, that is to say frequency-dependent, are additionally raised or possibly reduced.

Knowing the channel-specific, non-linear level-dependent gains G _k (S _k ) in the respective critical frequency bands determined as shown and knowledge of the deviations of the critical frequency bands CB _{kI of} the individual from those CB _{kN of} the standard, it is possible to make these deviations frequency-dependent through the gains G _k (S _k , f) to compensate for the hearing aid channels.

Of course, it is easily possible to experimentally determine all parameters α, T and CB defining the model according to (1) for the norm and for an individual and to deduce correctional interventions on the hearing aid directly from deviations of these coefficients. However, conditionally such a procedure is the channel-specific measurement of the individual, which, as has been mentioned, is hardly possible for clinical applications.

Starting from the procedure according to FIGS. 4 and 7, FIG. 10 shows a further development as a function block signal flow diagram in which the parameters α _k and CB _k can be determined using a single method. Not only is one critical band after the other examined in accordance with FIGS. 4 and 7, but also, with broadband acoustic signals, the loudness summation is recorded and the width of the individual critical bands is thus also determined as a variable by optimization.

The simulation model parameters of the standard, namely α _N , CB _kN , are stored in a memory unit 41 and, in a preferred embodiment, not the hearing thresholds T _{kN of} the standard, but rather the hearing thresholds T _{kI of} the individual to be examined, determined beforehand by audiometry and taken from a memory unit 43.

An individual is acoustically presented with signals A _Δk by a generator that is no longer shown here. The electrical signals corresponding to them in FIG. 10, also designated A _Δk , are fed to a frequency-selective power measurement unit 45. The channel-specific average powers are determined on the unit 45 in accordance with the critical frequency bands of the standard, frequency-selective, and a set of such power values S _{Δk is} output on the output side. These signals are stored in a memory unit 47 in a channel-specific manner and specifically for the signal A _Δk (A No.) that is presented in each case. When one of the signals A _Δk is presented in each case, all the coefficients stored in the storage unit 41 become initially unchanged, supplied via a unit 49 to be described later to the computing unit 51, an arithmetic module 53, as are the predominant signal A _.DELTA.k corresponding power signals S _.DELTA.k. The computing module 53 calculates the loudness L 'according to (1) from the norm parameters α _N , CB _kN and the individual hearing threshold _values T _kI , taking into account the loudness _summation , which would result for the norm if the latter had hearing thresholds (T _kI ) such as the individual.

For each signal A _{Δk presented} , assigned to the signal, the calculated value L ' _{N is} stored in a storage unit 55 on the output side of the computing module 53. Each of the presented broadband (Δk) signals A _Δk , as described with reference to FIGS. 4 and 7, is assessed or categorized by the individual in terms of loudness perception, the evaluation signal L _I , again assigned to the respective presented acoustic signals A _Δk , stored in a storage unit 57. Both when determining L ' _N and when determining L _I , the loudness _{summation is} taken into account arithmetically or by the individual due to the broadbandness Δk of the signals A _Δk presented.

After the presentation of a given number of signals A _Δk , the corresponding number of values L ' _{N is} stored in the memory unit 55, and the corresponding number of L _I values are also stored in the memory unit 57.

Now the presentation of acoustic signals is stopped, the individual is no longer burdened. All assigned L ' _N and L _I values, each of which forms a course as a function of the numbers of the previously presented acoustic signals A _Δk , are fed to a comparison unit 59 on the computing device 51 which shows the course of the difference Δ (L ' _N , L _I ) determined. This difference curve is fed to the parameter modification unit 49, in principle similar to the control difference signal in a follow-up control circuit.

The parameter modification unit 49 varies the start values α _N , CB _kN , but not the T _kI values, for all critical frequency bands, while simultaneously recalculating the updated L ' _N value until the difference signal Δ (L' _N , L _I ) runs within a predeterminable minimum course, which is checked on the unit 61.

If the termination criterion ΔR is not yet reached, further acoustic signals A _Δk must be processed.

Thus, on the simulation model according to (1) with the individual hearing thresholds T _kI, the standard parameters α _N and CB _kN entered as start values, taking into account the signals S _Δk corresponding to the channel-specific sound pressure values retrieved from memory 47, are varied according to predetermined search algorithms until a maximum permissible deviation between the L ' _N and the L _I course has been reached.

If the achievement of a predetermined maximum deviation criterion ΔR by the difference Δ (L ' _N , L _I ) occurring on the output side of the unit 59 is registered on a comparator unit 61, the search process is terminated; The α and CB values on the output side of the modification unit 49 correspond to those which, used in (1), result in loudness values corresponding optimally with the individually perceived values L _I for the acoustic signals A _Δk presented: by varying the standard parameters, the individual values in turn became individual determined.

From the output values of the modification unit 49 when the search is terminated and their difference from the start values α _N and CB _kN , manipulated variables are determined in order to set the amplification functions on the frequency-selective channels of the hearing aid corresponding to the critical frequency bands.

As can be seen, the procedure described is actually the search for a minimum point of a multi-variable function. In most cases, several sets of changed parameters will result in the minimum criterion specified by ΔR being met. The method described can therefore lead to the receipt of several such solution parameter sets, with the physical positions of the hearing aid being used in those sets which are physically sensible and, for example, the easiest to implement.

Solution parameter sets that can be excluded from the outset, which would lead, for example, to amplification profiles on the respective channels of the hearing aid that are extremely difficult or impossible to implement, can be excluded from the outset by corresponding specifications on the modification unit 49.

A shortening of the search process can also be achieved, for example for hearing-impaired individuals, by replacing the standard parameters α _N and CB _{kN with} the α _kI and CB _kI values estimated from the individual hearing thresholds T _kI for hearing _{impaired people} as search _starting values in the Storage unit 41 are stored, especially if the individual's hearing loss is determined from the outset.

Of course, the arithmetic unit 51 can also do the mentioned Include storage devices integrated in terms of hardware; their delimitation shown in dashed lines in FIG. 10 is to be understood, for example, including in particular the computing module 53 and the coefficient modification unit 49.

The previously described procedure according to FIGS. 4, 7 and 10 are primarily suitable for setting a hearing aid ex situ. The determined manipulated variables can be transferred directly electronically to a hearing device in situ, but the actual advantage of an in situ adaptation, namely the consideration of the fundamental hearing influence by a hearing device, is not taken into account: first, all the manipulated variables are determined without a hearing device, and then is set without further acoustic signal presentation.

If, however, the fundamental considerations in connection with FIGS. 4, 7 and 10 are rethought, it can be seen that the considerations made in connection in particular with the ex situ setting of a hearing aid readily refer to the “on-line” setting have a hearing aid transmitted in situ. Instead of a predefined loudness model being adapted to that of an individual or possibly vice versa according to the simulation model with predefined parameters and finally determining variables for the hearing aid, it is easily possible to adjust the hearing aid in situ for as long as until the loudness perceived by the individual matches the norm.

It is entirely possible to use the assessment of loudness perception by the individual to determine whether an incremental parameter change made on the hearing aid, in analogy to FIGS. 4 and 7, changes the Loudness perception against or away from the loudness of the norm. However, it should be avoided that an individual is unreasonably stressed in terms of time and concentration due to the hearing aid fitting.

With regard to the procedure explained with reference to FIG. 10, it can now be seen that this is optimally suited for in-situ hearing aid adaptation. The preferred procedure for this is to be explained with reference to FIG. 11, in which function blocks which correspond to those of FIG. 10 are provided with the same reference symbols. The procedure corresponds, with the differences described below, to that explained with reference to FIG. 10.

The acoustic signals A _Δk are fed to the hearing aid system HG with transducers 63 and 65 on the input and output sides and individual I, the latter loading the perceived L _I values into the memory 57 with the evaluation unit 5.

Exactly the same as was explained with reference to FIG. 10, the L _I value is stored in the memory 57 for each presented standard-acoustic, broadband signal A _Δk . With the power _values S _Δk from the storage unit 47 according to FIG. 10 and the standard parameter values from the storage unit 41, the loudness values L ' _{N are} initially determined on the computing module 53 according to (1) or (1 *), as was explained with reference to FIG. 10 , calculated and, specifically assigned to the presented signals A _Δk , stored in the memory unit 55. Via the comparison unit 59 and the modification unit 49, the standard parameters from the memory unit 41 are then modified, as described, until they, when used in (1) or (1 *), give L ' _N values with predeterminable accuracy corresponding to the L _I values in memory 57.

und

${\text{L'}}_{\text{N}}{\text{= L}}_{\text{I}}{\text{für alle A}}_{\text{Δk}}\text{.}$

Damit gilt aber auch:

${\text{α'}}_{\text{Nk}}{\text{= α}}_{\text{Ik}}{\text{, CB'}}_{\text{Nk}}{\text{= CB}}_{\text{Ik}}\text{.}$

The following then applies:

${\text{α '}}_{\text{Nk}}{\text{= α}}_{\text{N}}{\text{± Δ'α}}_{\text{k}}{\text{'CB'}}_{\text{Nk}}{\text{= CB}}_{\text{Nk}}{\text{± Δ'CB}}_{\text{k}}\text{,}$

and

${\text{L '}}_{\text{N}}{\text{= L}}_{\text{I.}}{\text{for all A.}}_{\text{Δk}}\text{.}$

But this also applies:

${\text{α '}}_{\text{Nk}}{\text{= α}}_{\text{Ik}}{\text{'CB'}}_{\text{Nk}}{\text{= CB}}_{\text{Ik}}\text{.}$

überträgt, wobei ${\text{ΔT}}_{\text{k}}{\text{= T}}_{\text{kI}}{\text{-T}}_{\text{kN}}$

gesetzt ist, das Gesamtsystem aus Hörgerät und Individuum eine Lautheit entsprechend der Norm wahrnimmt.However, this also means that when the hearing aid receives input signals with a corrective loudness ${\text{L}}_{\text{Cor}}{\text{= L}}_{\text{Cor}}{\text{(± Δα}}_{\text{k}}{\text{, ± ΔCB}}_{\text{k}}{\text{, ΔT}}_{\text{k}}\text{)}$

transmits, whereby ${\text{ΔT}}_{\text{k}}{\text{= T}}_{\text{kI}}{\text{-T}}_{\text{kN}}$

is set, the overall system of hearing aid and individual perceives a loudness according to the norm.

As has already been explained in principle with reference to FIG. 6c, the hearing aid HG has a number k _o frequency-selective transmission channels K between the converter 63 and converter 65. Actuators for the transmission behavior of the channels are connected to an actuating unit 70 via a corresponding interface. The latter are fed the initial manipulated variables SG _o previously determined as optimal.

Now that the changed parameters α ' _Nk , CB' _{Nk have been} determined for a predetermined number of presented normacoustic, broadband signals A _Δk by means of the computing module 53 and the modification unit 49, by means of which, according to FIG. 8, the Scaling curves N 'have been adapted to those of the individual I with a hearing aid HG that has not yet been adjusted, the parameter changes found will act ± Δα _k , ± ΔCB _k , ± ΔT _k or Parameters α _N , T _kN , CB _kN and α _kI , T _kI , CB _kI via the manipulated variable control unit 70 so as to control the hearing aid in such a way that its channel-specific frequency and amplitude transmission behavior for the signals A _Δk , on the output side, produce the correction loudness L _Kor .

While in the procedure according to FIG. 10 and with a view of FIG. 8, the parameters of the standard were changed until the scaling curves N 'match the scaling curves I and the hearing thresholds T _{kN were} not required for this, but only for the determination of the amplifications 6b, the hearing thresholds of the individual, stored in memory 43, and the standard hearing thresholds, stored in memory 44, are also used according to FIG. 11.

The control variable determination unit 70 according to FIG. 11, manipulated variable changes, is summarized from the parameter changes determined in FIG. 11 in analogy to the procedure according to FIG. 10, in order to convert N 'into I according to FIG. 8, and from the differences in the hearing thresholds ΔSG for the channel-specific frequency and amplitude transmission behavior of the hearing aid in such a way that the scaling curves of the individual I are brought with the hearing aid HG with the desired accuracy to the scaling curves N of the standard:

The loudness behavior of the hearing aid forms the intrinsic, i.e. The individual's "own" loudness perception depends on that of the norm, the loudness perception of the individual with a hearing aid becomes equal to that of the norm or can be specified in relation to that of the norm.

Compared to an "ex situ" setting of the transmission behavior of a hearing aid, the "in situ" setting shown, for example, with reference to FIG. 11 shows the essential The advantage is that the physical “in situ” transmission behavior of the hearing aid and, for example, the mechanical ear influencing by the hearing aid are also taken into account.

12a) and b) show two basic implementation variants of a hearing aid according to the invention, by means of simplified signal flow function block diagrams which can be set "ex situ", but preferably "in situ" as described.

The hearing aid, as shown in FIGS. 12 a) and b), should, when optimally set, transmit received acoustic signals with the correction loudness L _Kor to its output, so that the system hearing aid and individual has a perception that is equal to that of the standard or (ΔL in Fig. 12a) deviates from this by a predeterminable amount.

According to FIG. 12 a), channels 1 to k _{o are} provided on a hearing aid according to the invention, followed by an acoustic-electrical input converter 63, each assigned to a critical frequency band CB _kN . The entirety of these transmission channels forms the signal transmission unit of the hearing aid.

The frequency selectivity for channels 1 to k _o is implemented by filter 64. Each channel also has a signal processing unit 66, for example with multipliers or programmable amplifiers. The non-linear, band- or channel-specific amplifications described above are implemented on the units 66.

On the output side, all signal processing units 66 act on a summation unit 68, which in turn acts on the output side on the electrical-acoustic output transducer 65 of the hearing aid works. Until then, the two design variants according to FIGS. 12a) and 12b) match.

In the embodiment variant according to FIG. 12a), the principle of which is hereinafter referred to as the “correction model”, the converted acoustic input signals present on the output side of the converter 63 are converted into their frequency spectrum at a unit 64a. This creates the basis for subjecting the acoustic signals, in the frequency domain, to the loudness model according to (1) or (1 *) on a computing unit 53 ', parameterized with the correction parameters Δα _k , ΔCB _k , ΔT _k found as described above, that is according to the correction loudness L _KOR . The aforementioned channel-specific correction parameters and the corresponding correction loudness L _{KOR are converted} into actuating signals SG₆₆ on the computing unit 53 ', with which the units 66 are set.

The values .DELTA.SG supplied to the hearing aid according to FIG. 12a) according to FIG. 11 therefore essentially correspond to the channel-specific correction parameters in this embodiment variant. By controlling the transmission behavior of the hearing aid via the units 66, as a function of the acoustic input signals currently pending and the correspondingly valid correction parameters, it is achieved that the hearing aid transmits the input signals mentioned with the correction loudness L _KOR . The system individual with hearing aid thus perceives the required loudness, be it preferably the same as the standard or in this respect in a predetermined ratio.

In the embodiment variant according to FIG. 12b), which is called the “difference model” variant in the following, the spectra are formed from the converted acoustic input signals and the electrical output signals of the hearing aid at units 64a. On a computing unit 53a are due of the input spectra as well as the loudness model parameters of the norm N calculates the current loudness values which the norm would perceive on the basis of the input signals. Analogously, the loudness values that the individual perceives without the hearing aid, ie the intrinsic individual, are calculated on a computing unit 53b on the basis of the output signal spectra. To this end, the modeling parameters 53b are fed with the modeling parameters of the individual, which, as described above, were determined.

On the one hand, a controller 116 compares the loudness values L _N and L _I determined by standard and individual modeling and, channel-specifically, the parameters of the standard model and the individual model and, on the output side, sends control signals SG₆₆ to the transmission units 66 in accordance with the determined differences in such a way that the modeled Loudness L _I becomes equal to the currently required standard loudness L _N.

In contrast to the correction model variant of FIG. 12a), controller 116 first determines the necessary correction loudness L _{KOR in} accordance with FIG. 12b).

In the case of the differential model variant according to FIG. 12b), the hearing aid transmission is set with the units 66 in such a way that the acoustic signals currently pending are transmitted with the correction loudness, so that the loudness is modeled on the output signals in accordance with the perception behavior of the individual ( 53b), results in a loudness corresponding to that which is perceived by the standard or which can be specified in this respect.

• dass, wie anhand der Fig. 1 bis 11 erläutert, ausgehend von einem gegebenen mathematischen Norm-Lautheitsmodell, Parameteränderungen ermittelt werden, welche dem Lautheits-Empfindungsunterschied von Norm und Individuum entsprechen. Damit sind Modellunterschiede und Individuummodell bekannt.
• An einem Hörgerät wird dasselbe mathematische Modell vorgesehen.
• Das Lautheitsmodell am Hörgerät wird in Funktion der Parameterunterschiede (Δ) betrieben, welche das Lautheitsmodell des Individuums demjenigen der Norm angleichen, wozu die gefundenen Modell-Parameterunterschiede und/oder die Norm-Parameter und die Individuum-Parameter dem Hörgerät zugespiesen werden.
• Am Hörgerätemodell wird im letzterwähnten Fall laufend überprüft, ob die aus den momentanen Eingangssignalen nach dem Modell der Norm berechnete Lautheit auch der durch das Individuum-Modell aufgrund der Ausgangssignale errechneten entspricht. Aufgrund der Modell-Parameterunterschiede und gegebenenfalls der modellierten Lautheitsunterschiede wird die Uebertragung am Hörgerät in regelndem Sinne so geführt, dass modellierte Lautheiten L_I, L_N in vorgebbare Relation kommen, vorzugsweise gleich werden.

In summary, it can therefore be stated:

• that, as explained with reference to FIGS. 1 to 11, starting out From a given mathematical norm-loudness model, parameter changes are determined which correspond to the loudness-sensation difference between norm and individual. Model differences and individual models are thus known.
• The same mathematical model is provided on a hearing aid.
• The loudness model on the hearing aid is operated as a function of the parameter differences (Δ) which match the loudness model of the individual to that of the standard, for which purpose the model parameter differences found and / or the standard parameters and the individual parameters are fed to the hearing aid.
• In the latter case, the hearing aid model continuously checks whether the loudness calculated from the current input signals according to the standard model also corresponds to the loudness calculated by the individual model based on the output signals. On the basis of the model parameter differences and, if applicable, the modeled loudness differences, the transmission on the hearing aid is carried out in a regulating sense in such a way that modeled loudnesses L _I , L _{N come} into a predefinable relation, preferably become the same.

In retrospect, for example in FIG. 10 or 11, it is readily apparent that the functions of the "ex situ" processing units described there, in particular the computing units 53, the modification units 49 and 70, are performed directly by the controller unit 71 on the hearing aid can. The combination of the procedure according to FIG. 11 with a hearing device according to FIG. which both compute the same loudness model, time sequentially with different parameters.

An embodiment of a hearing aid according to the invention, combined from the procedure according to FIG. 11 and the structure according to FIG. 12a), is shown in FIG. 13. The same position signs as in FIGS. 11 and 12 are used for the same function blocks. For reasons of clarity, only one channel X of the hearing aid is shown. At the beginning, a switchover unit 81 connects the storage unit (41, 43, 44) according to FIG. 11, shown here as a unit, to the unit 49. A switchover unit 80 is in the position shown, i.e. is open, a switchover unit 84 is also initially effective in the position shown.

In these switching positions, the arrangement works exactly as shown in FIG. 11 and explained in this context. After the adjustment method explained with reference to FIG. 11 has been run through, the determined parameter changes Δα _k , ΔCB _k , ΔT _k , which convert the individual loudness model (I) into the standard loudness model (N), when the hearing aid is put into operation by switching over the switching unit 80 in the storage unit 41 ', 43', 44 'acting in the same way as the storage unit 41, 43, 44 is loaded. The switching unit 81 is switched to the output of the last-mentioned storage unit. At the same time, the modification unit 49 is deactivated (DIS), so that it directly supplies the data from the storage unit 41 'to 44' unmodified and permanently to the computing unit 53c.

The switchover unit 84 is switched over so that the output on the arithmetic unit 53c, now acting as arithmetic unit 53 'according to FIG. 12a), via the manipulated variable control unit 70a onto the transmission path with the units 66 of the Hearing aid works. The ΔZ _k parameters Δα _k , ΔCB _k , ΔT _k , as shown in dashed lines, act together with L _KOR on the manipulated variable control unit 70a.

In this way, the loudness model arithmetic unit 53c integrated in the hearing aid is initially used to determine the model parameter changes Δα _k , ΔCB _k , ΔT _k required for correction and then, in operation, to guide the transmission manipulated variables of the hearing aid in a time-variable manner - in accordance with the current acoustic signals Relationships - used.

Sound optimization

The determination of the correction loudness model parameters on the hearing aid and thus the necessary manipulated variables for generally non-linear channel-specific amplifications, e.g. for a hearing impaired person, allows different target functions, or the loudness requirements as a target function, as mentioned, can be achieved with different sets of correction loudness model parameters and therefore manipulated variables ΔSG₆₆.

In general, one tries to to rehabilitate the hearing impaired so that he feels like the norm again. This goal was achieved according to the previous explanations regarding loudness. The goal, namely that the individual perceives the same loudness perception as the norm with the hearing aid, does not necessarily have to be the optimum of the individual hearing needs, in particular the sound type.

One must assume that individual deviations from the stated goal, ie to adjust the loudness to the isophones on average normal hearing, in practice will be perceived as optimal, if one even wants to consider a fine adjustment that takes this into account, namely optimization of the hearing aid parameters also for optimal acoustic sound perception.

Experience has shown that so-called sound parameters are mainly associated with the frequency response of the hearing aid. In the range of high, medium and low frequencies, the gain should therefore sometimes be raised and / or lowered in order to influence the well-being of the device, as is common with hi-fi systems.

However, if the gain is increased on a hearing aid that is optimally set with respect to the isophones of the standard as described so far, that is to say in certain transmission channels, the correction loudness changes.

This creates the further task of changing the correction parameter set used for a loudness-optimized hearing aid in such a way that on the one hand the perception of sound is changed, and on the other hand the previously achieved goal, namely individual perception of loudness with the hearing aid as the norm, is retained.

Due to the multi-parameter optimization task, which leads to the fulfillment of the loudness requirement, as previously mentioned, several parameter sets can lead to the solution, i.e. it is entirely possible to specifically change parameters of the correction loudness model and to ensure that the loudness requirement is maintained by changing other model parameters accordingly.

This will be explained with reference to FIG. 14, starting from FIG. 11 become.

Fig. 14 shows the measures to be taken in addition to the precautions of Fig. 11; the same function blocks, which have already been listed in FIG. 11 and thus explained, have the same item numbers.

It goes without saying that the following explanations also apply to a system according to FIG. 13 and to the positioning of the hearing aids according to FIGS. 12a), b). For reasons of clarity, however, the measures to be taken are shown starting from FIG. 11.

With regard to sound perception, there are evaluation criteria such as those described by Nielsen, namely sharp, shrill, dull, clear, reverberant, to name just a few.

In analogy to the quantification of loudness perception or to loudness scaling, as was explained with reference to FIG. 1, a sound sensation structured according to specific categories can also be numerically scaled, for example according to the criteria known from Nielsen. 14 and 11, after hearing device HG has been set by finding a correction parameter set (Δα _k , ΔCB _k , ΔT _k ) such that the individual with the hearing device has at least approximately the same loudness perception as the norm, the individual states: for example, in the case of the same broadband norm-acoustic signals A _{Δk presented} , on a sound scaling unit 90. A numerical value is assigned to each sound category on the unit 90. On a differential unit 92, the individually quantified sound sensation KL _I with the, for example, statistically determined sound sensation KL _N becomes the norm for the same acoustic Signals A _Δk compared. These are stored in a memory unit 94 so that they can be called up.

Now, however, conclusions can be drawn directly from the individual's sound sensation regarding the spectral composition of the signal he perceives. For example, if the sound sensation of the individual with the loudness-matched hearing aid is too high, it can be readily seen that the gain on at least one of the hearing-frequency channels of the hearing aid HG has to be reduced. The resulting change in loudness must, however, be reversed by intervening in channels involved in the formation of loudness, namely with corresponding amplification changes, in order not to continue to reveal the previously achieved goal. Thus, if the sound sensation of the individual differs from that of the standard with a hearing aid that is matched to loudness, then a sound characterization unit 96, for example between the comparison unit 59 and the parameter modification or incrementation unit 49, is activated according to FIG. 14, which limits the degree of freedom of the parameter modification on the unit 49 , ie changes one or more of the parameters mentioned, regardless of the minimal difference obtained at unit 59, and keeps them constant.

Now the error criterion ΔR, which is no longer shown in FIGS. 11 and 14, has to be fulfilled recently as an abort criterion according to FIG. 10; if the mentioned parameter is kept, the still free parameters are changed via unit 59 until loudness is again felt according to the standard - L _I = L ' _N - but now with a changed sound.

The sound characterization unit 96 is preferably connected to an expert database, shown schematically in FIG. 14 at 98, which contains the information regarding individual Sound sensation deviation from the norm is supplied. Information, for example, is stored in the expert database 98
"shrill at A _Δk is the result of too much amplification in channels no. ..."

If "perceived as shrill, then, starting from the expert database and the sound characterization unit 96, the gain in one or more of the higher-frequency hearing aid channels is withdrawn, with which the termination criterion ΔR according to FIG. 10 is no longer met at the comparison unit 59 and a new search cycle for uses the correction model parameters, but with the retraction of the gain in higher-frequency hearing aid channels prescribed by the expert system.

A specific constellation of simultaneously prevailing correction _coefficients Δα _k , ΔCB _k and ΔT _k in a critical frequency band k can be regarded as a band-specific state vector Z _k (Δα _k , ΔCB _k , ΔT _k ) of the correction loudness model. The entirety of all band-specific state vectors Z _k forms the band-specific state space, which is three-dimensional in the case considered here. Band-specific state vectors Z _{k are} primarily responsible for every sound feature that can occur during sound scaling, with "shrill and" muffled "in high-frequency critical bands. This expert knowledge must be stored as rules in the sound characterization unit 96 or the expert system 98.

Are the band-specific correction state vectors Z _k , which give a loudness perception of the individual with the hearing aid essentially the same as that of the norm, as before has been found, a changed state vector Z ' _{k must be} searched for in at least one of the critical bands in order to change the sound. When changing the one band-specific state vector, it must either be changed further so that the loudness remains the same, or at least one other band-specific state vector must also be changed. The parameters of the correction loudness model on the hearing device thus result, based on the parameters of the standard, from a first incremental change “Δ” for conforming loudness adjustment and from second incremental changes δ for sound matching.

The correction loudness model on the hearing aid, for example according to FIG. 12 a), therefore uses parameters of the type

${\text{α}}_{\text{COR}}{\text{= ± Δα}}_{\text{k}}{\text{± δα}}_{\text{k}}{\text{; CB}}_{\text{COR}}{\text{= ± ΔCB}}_{\text{k}}{\text{± δCB}}_{\text{k}}{\text{; T}}_{\text{COR}}{\text{= ± δT}}_{\text{k}}\text{.}$

With each newly found or controlled band-specific state vector on the hearing device model, Z ' _k , which is intended to impart a new timbre to the individual, the corresponding manipulated variables according to FIGS. 12a), 12b) or 13 are switched to the actuators on the hearing device channels and the hearing device thereby readjusted, whereupon the individual continues to assess the sound quality while continuing to comply with the norm of loudness perception and accordingly inputs it at the unit 90 according to FIG. 14. This process is repeated until new δα _k , δCB _k and δT _{k are} searched for with the correct sign, until the individual equipped with the hearing aid perceives the presented acoustic signals satisfactorily, for example also evaluates their sound quality the same as the norm.

Instead of an absolute statement regarding sound quality, which is based on the statement of normal hearing (memory 94) in the interactive method described above, various iteratively comparative, relative test methods, for example according to Neuman and Levitt, have also proven their worth for sound sensation optimization. It is therefore quite possible to calculate a large number of related channel-specific state vector sets, each of which meets the loudness criteria, as explained, by triggering a new calculation cycle each time the termination criterion ΔR according to FIG. 10 is reached, for example with a changed one channel-specific state vector. The individual can then, for example, in a systematic selection process from the found sets of channel-specific state vectors that meet all the loudness requirements, determine the set that optimally satisfies them.

In FIG. 15, again in a functional block representation, the hearing aid according to the invention according to FIG. 12b) (model difference variant) is shown in a form which is preferably implemented. In order to facilitate the overview, the same reference numerals are used as were used for the hearing aid according to the invention according to FIG. 12b).

The output signal of the input converter 63 of the hearing aid is subjected to a time / frequency transformation at a transformation unit TFT 110. The resulting signal, in the frequency domain, is transmitted in the multi-channel time-variant loudness filter unit 112 with the channels 66 to the frequency / time domain FTT transformation unit 114 and from there, in the time domain, to the output converter 65, for example a loudspeaker or another stimulus transducer for the Individual. On a computing part 53a the loudness L _{N is} calculated according to the input signal in the frequency domain and the standard model parameters in accordance with Z _kN .

wird.The individual loudness L _{I is} calculated analogously on the output side of the loudness filter 112. The loudness values L _N and L _I are supplied to the controller unit 116. The controller unit 116 provides the actuators, such as the multipliers 66a or programmable amplifiers, on the loudness filter 112 such that

${\text{L}}_{\text{I.}}{\text{= L}}_{\text{N}}$

becomes.

With this hearing aid according to the invention, the individual loudness is corrected to the standard loudness by adapting the isophones of an individual to those of the standard.

Loudness-corrected frequency masking

Even if the objective function “standard loudness” and possibly also optimization of sound perception can be achieved with the hearing aid according to the invention, as shown for example in FIG. 15, the intelligibility of speech is not yet optimal. This stems from the masking behavior of the human ear, which is different in the case of damaged individual hearing than in the norm. The frequency masking phenomenon states that quiet tones in close frequency neighborhood are faded out from loud tones, i.e. they do not contribute to loudness perception.

If the intelligibility is to be increased further, it must be ensured that those spectral components, which are unmasked in the norm, i.e. perceived, are also perceived in the case of possibly damaged individual hearing, which is usually characterized by a widened masking behavior. Frequency components that are unmasked in normal hearing were usually masked in the case of damaged hearing.

FIG. 16 shows, based on the representation of the hearing aid according to the invention according to FIG. 15 described so far, a further development in which, in addition to the loudness correction of the individual, a masking correction for a hearing-impaired individual, and hence frequency demasking, is carried out. It should be noted in advance that changing the masking behavior of the hearing aid and thus its frequency transmission behavior also changes the loudness transmission, which means that the loudness transmission must be iteratively recreated after each change in the frequency masking behavior.

16, the input signal of the hearing aid is supplied in the frequency domain to a standard masking model unit 118a, whereupon the input signal is masked as in the standard. How the masking model is determined will be explained later.

The output signal of the hearing aid in the frequency range is, analogously, supplied to the individual masking model unit 118b, whereupon the output signal of the hearing aid is subjected to the masking model of the intrinsic individual. The input and output signals masked with the models N and 1 are fed to the masking controller 122 and compared there. In the function of the comparison results, the controller 122 intervenes in a regulating manner on a masking filter 124 until the masking "hearing aid transmission and individual "is aligned with that of the norm.

The multichannel time-variable loudness filter 112 is followed by the likewise multichannel time-variable masking filter 124, which, as mentioned, is set in function of the difference determined at the masking controller 122 such that the norm-masked input signal at unit 118a equals the "individual + hearing aid" -masked output signal at unit 118b will. If the transmission behavior of the hearing aid has now been changed via the masking controller 122 and the masking filter unit 124, the correction loudness L _{KOR of} the transmission no longer corresponds to the required one, and the loudness controller 116 adjusts the manipulated variables on the multi-channel time-variable loudness filter 112, that the controller 116 again determines the same loudness L _I , L _N.

Masking correction via controller 122 and loudness tracking via controller 116 are thus carried out iteratively, the loudness model used, defined by the state vectors Z _LN , Z _LI , remaining unchanged. It is only when both the loudness controller 116 and the masking controller 122 that the iterative matching of the filters 112 and 124 achieves the same within narrow tolerances, is the transmitted signal at the frequency / time transformation unit 114 converted back into the time domain and transmitted to the individual .

Analogous to the loudness model, the frequency _{masking model is} parameterized by state vectors Z _FMN or Z _FMI .

17, based on the masking behavior of normal hearing N, for example hearing impaired individuals I are explained and, going back from the latter, the masking correction is explained in a highly simplified representation.

If, according to the representation N of FIG. 17, a static acoustic signal is presented to the human ear, for example with the three frequency components f 1-f 3 shown, a masking curve F _{fx is} assigned to each frequency component according to its loudness. To the sound and loudness perception of the presented broadband signal, for example with the frequency components f₁-f₃, only the level components, which exceed the masking limits, corresponding to the F _f functions, contribute outstanding levels. In the constellation shown, the norm perceives a loudness to which the _unmasked components L _f1N -L _f3N contribute. The slopes m _unN and m _{obN of} the masking _curves F _f are essentially _independent of frequency and level if, as shown, the frequency scaling takes place in "bark", according to E. Zwicker (in critical bands).

In the case of a hearing-impaired individual I, the masking curves F _f are broadened as far as the gradients m are concerned, and they are also raised. This can be seen from the illustration for a hearing-impaired individual I below in Fig. 17, according to which, with the same acoustic signals presented with the frequency components f₁-f₃, the component on the frequency f₂ is not perceived and thus also contributes to the perceived loudness. The frequency masking behavior of the standard N is again shown in dashed lines in characteristic I of FIG. 17.

It is now a question of realizing a filter characteristic for the individual I by means of "frequency demasking filtering" on a hearing aid, which filter characteristics reflect the masking behavior of the individual corrected to that of the norm. As shown in principle in FIG. 17 at 126, this is preferably implemented in each channel of the hearing aid assigned to a critical frequency band, which as a whole, with frequency-dependent amplification G ′, in particular raises the frequency components masked out in the damaged individual so that the same frequency components as with the norm, contribute equally to the sound perception and loudness perception of the individual. The correction of the L _f1I , L _f3I shares to the L _f1N , L _f3N values is achieved by the loudness _correction - different T _KI , T _kN .

In the case of non-stationary signals, i.e. if the frequency components of the presented acoustic signal vary in time, the total masking limit FMG formed by all frequency-specific masking characteristic curves F _f naturally also varies over the entire frequency spectrum, with which the filter 126 or the channel-specific filter must be guided in a time-variable manner.

The frequency masking model for the standard is known from E. Zwicker or from ISO / MPEG according to the literature reference below. However, the applicable individual frequency masking model with FMG _I must first be determined in order to be able to carry out the individually necessary correction, as shown schematically with the unmasking filter 126 in FIG. 17.

Furthermore, in the hearing aid according to the invention, frequency components which are masked according to the frequency masking model of the standard, i.e. which do not contribute to loudness, are not taken into account at all, i.e. not broadcast.

18 how the individual masking model FMG _{I is} determined on an individual becomes.

Narrow band noise R _o , preferably centered with respect to the center frequency f _{o of} a critical frequency band CB _{k of} the standard or, if already determined as described above, the individual, is presented to the individual via headphones or, and preferably, via the already loudness-optimized hearing aid. A sinusoidal signal, preferably at the center frequency f _o , is added to the noise R _o , as are sinusoidal signals at f _un and f _ob above and below the noise spectrum. These test sinus signals are added sequentially in time. By varying the amplitude of the signals to f _un , f _o and f _ob , it is determined when the individual to whom the noise R _{o is} presented perceives a change in this noise. The corresponding perception _limits, designated A _Wx in FIG. 18, determine three points of the frequency _{masking behavior} F _{foI of} the individual. In this case, certain estimates are preferably used in advance in order to shorten the investigation process. The masking at the center frequency f _o is initially estimated to be -6dB for the hearing impaired. The frequencies f _un and f _ob are chosen to be offset by one to three critical bandwidths with respect to f _o . This procedure is preferably carried out at two to three different center frequencies f _o , distributed over the hearing range of the individual, in order to determine FMG _I , the frequency _{masking model of} the individual or its parameters, such as in particular m _obf , m _unf .

FIG. 19 schematically shows the experimental setup for determining the frequency masking behavior of an individual according to FIG. 18. Noise center frequency f _o , noise bandwidth B and the average noise power A _{N are} set on a noise generator 128. On an overlay unit 130, the output signal of the noise generator 128 is superimposed with the respective test sinus signals, which are set on a sine generator 132. Amplitude A _S , frequency f _{S can be} set on the test sine generator 132. As will be explained with reference to FIG. 20, the test sine generator 132 is preferably operated in a clocked manner, for which purpose it is activated cyclically, for example via a clock generator 134. The superimposition signal is fed to the individual via an amplifier 136 via calibrated headphones or, and preferably, directly via the hearing aid according to FIG. 16, which is still to be optimized with respect to frequency masking.

According to FIG. 20, the noise signals R _{o are} presented to the individual, for example every second, and the respective test sinusoidal signal TS is added to one of the noise packets. The individual is asked whether and, if so, which of the noise packages sounds different from the others. If all noise packets sound the same to the individual, the amplitude of the test signal TS is increased until the corresponding noise packet is perceived differently from the others, then the associated point A _{W is found} on the frequency masking characteristic FMG _I according to FIG. 18. The unmasking model according to block 126 of FIG. 17 can be determined from the masking model of the individual determined in this way and the known standard.

With reference to FIG. 16, the TARGET masking is actually calculated at block 118a depending on the acoustic signal presented, and the filter 124 in the signal transmission path is adjusted via the masking controller 122 until the masking on it and on the individual - model on 118b - provides the same result as required by the leadership masking model in block 118a. As mentioned, the frequency masking correction generally also changes the loudness transmission, so that loudness control and frequency masking control are carried out alternately until both criteria are met with the required accuracy, only then is the "virtually present" acoustic signal converted back into the time range via block 114 and transmitted to the individual.

At this point it must further be noted that it is quite possible to at least estimate the latter from audiogram measurements and / or the loudness scaling according to FIG. 3 instead of the actual measurement of the individual frequency masking behavior. If approximated estimates are assumed for the model identification of the individual, the identification process (FIGS. 18 to 20) is considerably shortened.

Loudness corrected time masking

Even if the loudness that an individual perceives with the hearing aid matches the loudness perceived by the standard and, as has been described, the frequency masking behavior of the hearing aid system with individual is also aligned with the frequency masking behavior of the standard, which is also achieved with the measures described above , speech intelligibility does not remain optimal. This is because human hearing, as a further psycho-acoustic perceptual variable, also has a masking behavior in time that differs in the norm from time masking behavior in an individual, in particular an individual with hearing loss.

While the frequency masking behavior says that if there is a spectral component of an acoustic signal with high level, simultaneous spectral components with low levels and in close frequency neighborhood of the high level component may not contribute to the perceived loudness, it follows from the masking behavior in time that after a loud acoustic signal there may be no noise. That is why slower speaking is helpful for unmasking the time of a hearing impaired person.

In analogy to the problems identified above and solved regarding loudness, sound optimization and frequency masking, it is therefore a matter of further increasing the intelligibility to allow signal sections which are time-unmasked by the standard to be perceived unmasked by the individual with the aid of a hearing device according to the invention.

When taking into account or correcting the time masking behavior on a hearing aid designed as described so far, it should be borne in mind that the procedure described so far is based on the processing of individual spectra. Interactions between temporally following spectra were not to be taken into account. In contrast to this, when considering the time masking effect, a causal connection has to be established between currently pending acoustic signals and future pending acoustic signals. MaW is a further developed hearing aid, which also takes into account the time masking behavior, in principle equipped with time-variable time delay measures in order to be able to take into account and control the effects of a past acoustic signal on a subsequent one. However, this also means that the loudness correction and frequency masking correction, based on individual spectra, as mentioned, must be included in the time so that associated input and output spectra form the loudness and frequency masking corrections stay synchronized.

Again, a change or correction of the temporal signal sequence, which is necessary for a time masking correction, changes the current loudness in each case, with which the loudness correction, as has already been carried out in connection with the frequency masking correction, must be tracked.

21, starting from the above-described hearing aid structure, in particular according to FIG. 16, shows its modification to take time masking corrections into account. After the time / frequency transformation at the unit 110, signal spectra accumulated sequentially in time are stored in a spectrum / time buffer 140 (waterfall spectrum display). Optionally, the spectrum-over-time representation can also be calculated with the Wigner transformation (see Ref. 13, 14). A plurality of input spectra obtained and stored sequentially in time are processed on the standard loudness computing device 53 ' _a - acting for the individual spectra in frequency analogously to the computing device 53a of FIG. 16 - and the L _{N time} image is fed to the controller unit 116a.

The frequency / time inverse transformation unit 114 (Wigner inverse transformation or Wigner synthesis) is preceded by a spectrum / time buffer 142 which acts analogously to the buffer 140.

Analogously, a further computing device 53 ′ _{b determines} the time image of the L _I values determined on the basis of the spectra. This time image is compared with the time image of the L _N values at controller 116a, and the comparison result is used to control a multi-channel loudness filter unit 112a with controlled, time-variable dispersion (phase shift, time delay). The filter 112a thus ensures that the temporal correction loudness image of the transmission with the loudness image of the individual corresponds to that of the norm.

The 142 respectively stored spectra in the buffers 140, the total of signals over a predetermined time period, for example from 20 to 100 msec depict, time and frequency masking model computers for the standard 118 _'a and the individual 118' are further _b supplied to the are parameterized with the norm and individual parameters or state vectors, Z _FM , Z _TM . Both frequency masking model F _N , analogous to FIG. 16, and time masking model T _{M are} implemented therein. The outputs of the computers 118 ' _a , 118' _b act on a masking controller unit 122a, the latter acting on the multi-channel unmasking filter 124a, which can now also be used to control the dispersion in a time-variable manner in addition to 124 from FIG. 16. By modeling computer 118 _'a, 118' _b, and the controller unit 122 is the filter unit 124 so as to frequency transfer and time response controlled so that the frequency and time corrected masked temporal Eingangsspektralbild with the individually modeled (118 _'b) of the output Zeitspektralbildes matches.

The control of the loudness filter 112a and the masking correction filter 124a is preferably carried out alternately, until both assigned controllers 116a and 122a detect predetermined minimum deviation criteria. Only then are the spectra in the buffer unit 142 converted back into the time domain in the correct time sequence at the unit 114 and transmitted to the individual wearing the hearing aid.

Fig. 21 shows a hearing aid structure in which loudness correction, frequency masking correction and time masking correction on signals converted into the frequency range.

A technically possibly simpler embodiment variant according to FIG. 22 consistently takes into account time phenomena on signals in the time domain and phenomena with regard to frequency response on signals in the frequency domain. For this purpose, a time masking correction unit 141 is connected upstream of the time / frequency transformation unit 110, which according to the embodiment of FIG. 16 preferably carries out an instantaneous spectrum transformation, as shown schematically, or, if necessary also as a supplement or replacement, between reverse transformation unit 114 and output transducer 65, such as speakers, stimulator, e.g. an electrode-stimulated cochlear implant.

Between the transformation units 110 and 114, the signal processing in block 117 takes place in accordance with the processing between 110 and 114 of FIG. 16.

The time mask correction unit designated 140 in FIG. 22 is shown in more detail in FIG. 23. It comprises a time-loudness model unit 142, by means of which, preferably as a power integral, the course of the loudness is tracked over the time of the acoustic input signal. Analogously, the instantaneous loudness of the signal in the time domain before its conversion is determined at the time / frequency transformation unit 110 in a further time-loudness model unit 142. The loudness curves in the time of the mentioned input signal and the mentioned output signal are compared on a (simplified) time-loudness controller 144, and on a filter unit 146, namely essentially a gain control unit GK, the loudness of the output signal over which Time considered, that of the input signal aligned.

To carry out the time masking correction, the input signal is fed to a time buffer unit 148, which, according to W. Verheist, M. Roelands, "An overlap-add technique based on waveform similarity ...", ICASSP 93, pp. 554-557, 1993, WSOLA Algorithms or, according to E. Moulines, F. Charpentier, "Pitch Synchronous Waveform Processing Techniques for Text to Speech Synthesis Using Diphones", Speech Communication Vol. 9 (5/6), pp. 453-467, 1990, PSOLA- Algorithms are used.

On a standard time masking model unit 150 _N , the standard time masking to be described is modeled on the input signals, on the further unit 150 _I , on the output signals of the time buffer unit 148, the individual time masking. The time maskings modeled on the signals on the input and output sides of the time buffer unit 148 are compared on a time masking control unit 152, and according to the comparison result, the signal output on the time buffer unit 148 is time-controlled via the algorithms mentioned, preferably used, ie the transmission via the time buffer 148 controlled time-variable expansion factor or delay.

The time masking behavior of the standard is again known from E. Zwicker. The time masking behavior of an individual will be explained with reference to FIG. 24.

24, when the norm over the time t an acoustic signal A₁ is presented, a second, subsequently presented acoustic signal A₂ is only perceived if its level is above the dashed time masking limit TMG _N. The course of this masking limit decay is primarily given by the level of the acoustic signal currently being presented. If signals with different loudness follow, an enveloping TMG results from all TMGs triggered individually by the signals.

In Fig. 24, the time masking limit curve ZMG, for example of a hearing-impaired individual, is shown under representation I with the same, schematically represented acoustic signals A 1 and A 2. It can be seen that the hearing-impaired person may not notice the second signal A₂ in time. The dot-time masking behavior TMG _N , assumed for example, of the curve N is again shown in dash-dotted lines in the course of I. From the difference it can be seen that a time masking correction basically involves either delaying the second signal A₂ on the individual - with the hearing aid - until his individual time masking limit has fallen sufficiently far, or the signal A₂ so too reinforce that the individual is also above his time masking limit.

If in the course of N the perceived area of the signal A₂ is designated L, the last-mentioned procedure on the individual shows that A₂ has to be strengthened so that in the best case the same perceived area L is above the time masking limit of the individual.

In any case, as can also be seen from the explanations relating to FIGS. 21 to 23, corrective interventions which relate to acoustic signals occurring in the future must be carried out from current acoustic signal profiles, shifted in time.

The decay time T _{AN of} the time masking limit TMG _N on the Norm is essentially independent of the level or loudness of the signal triggering the time masking, as shown in Fig. 24 of A₁. This also applies to people with hearing loss, so that in most cases it is sufficient to determine the decay time T _{AI of} the time masking limit TMG _I regardless of the level.

25, to determine the individual time masking limit decay time T _AI, the individual is presented with a click-free and click-free exposing narrowband noise signal R _o . After exposure of the noise signal R _o a test sinusoidal signal with Gaussian wrap-around him will be presented after a set interval T _Paus. A point corresponding to A _{ZM of} the individual time masking limit TMG _{I is} determined by varying the envelope amplitude and / or the pause time T _Paus . Further changes in the pause time and / or the envelope amplitude of the test signal determine two or more points of the individual time masking limit.

This is done, for example, using a test arrangement as shown in FIG. 19, but using a test sine generator 132 which emits a Gauss-encased sine signal. The individual is asked at which pair of values T _Paus and amplitude of the Gauss envelope the test signal after the noise signal is currently being perceived.

Here, too, the individual masking behavior can also be estimated from diagnostic data, which results in a significant reduction in the time for the identification of the individual time masking model TMG _I. As mentioned, the essential parameter of this model is the decay time T _AN or T _AI .

Literature:

1)

E. Zwicker, psychoacoustics, Springer Verlag Berlin, university text, 1982

2)

O. Heller, Hörfeldaudiometrie with the procedure of the division of categories, Psychological contributions 26, 1985

3)

A. Leijon, Hearing Aid Gain for Loudness-Density Normalization in Cochlear Hearing Losses with Impaired Frequency Resolution, Ear and Hearing, Vol. 12, NO. 4, 1990

4)

ANSI, American National Standard Institute, American National Standard Methods for the Calculation of the Articulation Index, Draft WG S3.79; May 1992, V2.1

5)

B.R. Glasberg & B.C.J. Moore, Derivation of the auditory filter shapes from notched-noise data, Hearing Research, 47, 1990

6)

P. Bonding et al., Estimation of the Critical Bandwidth from Loudness Summation Data, Scandinavian Audiolog, Vol. 7, No. 2, 1978

7)

V. Hohmann, dynamic compression for hearing aids, psychoacoustic fundamentals and algorithms, dissertation UNI Göttingen, VDI-Verlag, series 17, No. 93

8th)

A.C. Neuman & H. Levitt, The Application of Adaptive Test Strategies to Hearing Aid Selection, Chapter 7 of Acoustical Factors Affecting Hearing Aid Performance, Allyn and Bacon, Needham Heights, 1993

9)

ISO / MPEG standards, ISO / IEC 11172, 1993-08-01

10)

PSOLA, E. Moulines, F. Charpentier, Pitch Synchronous Waveform Processing Techniques for Text to Speech Synthesis Using Diphones, Speech Communication Vol. 9 (5/6), pp. 453-467, 1990

11)

WSOLA, W. Verheist, M. Roelands, An overlap-add technique based on waveform similarity ..., ICASSP 93, pp. 554-557, 1993

12)

Lars Bramsløw Nielsen, Objective Scaling of Sound Quality for Normal-Hearing and Hearing-Impaired Listeners, The Acoustics Laboratory, Technical University of Denmark, Report No. 54, 1993

13)

B.V.K. Vijaya Kumar, Charles P. Neuman and Keith J. DeVos, Discrete Wigner Synthesis, Signal Processing 11 (1986) 277-304, Elsevier Science Publishers B.V. (North Holland)

14)

Françoise Peyrin and Rémy Prost, A Unified Definition for the Discrete-Time, Discrete-Frequency, and Discrete-Time / Frequency Wigner Distributions, p. 858 ff., IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol. ASSP-34 , No. August 4, 1986

Claims (40)

Method for adapting a hearing aid (HG) to an individual (I), characterized in that one - Quantified at least one psycho-acoustic perception variable (L, F _f , ZMG) of a norm (N) on given acoustic signals; - quantifies the same psycho-acoustic perceptual quantity (L, F _f , ZMG) as the individual (I) perceives with given acoustic signals; - from deviations of the mentioned quantified psychoacoustic perceptual values, the hearing aid adjusts or conceives for the individual in such a way that the psychoacoustic perceptual size, as perceived by the individual with the hearing aid, is at least approximated to that as perceived by the norm, in a more definable manner Relation stands. A method according to claim 1, characterized in that the predeterminable relation is equality. Method according to one of claims 1 or 2, characterized in that the quantifications, the determination of the deviations are carried out with a device separate from the hearing device and the acoustic signals are presented to the individual without a hearing device for quantification. Method according to one of claims 1 or 2, characterized in that one carries out the quantification, the determination of the deviations with a device separate from the hearing aid and the acoustic signals with the individual Presented hearing aid for quantification and preferably created a controllable connection between the device and hearing aid for the transfer of data that depend on the deviations. Method according to one of claims 1 to 4, characterized in that the quantification of the psycho-acoustic perception by the individual is terminated when the deviations are determined with a predeterminable (ΔR) accuracy. Method according to one of Claims 1 to 5, characterized in that the number of quantities to be quantified by the individual is reduced by estimating his perception in advance, preferably on the basis of diagnostic information, and checking the estimation by means of the quantification and, if necessary, specifying it. Method according to one of claims 1 to 6, characterized in that at least one from the group loudness, frequency masking and time masking is used as the psycho-acoustic perception variable. Method according to one of Claims 1 to 7, characterized in that a model (11; 53; 53 '; 118, 120; 53a, 118a; 150) is created and its model for determining the dependence of the psycho-acoustic perception variable on acoustic signals On the one hand, parameters are determined in such a way that the modeled psycho-acoustic quantity is at least approximated due to acoustic signals equal to that perceived by the norm for these acoustic signals, on the other hand so that the modeled psycho-acoustic quantity is at least approximated equal to that perceived by the individual and that one from parameter differences of the two models on conception or setting of the hearing device closes or the transmission of the hearing device leads with the determined differences. Method according to one of Claims 1 to 8, characterized in that a model (11; 53; 53 ') is created to determine the dependency of the psycho-acoustic perception variable on acoustic signals and the parameters thereof are determined in such a way that the model is modeled on the basis of the acoustic signals The psycho-acoustic size is the same as that perceived by the norm for the acoustic signals mentioned, that the psycho-acoustic size perceived by the individual without a hearing aid is further quantified for acoustic signals (5) and the specific model parameters on the model are changed so that they are calculated modeled psycho-acoustic quantity corresponds to a predeterminable amount with that quantified by the individual. Method according to one of claims 8 or 9, characterized in that the determination of the parameters for the modeling of the size perceived by the individual is terminated when the parameters determine the model with predeterminable accuracy. Method according to one of claims 8 to 10, characterized in that the determination of the parameters begins with estimates for this. Method according to one of claims 8 to 11, characterized in that only parameters are determined which determine the modeling with predeterminable accuracy. Method according to one of claims 8 to 12, characterized in that the model (53 '; 118, 120; 53a, 118a; 150) implemented and its parameters for the formation of a correction model, corresponding to the differences or changes mentioned, set. Method according to one of Claims 8 to 12, characterized in that the model for the norm and for the individual is implemented on the hearing device, one is applied to input and output signals of the hearing device and the hearing device transmission is dependent on modeling differences. Method according to one of Claims 8 to 14, characterized in that a model (1) is selected in which changes in parameters (α, CB, T) result in the same changes in the modeled psycho-acoustic variable as changes in assigned physical manipulated variables (66 ) Changes in the psycho-acoustic size at the transmission link on the hearing aid result. Method according to one of claims 8 to 15, characterized in that a plurality of sets of parameter changes which meet the stated conditions are determined and that set is used for the conception or positioning of the hearing aid or for guiding its transmission, which is used for the individual with the hearing aid gives an individually satisfactory sound impression. Method according to one of claims 8 to 16, characterized in that the loudness is used as the psycho-acoustic perceptual variable and by this

is modeled, in which mean: k: running parameters with 1 ≦ k ≦ k _o , numbering of the number k _o considered critical bands; CB _k : spectral width of the considered critical band with the number k; α _k : increase in a linear approximation of the loudness perception scaled in categories when the level of a presented sinusoidal or narrow-band acoustic signal is logarithmically applied, the frequency of which lies approximately in the middle of the band of the critical band CB _k under consideration; T _k : hearing threshold for the sine signal mentioned; S _k : the mean sound pressure level of a presented acoustic signal in the considered critical frequency band CB _k ; and where appropriate the model for level-dependent α _{k is} expanded. Method according to claim 17, characterized in that the hearing thresholds are taken into account individually when modeling the individual, preferably also the α _k and optionally also the CB _{k are} taken into account individually. Method according to one of claims 17 or 18, characterized in that the frequency and / or time masking is additionally used as a psycho-acoustic perception variable. Method according to one of claims 1 to 19, characterized in that the dependence on acoustic signals of a psycho-acoustic size on the hearing aid for the norm and for an individual is modeled and the models on the acoustic signals corresponding electrical input and / or output signals of the Hearing aid used in the time domain and / or in the frequency domain. Method according to Claim 19, characterized in that at least one loudness model and at least one masking model are used intermittently on the hearing aid for guiding transmission manipulated variables. Method according to one of claims 1 to 21, characterized in that the time masking is used as a psycho-acoustic perception variable and this is taken into account on the hearing aid with a controlled time-variable transmission delay, preferably using WSOLA algorithms. Device for adapting a hearing aid to an individual, characterized in that it comprises: - At least one arithmetic unit (11; 53; 53 '; 118, 120; 53a, 118a; 150), in which a model (L, F _f , ZMG) is implemented which shows the dependence of a psycho-acoustic perception size of humans on acoustic signals modeled, and with which, on the input side, an input for signals dependent on acoustic signals is operatively connected, - A comparison unit (15; 59; 116; 122; 116a, 122a; 152), the input of which is operatively connected to the output of the computing unit, with another input of the comparison unit can be operatively connected to an input for the input of a quantified psycho-acoustic perceptual quantity, the comparison unit emits signals on the output side for the design or for the setting or for guiding the transmission behavior of the hearing device. Device according to claim 23, characterized in that a storage unit with fixed data is connected to the computing unit on the input side and the output of the comparison unit acts on a control input of a data modification unit, by means of which the data supplied by the storage unit of the computing unit are changed as a function of the signal at the comparison unit output. Apparatus according to claim 24, characterized in that the output of the comparison unit acts on a threshold value unit, the output of which activates or shuts down the modification unit, the threshold value unit being supplied with a predeterminable threshold value signal. Device according to one of claims 23 to 25, characterized in that the device on the hearing aid comprises at least one computing unit which is connected on the input side to a storage unit and to which signals are supplied as a function of the input and / or output signals of the hearing aid, the computing unit on the output side acts on actuators for transmission to the hearing aid. Apparatus according to claim 26, characterized in that both input and output signals are supplied to the computing unit and signals act on the actuators as a function of a difference in the computing unit output signal, resulting in each case with the input and output signals. Device according to one of claims 22 to 27, characterized in that at least one model is implemented on the computing unit, which models at least one of the psycho-acoustic perception variables loudness, frequency masking, time masking, preferably models at least the loudness. Apparatus according to claim 28, characterized in that a computing unit is provided remote from the hearing aid, on which a storage unit for fixed data acts on the input side via a data modification unit, the comparison unit acting on the output side on a control input on the data modification unit, and a signal generator is also provided, which on the one hand acts on an output control input on the storage unit, on the other hand acts on an electrical / acoustic converter, the computing unit modeling a psycho-acoustic variable, parameterized with the modified data supplied by the storage unit. Apparatus according to claim 29, characterized in that the comparison unit is operatively connected on the input side to a category scaling unit on which the perception can be categorized individually. Apparatus according to claim 28, characterized in that at least one computing unit is provided on the hearing aid, on which the model is implemented, and that a storage unit for parameter data is assigned to it, on the output side acting on actuators for signal transmission on the hearing aid. Apparatus according to claim 31, characterized in that at least two data records are stored on the storage unit, each of which acts on the computing unit with the input and output signals of the hearing aid, on which the modeling difference is formed, depending on which the computing unit acts on the actuators. Device according to one of claims 23 to 32, characterized in that a loudness model according to at least one computing unit

is implemented in what mean. k: running parameters with 1 ≦ k ≦ k _o , numbering of the number k _o considered critical bands; CB _k : spectral width of the considered critical band with the number k; α _k : increase in a linear approximation of the loudness perception scaled in categories when the level of a presented sinusoidal or narrow-band acoustic signal is logarithmically applied, the frequency of which lies approximately in the middle of the band of the critical band CB _k under consideration; T _k : hearing threshold for the sine signal mentioned; S _k : the mean sound pressure level of a presented acoustic signal in the considered critical frequency band CB _k ; and if applicable the implemented model takes the level dependence of α _k into account. Device according to one of claims 23 to 33, characterized in that an intermediate storage unit (55, 57) is connected upstream of both inputs of the comparison unit. Device according to one of claims 24 to 34, characterized in that the computing unit is supplied with an input for acoustic signals via a power generation unit (45, 47). Apparatus according to claim 26, characterized in that the transmission link (117) is arranged on the hearing aid between a time domain-in frequency domain transformation unit (110) and a frequency domain-in time domain transformation unit (114) and the computing unit with transmission link input and Output is functionally connected. Apparatus according to claim 36, characterized in that a further transmission link (148) is provided in front of the time domain-in-frequency domain transformation unit (110) and a computing unit (150) on the input side with both the input and the output of the further transmission link (148 ) is functionally linked and carries out modeling on the basis of the output and input signals of the further transmission path (148), a comparison unit (152) comparing the modeling results and controlling the further transmission path (148) on the output side. Apparatus according to claim 37, characterized in that the further transmission link comprises controllable time delay means, preferably with a WSOLA algorithm. Hearing aid, characterized in that it comprises a computing unit which models the perception of at least one psycho-acoustic variable by humans in response to received acoustic signals. Hearing aid according to claim 39, characterized in that the computing unit calculates the model with at least two parameter sets, each based on hearing aid input and output signals, and provides the transmission between input and output signals as a function of the model difference.

Images