EP4626037A1

EP4626037A1 - Using machine learning algorithm for determining parameters of a sound transfer system of a hearing device

Info

Publication number: EP4626037A1
Application number: EP24167123.9A
Authority: EP
Inventors: Ullrich Sigwanz; Stefan Pislak; Richard Rau; Robert Baur
Original assignee: Sonova AG
Current assignee: Sonova Holding AG
Priority date: 2024-03-28
Filing date: 2024-03-28
Publication date: 2025-10-01

Abstract

A method for determining at least one parameter (40) of a sound transfer system (28) between a loudspeaker (18) and a sound opening (24) of a hearing device (10) comprising: receiving a feedback threshold curve (34) of the hearing device (10) and/or a transfer function (36) between the loudspeaker (18) and a microphone (11) of the hearing device (10); and inputting the feedback threshold curve (34) and/or the transfer function (36) into a machine learning algorithm (38) and determining the at least one parameter (40) of the sound transfer system (38) by the machine learning algorithm (38).

Description

FIELD OF THE INVENTION

The invention relates to a system, method, computer program and computer-readable medium for determining at least one parameter of a sound transfer system between a loudspeaker and a sound opening of a hearing device. The invention also relates to a training method, system, computer program and computer-readable medium for training a machine learning algorithm used for determining the at least one parameter.

BACKGROUND OF THE INVENTION

Hearing devices are generally small and complex devices. Hearing devices can include a processor, microphone, speaker, memory, housing, and other electronical and mechanical components.
Some hearing devices have custom made earpieces with loudspeakers, which are also called receivers. Such earpieces may show deviations in their measured responses compared to their data sheet values. This is due to the way these ear pieces are built. A hollow custom shell usually contains an assembly of a receiver and a sound tube that is acoustically connecting the receiver's sound outlet with the sound opening in the shell. In most cases, these openings and/or outlets are protected from earwax by a removable mesh, which is called wax filter or wax protection.
Measurements have shown that the types and/or flavors of wax filters and the lengths of the sound tubes have impact on the frequency responses of the earpiece. The variations in tube lengths and the final choices of wax filters may be therefore a source of fitting errors due to the unknown acoustic effects of the selected components. Particularly for detachable earpieces, catching the effect of these deviations may be very important, especially when an old earpiece is replaced with a new one that has a different geometry.
In EP 1 830 602 A1 a real ear acoustic coupling quantity representative of the acoustic coupling of a hearing aid to the user's ear or an anatomical transfer quantity is obtained from a transfer function representative of an acoustic transfer from the receiver to the outer microphone such as a signal feedback threshold gain. The obtained quantity may be used for setting a fitting parameter of the hearing instrument, for example a gain correction. The transfer function may be obtained with an artificial neuronal network.

DESCRIPTION OF THE INVENTION

It is an objective of the invention to simplify and/or improve the fitting of a hearing device.
This objective is achieved by the subject-matter of the independent claims. Further exemplary embodiments are evident from the dependent claims and the following description.
A first aspect of the invention relates to a method for determining at least one parameter of a sound transfer system between a loudspeaker and a sound opening of a hearing device.
A hearing device is adapted for receiving a sound signal, processing the sound signal and outputting the sound to an ear of the user, i.e. the user wearing the hearing device. A hearing aid is a hearing device adapted for processing the sound signal, such that a hearing loss of the user is compensated. The processed sound signal is output by the loudspeaker, which also may be called receiver.
The hearing device may comprise a shell and/or housing and/or dome, which has a sound opening, which opens towards an exterior of the hearing device and in particular into the ear canal of the user. The sound opening may be protected by a filter and/or may be connected via tube with the loudspeaker. The filter may be a mesh protecting the tube from ear wax of the user. In general, all components of the hearing device between the loudspeaker and the sound opening may be seen as the physical sound transfer system of the hearing device.
For fitting the hearing device, i.e. adapting the hearing device to a hearing loss of the user, information of the sound transfer system how the sound transfer system alters the sound generated by the loudspeaker may be important. Such information can be derived from parameters of the sound transfer system, such that its geometric shape, its type, etc.
The method may be performed by a fitting station, i.e. a computer device at an office of a hearing care specialist, and in particular a computer program run in the fitting station. Optionally, parts of the method may be run by the hearing device. The method may be performed during fitting of the hearing device by the hearing care specialist, who uses the fitting station.
According to an embodiment, the method comprises: receiving a feedback threshold curve of the hearing device and/or a transfer function between the loudspeaker and a microphone of the hearing device; and determining the at least one parameter of the sound transfer system by a machine learning algorithm based on the feedback threshold curve and/or the transfer function input into the machine learning algorithm.
The feedback threshold curve is indicative of a feedback between the loudspeaker and the microphone of the hearing device. The feedback threshold curve provides a plurality of pairs of a frequency value and a feedback threshold value at this frequency value. The feedback threshold value is a gain value at which the hearing device starts to generate a feedback between its loudspeaker and its microphone at the specific frequency. A common but not exclusive way of determining the feedback threshold of a hearing system uses the method of determining the open loop gain. The system then will directly or indirectly determine, which electric gain will produce an overall acoustic gain so that the attenuation of acoustic path from the receiver outlet to the microphone inlet is just compensated. The feedback threshold value at a frequency value and in particular the feedback threshold curve may be measured by the hearing device itself.
The transfer function between is indicative of a sound transfer between the loudspeaker and the microphone of the hearing device. The transfer function provides a plurality of pairs of a frequency value and a gain value of a sound signal between the loudspeaker and the microphone at this frequency value. The gain value may indicate how the sound signal at the frequency value is amplified or attenuated between the loudspeaker and the microphone. The gain value at a frequency value and in particular the transfer function may be measured by the hearing device itself.
As have been recognized by the applicant, the feedback threshold curve and the transfer function contain much information about the parameters of the sound transfer system, which parameters are useful for fitting the hearing device. Therefore these sets of frequency value and gain value pairs are used as input values of a machine learning algorithm. The feedback threshold curve and/or the transfer function are input into the machine learning algorithm and the at least one parameter of the sound transfer system is determined by the machine learning algorithm. The machine learning algorithm has been trained to output one or more parameters of the sound transfer system based on the feedback threshold curve and/or the transfer function.
The one or more parameters may be used in the fitting station to determine a frequency dependent gain value of the sound transfer system, which may be considered when setting the frequency dependent gain value of the electronic sound processing system of the hearing device. For example, the frequency dependent gain value of the sound transfer system may be subtracted from the frequency dependent gain value of the electronic sound processing system, such that a desired overall gain is reached.
In this way, a better fitting for a user using a customized earpiece may be achieved, since the fitting accounts for the specific sound transfer system of the earpiece. This may result in better acceptance and better audiological performance of the hearing device.
According to an embodiment, the at least one parameter comprises a geometrical parameter of the sound transfer system. Geometrical parameters may be shapes, sizes, lengths, diameters, volumes, etc. of parts of the sound transfer system. Examples for such geometrical parameters are the length of the tube, the (for example inner) diameter of the tube, the volume of the tube, the size of the sound opening, the shape of the sound opening, the mesh size of a filter in the sound opening, etc.
Beside couplings with a constant diameter over the entire length of the tube, there exist also couplings with a varying diameter. The at least one parameter also may comprise varying diameters.
For example, the sound transfer system comprises a tube interconnecting the loudspeaker and the sound opening, wherein the at least one parameter comprises a geometrical parameter of the tube and/or the sound opening.
According to an embodiment, the at least one parameter comprises a type of a part of the sound transfer system. The type may be a number and/or alphanumeric code indicating which standard part was used for the part of the sound transfer system. Examples for such types may be the type of the tube material, the type of the filter, etc.
For example, the sound transfer system comprises a filter in the sound opening, wherein the at least one parameter comprises a filter type of the filter and/or a frequency response of the filter.
As already mentioned, the feedback threshold curve and/or the transfer function may be determined with the hearing devices. This makes it possible that the determination of the at least one parameter of the sound transfer system and the fitting of the hearing device can be performed at the same place and/or at the same time.
According to an embodiment, the feedback threshold curve is determined with the hearing device by controlling the hearing device to perform the following steps: acquiring sound with a microphone of the hearing device; processing the sound with the hearing device; outputting the processed sound with the loudspeaker; monitoring the sound level and determining at which sound level a feedback of the sound takes place. The sound level at which a feedback between loudspeaker and microphone takes place may be determined at a plurality of frequencies. To this end, a sound with a specific frequency and specific sound level may be generated with the loudspeaker. The processing of the sound may be performed with a standard gain.
According to an embodiment, the transfer function is determined with the hearing device by controlling the hearing device to perform the following steps: generating sound with the loudspeaker; acquiring the sound with a microphone of the hearing device; determining the transfer function by determining one or more difference between the generated sound and the acquired sound. The difference could be a magnitude and/or phase difference. The gain values of the transfer function may be determined at a plurality of frequencies. To this end, a sound with a specific frequency and specific sound level may be generated with the loudspeaker. The processing of the sound may be performed with a standard gain. The transfer function then may be based on a ratio of the sound signal output by the loudspeaker and the sound signal acquired by the microphone. Also a phase shift of the sound signal between the loudspeaker and the microphone may be considered as a part of the transfer function.
According to an embodiment, the machine learning algorithm is executed by a fitting station or alternatively in the hearing device itself. A fitting station may be a PC or comparable computing device, which is adapted for communicating with the hearing device, receiving parameters and measurement values from the hearing device and setting fitting parameters in the hearing device. The machine learning algorithm may be run in a software module in the fitting station. In this way, the determination of the at least one parameter of the sound transfer system may be performed during a fitting process of the hearing device, when the user has equipped the hearing device in his or her ear.
According to an embodiment, the machine learning algorithm is an artificial neuronal network. In general, the machine learning algorithm also may be implemented as a convolutional neuronal network (CNN), a multi-layer perceptron (MLP), a Gaussian process, a polynomial regression, a regression tree or a combination of these. An MLP is suitable for regression prediction problems, where a real-valued quantity is predicted given a set of inputs. An MLP is a multi-layer feed-forward neural network with one input layer, one or more hidden layers and an output layer. Each layer, except the input layer, processes the information from the previous layer and sends the results to the next layer. Generally, a three-layer neural network may approximate the nonlinear mapping with arbitrary precision. The MLP may be implemented as a regression prediction problem. The MLP may predict a set of continuously numeric values.
In particular, artificial neuronal networks, in particular with a plurality of layers or concatenated architectures, are well suited for modelling complex functions, such as the at least one parameter of the sound transfer system in dependence of the feedback threshold curve and/or the transfer function.
According to an embodiment, the method further comprises: determining an acoustic coupling curve of an acoustic coupling of the sound transfer system of the hearing device. The acoustic coupling curve may be determined by the fitting station. From the at least one parameter, such as a length and/or diameter of the tube, the diameter of the sound opening and/or the mesh type of the filter, an acoustic coupling between the loudspeaker and the sound opening may be determined. Such an acoustic coupling curve may be modeled as a transfer function and/or a frequency dependent gain value and/or frequency dependent phase shift value.
The acoustic coupling curve may be determined by a mathematical model, which is encoded as algorithm. The mathematical model may be based on formulas depending on the at least one parameter. In particular, the acoustic coupling curve may be the so called ECLD (ear coupler level difference) of an acoustic coupling of a hearing device.
According to an embodiment, the method further comprises: adjusting fitting parameters of the hearing device based on the acoustic coupling curve. The fitting parameters may be determined by the fitting station and sent to the hearing device.
The frequency specific gain values applied to a sound signal acquired by the microphone, which are generated by the hearing device by electronically processing the sound signal, can be adjusted to the acoustic coupling of the sound transfer system. For example, an overall gain of the hearing device is the gain produced electronically, provided by the amplifier of the hearing device and the gain produced physically by the sound transfer system.
A further aspect of the invention relates to a training method for training a machine learning algorithm for determining at least one parameter of a sound transfer system between a loudspeaker and a sound opening of a hearing device. The training method may be performed at a manufacture of the hearing device.
According to an embodiment, the training method comprises: receiving a dataset with records, each record comprising at least one parameter of the sound transfer system, and a feedback threshold curve of the hearing device and/or a transfer function of the hearing device, which have been acquired for a hearing device with a sound transfer system having the at least one parameter; and training the machine learning algorithm with the dataset.
The dataset may have been acquired by the manufacture of the hearing device, which has collected a lot of data for hearing devices such as described herein. For example, a suitable dataset may contain at least 500 records, wherein each record comprises a feedback threshold curve acquired for a hearing device, a length of the tube for the hearing device and a filter type of the filter in the sound opening of the hearing device. In the case of an artificial neuronal network, the training comprises the step of determining The weights of the neurons of the artificial neuronal network.
A further aspect of the invention relates to a computer program for determining at least one parameter of a sound transfer system between a loudspeaker and a sound opening of a hearing device with a machine learning algorithm and/or for training the machine learning algorithm, which, when being executed by at least one processor, is adapted to carry out the steps of the method of one of the previous claims.
For example, the computer program for determining the at least one parameter may be executed in a fitting station and optionally a hearing device in data communication with the fitting station. The computer program for training the machine learning algorithm may be executed in a computing device of a hearing device manufacture.
A further aspect of the invention relates to a computer-readable medium, in which such a computer program is stored. In general, a computer-readable medium may be a hard disk, a USB (Universal Serial Bus) storage device, a RAM (Random Access Memory), a ROM (Read Only Memory), an EPROM (Erasable Programmable Read Only Memory) or a FLASH memory. A computer-readable medium may also be a data communication network, e.g., the Internet, which allows downloading a program code. The computer-readable medium may be a non-transitory or transitory medium. The computer-readable medium may be a memory of the fitting station, the hearing device and/or the computing device of the hearing device manufacture.
A further aspect of the invention relates to a system for determining at least one parameter of a sound transfer system between a loudspeaker and a sound opening of a hearing device. The system comprises a fitting station for performing the method as described herein, and optionally a hearing device for determining the feedback threshold curve and/or the transfer function.
It has to be understood that features of the method as described in the above and in the following may be features of the system, computer program and the computer-readable medium as described in the above and in the following, and vice versa.
These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.

BRIEF DESCRIPTION OF THE DRAWINGS

Below, embodiments of the present invention are described in more detail with reference to the attached drawings.

Fig. 1 shows a hearing device with an earpiece.
Fig. 2 shows a diagram illustrating a method as described herein.
Fig. 3 shows a diagram with a transfer function between a loudspeaker and a microphone of a hearing device.
Fig. 4 shows a diagram illustrating an acoustic coupling of a sound transfer system.

The reference symbols used in the drawings, and their meanings, are listed in summary form in the list of reference symbols. In principle, identical parts are provided with the same reference symbols in the figures.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

Fig. 1 shows a hearing device 10, in particular a hearing aid, which comprises a part 12 to be worn behind the ear and an earpiece 14 to be plugged into the ear canal of the ear of a user. The part 12 comprises a microphone 11, sound processor 13 and further components to process the sound signal from the microphone 11 and to compensate hearing deficiencies of a user. The part 12 is connected via a cable 16 with the earpiece 14, which comprises a loudspeaker and/or receiver 18, which is connected to the cable and which generates sound, which is input into a sound tube 20.
The loudspeaker 18 and sound tube 20 are accommodated in a hollow custom shell 22, which has been customized to an inner shape of the ear canal of the user. The sound tube 20 is acoustically connecting the loudspeaker 18 with a sound opening 24 in the shell 22. The sound opening and/or sound outlet 24 is protected from earwax by a removable mesh and/or filter 26. The tube 20, the sound opening 24 and the filter 26 form a sound transfer system 28, which transfers sound from the loudspeaker 18 into the ear canal of the user.
As an alternative, the hearing device 10 is an in-the-ear device, wherein all components of the part 12 are additionally integrated into the shell 22.
Fig. 2 shows a system 30 comprising a fitting station 32 and a hearing device 10, such as shown in Fig. 1.
Fig. 2 also illustrates a method for determining at least one parameter 40 of a sound transfer system 28 between a loudspeaker 18 and a sound opening 24 of a hearing device 10. The method is performed by the fitting station 32, such as a computer device in an office of a hearing care specialist, and in particular a computer program run in the fitting station 32. Steps of the method may be run by the hearing device 10. The method may be performed during fitting of the hearing device 10 by the hearing care specialist, who uses the fitting station 32.
In the beginning, a feedback threshold curve 34 of the hearing device 10 and/or a transfer function 36 between the loudspeaker 18 and a microphone 11 of the hearing device 10 is determined by the hearing device 10. The fitting station 32 may control the hearing device 10 to determine such data and to send it to the fitting station 32. This makes it possible that the determination of the at least one parameter 40 of the sound transfer system 38 and the fitting of the hearing device 10 can be performed at the same place and/or at the same time.
For example, a feedback threshold curve 34 may be determined with the hearing device 10 by: acquiring sound with a microphone 11 of the hearing device 10; processing the sound with the hearing device 10 and in particular the sound processor 13; outputting the processed sound with the loudspeaker 18; and monitoring the sound level and determining at which sound level a feedback of the sound takes place. The sound level at which a feedback between loudspeaker 18 and microphone 11 takes place may be determined at a plurality of frequencies. To this end, a sound with a specific frequency and specific sound level may be generated with the loudspeaker 18. The processing of the sound may be performed with a standard gain, i.e. a constant gain curve.
The feedback threshold curve 34 provides a plurality of pairs of a frequency value and a feedback threshold value at this frequency value. The feedback threshold value is a gain value at which the hearing device 10 starts to generate a feedback between its loudspeaker 18 and its microphone 11 at the specific frequency value. The feedback threshold value at a frequency value may be measured by the hearing device 10 itself as described above.
As a further example, a transfer function 36 may be determined with the hearing device 10 by: generating sound with the loudspeaker 18; acquiring the sound with a microphone 11 of the hearing device 10; and determining the transfer function 36 by determining a difference between the generated sound and the acquired sound. The gain values of the transfer function 36 may be determined at a plurality of frequencies. To this end, a sound with a specific frequency and specific sound level may be generated with the loudspeaker 18. The processing of the sound may be performed with a standard gain, i.e. a constant gain curve. The transfer function 36 then may be based on a ratio of the sound signal output by the loudspeaker 18 and the sound signal acquired by the microphone 11. Also a phase shift of the sound signal between the loudspeaker 18 and the microphone 11 may be considered as a part of the transfer function.
The transfer function 36 provides a plurality of pairs of a frequency value and a gain value of a sound signal between the loudspeaker 18 and the microphone 11 at this frequency value. The gain value may indicate how the sound signal at the frequency value is amplified or attenuated between the loudspeaker 18 and the microphone 11. The gain value at a frequency value may be measured by the hearing device 10 itself.
The feedback threshold curve 34 of the hearing device 10 and/or the transfer function 36 between the loudspeaker 18 and a microphone 11 of the hearing device 10 is then received in the fitting station 32 and input into a machine learning algorithm 38. The machine learning algorithm 38 then determines the at least one parameter 40.
Fig. 3 shows an example of several transfer functions 36, which have been acquired for hearing devices 10 with different sound transfer systems 28. It can be seen that different sound transfer systems 28 result in different transfer functions 36 with different peaks. The parameters 40 varied for the several transfer functions 36 are the tube 20 and the type of filter 26. Feedback threshold curves 34 have a similar shape. The feedback threshold curves 34 and the transfer functions 36 contain much information about parameters 40 of the sound transfer system 28, which parameters are useful for fitting the hearing device 10.
Fig. 4 shows an example of several acoustic coupling curves 42, which have been determined for different sound transfer systems 28 with different parameters 40. It can be seen that different sound transfer systems 28 result in different acoustic coupling curves 42 with different peaks. Again, the parameters 40 varied for the several acoustic coupling curves 42 are the length of the tube 20 and the type of filter 26.
In particular, Fig. 4 shows how different tube lengths and filter types change the frequency response of a sound transfer system 28 measured on a HA-1 coupler. These deviations also translate in response-changes in real ear conditions. As one can see, there can be differences of up to 20dB. These spikes will deteriorate the perceived sound quality when not adequately taken care of.
What one can also see is that the peaks of Fig. 4 and the peaks of Fig. 3 are correlated. Since there are parameters 40, such as the tube length and the filter type, from which the acoustic coupling curves 42 can be estimated by calculation, training of a machine learning algorithm 38 to determine these parameters 40, when a feedback threshold curve 34 and/or a transfer function 36 is input work very well.
The training already may be performed at a manufacture of the hearing device 10. The machine learning algorithm 38 and in particular its weights may be saved in the fitting station 32.
The training may be based on a dataset with records, each record comprising at least one parameter 40 of a sound transfer system 28, and a feedback threshold curve 34 of the hearing device 10 and/or a transfer function 36 of the hearing device 10, which have been acquired for a hearing device 10 with a sound transfer system 28 having the at least one parameter 40. The dataset may have been acquired by the manufacture of the hearing device 10, which has collected a lot of data for hearing devices 10 such as described herein. As an example, each record may comprise a feedback threshold curve 34 acquired for a hearing device 10, a length of the tube 20 and a filter type of the filter 26.
The machine learning algorithm 38 may be an artificial neuronal network. Artificial neuronal networks, in particular with a plurality of layers, are well suited for modelling complex functions, such as the at least one parameter 40 of the sound transfer system 28 in dependence of the feedback threshold curve 34 and/or the transfer function 36.
Examples for the at least one parameter 40 may be provided by geometric properties and/or types of parts 20, 24, 26 of the sound transfer system 28.
The at least one parameter 40 may comprise a geometrical parameter of the sound transfer system 28. Geometrical parameters 40 may be shapes, sizes, lengths, diameters, volumes, etc. of parts of the sound transfer system 28. Examples for such geometrical parameters are the length of the tube 20, the diameter of the tube 20, the volume of the tube 20, the size of the sound opening 24, the shape of the sound opening 24, the mesh size of a filter 26 in the sound opening 24, etc.
The at least one parameter 40 may comprise a type of a part 20, 24, 26 of the sound transfer system 28. The type may be a number and/or alphanumeric code indicating which standard part was used for the part 20, 24, 26 of the sound transfer system 28. Examples for such types may be the type of the tube material, the type of the filter 26, etc.
As already mentioned, the one or more parameters 40 may be used in the fitting station 32 to determine a frequency dependent gain value of the sound transfer system 28, which may be considered, when adjusting the frequency dependent gain value of the electronic sound processing system and/or the sound processor 13 of the hearing device 10.
The at least one parameter 40 is received in module 44 of the fitting station, which determines an acoustic coupling curve 42 of an acoustic coupling of the sound transfer system 28 of the hearing device 10. From the at least one parameter 40, such as a length and/or diameter of the tube 20, the diameter of the sound opening 24 and/or the mesh type of the filter 26, an acoustic coupling between the loudspeaker 18 and the sound opening 24 may be determined. Such an acoustic coupling curve 42 may be modeled as a transfer function and/or a frequency dependent gain value and/or frequency dependent phase shift value. Fig. 4 shows such acoustic coupling curves 42.
An acoustic coupling curve 42 may be determined by a mathematical model, which is encoded as algorithm in the module 44. The mathematical model 42 may be based on formulas depending on the at least one parameter 40. In particular, the acoustic coupling curve 42 may be the so called ECLD ear coupler level difference of an acoustic coupling of a hearing device 10.
Module 48 of the fitting station 32 receives the acoustic coupling curve 42 and adjusts fitting parameters 46 of the hearing device 10 based on the acoustic coupling curve 42. The fitting parameters 46 may be determined by the fitting station 32 and sent to the hearing device 10.
The frequency specific gain values applied to a sound signal acquired by the microphone 11, which are generated by the hearing device 10 by electronically processing the sound signal with the sound processor 13, can be adjusted to the acoustic coupling 42 of the sound transfer system 28. For example, an overall gain of the hearing device 10 is the gain produced electronically, provided by the amplifier of the hearing device 10 controlled by the sound processor 13 and the gain produced physically by the sound transfer system 28.
With the method described herein, no other additional means except the hearing device 10 and the fitting station 10 are needed to detect the coupling properties of the sound transfer system 28, as a prerequisite for achieving a good fitting. The method reduces the need to pursue expensive or time-consuming real ear measurements, extensions of, or additions to the manufacturing processes and inclusion of non-volatile memory means within the earpiece 14.
The method may enable much better accepted fittings for receiver shell combinations and for acoustic couplings comprising a changeable wax filter system of different types.
Due to a better fitting, the method may increase speech understanding and may increase acceptance by enabling the use of all frequency shaping means to acoustically match targets. An improved acoustical coupling will typically result in better speech intelligibility for soft speech, since it is less likely that some parts of speech are below the hearing threshold, and a better loudness tolerance of loud sounds, since it is less likely that some parts of loud sounds will exceed the uncomfortable level.
While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments. Other variations to the disclosed embodiments can be understood and effected by those skilled in the art and practicing the claimed invention from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or controller or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.

LIST OF REFERENCE SYMBOLS

10: hearing device
12: behind-the-ear part
14: earpiece
16: cable
18: loudspeaker/receiver
20: sound tube
22: shell/housing
24: sound opening
26: filter
28: sound transfer system
30: system
32: fitting station
34: feedback threshold curve
36: transfer function
38: machine learning algorithm
40: parameters of sound transfer system
42: acoustic coupling curve
44: module
46: fitting parameters
48: module

Claims

A method for determining at least one parameter (40) of a sound transfer system (28) between a loudspeaker (18) and a sound opening (24) of a hearing device (10), the method comprising:
receiving a feedback threshold curve (34) of the hearing device (10) and/or a transfer function (36) between the loudspeaker (18) and a microphone (11) of the hearing device (10);

and determining the at least one parameter (40) of the sound transfer system (28) by a machine learning algorithm (38) based on the feedback threshold curve (34) and/or the transfer function (36) input into the machine learning algorithm (38).
The method of claim 1,
wherein the at least one parameter (40) comprises a geometrical parameter of the sound transfer system (28); and/or

wherein the at least one parameter (40) comprises a type of a part (26) of the sound transfer system (28).
The method of claim 1 or 2,
wherein the sound transfer system (28) comprises a tube (20) interconnecting the loudspeaker (18) and the sound opening (24);

wherein the at least one parameter (40) comprises a geometrical parameter of the tube (20) and/or the sound opening (24).
The method of claim 3,
wherein the geometrical parameter is a length and/or diameter of the tube (20).
The method of one of the preceding claims,
wherein the sound transfer system (28) comprises a filter (26) in the sound opening (24);

wherein the at least one parameter (40) comprises a filter type of the filter (26) and/or a frequency response of the filter (26).
The method of one of the preceding claims,
wherein the feedback threshold curve (34) is determined with the hearing device (10) by:
acquiring sound with a microphone (11) of the hearing device (10);

processing the sound with the hearing device (10);

outputting the processed sound with the loudspeaker (18);

monitoring the sound level and determining at which sound level a feedback of the sound takes place.
The method of one of the preceding claims,
wherein the transfer function (36) is determined with the hearing device (10) by:
generating sound with the loudspeaker (18);

acquiring the sound with a microphone (11) of the hearing device (10);

determining the transfer function (36) by determining one or more differences between the generated sound and the acquired sound.
The method of one of the preceding claims,
wherein the machine learning algorithm (38) is executed by a fitting station (32) or by the hearing device (10).
The method of one of the preceding claims,
wherein the machine learning algorithm (38) is at least one of an artificial neuronal network, a convolutional neuronal network, a multi-layer perceptron, a Gaussian process, a polynomial regression, a regression tree.
The method of one of the preceding claims, further comprising:
determining an acoustic coupling curve (42) of an acoustic coupling of the sound transfer system (28) of the hearing device (10).
The method of claim 10, further comprising:
adjusting fitting parameters (46) of the hearing device (10) based on the acoustic coupling curve.
A training method for training a machine learning algorithm (38) for determining at least one parameter (40) of a sound transfer system (28) between a loudspeaker (18) and a sound opening (24) of a hearing device (10), the training method comprising:
receiving a dataset with records, each record comprising at least one parameter (40) of a sound transfer system (28), and a feedback threshold curve (34) of the hearing device (10) and/or a transfer function (36) of the hearing device (10), which have been acquired for a hearing device (10) with a sound transfer system (28) having the at least one parameter (40);

training the machine learning algorithm (38) with the dataset.
A computer program for determining at least one parameter of a sound transfer system (28) between a loudspeaker (18) and a sound opening (24) of a hearing device (10) with a machine learning algorithm (38), which, when being executed by a processor, is adapted to carry out the steps of the method of one of the previous claims.
A computer-readable medium, in which a computer program according to claim 13 is stored.
A system (30) for determining at least one parameter of a sound transfer system (28) between a loudspeaker (18) and a sound opening (24) of a hearing device (10), the system comprising:
a fitting station (32) or hearing device (10) for performing the method of one of claims 1 to 11;

a hearing device (10) for determining the feedback threshold curve (34) and/or the transfer function (36).