Fitting agent for hearing device with environmental model

Abstract

A fitting agent for a hearing device system is disclosed, the hearing device system comprising a hearing device worn by a hearing device user, where the fitting agent comprises one or more processors configured to: initialize a user model and an environment model, the user model comprising a plurality of user preference functions and associated user response distributions; acquiring environmental data indicating a current environment; determining a first initial environment probability of the first environment and a second initial environment probability of the second environment based on the environment data and the environment model, obtaining test settings including primary test settings and secondary test settings of the hearing device based on the first initial environment probability and the second initial environment probability; presenting the test settings to the hearing device user; obtaining a user input of a preferred test setting indicating a preference for the primary test setting or the secondary test setting; and updating the user model based on the preferred test settings and the environmental data to provide an updated user model.

Description

[0001] The present disclosure relates to hearing devices and in particular to related tools, methods and systems for one or more of determining, tuning, fitting and optimizing hearing device parameters. Accordingly, a fitting agent for a hearing device and related methods are provided, in particular methods for updating a user model. Background Art

[0002] Fitting and tuning of hearing devices or hearing aids has always been considered a tedious task for healthcare professionals (HCPs). Traditional methods for fitting hearing device parameters rely on compensating for the user's hearing loss based on an audiogram by applying rules such as NAL-NL1 or NAL-NL2. However, these rules do not take into account specific user preferences.

[0003] Recent approaches involve preference learning of hearing devices.

[0004] EP 3 493 555 A1 relates to a method for tuning hearing device parameters of a hearing device and a hearing device. The method comprises: initializing a model; obtaining an initial test setting defined by one or more initial test hearing device parameters; designating the initial test setting as a primary test setting; obtaining a secondary test setting based on the model; outputting a primary test signal according to the primary test setting; outputting a secondary test signal according to the secondary test setting; detecting user input of a preferred test setting; updating the model based on the primary test setting, the secondary test setting and the preferred test setting; and updating the hearing device parameters of the hearing device based on the hearing device parameters of the preferred test setting based on a determination that a tuning criterion is met. Summary of the invention

[0005] Challenges remain in improving tools, methods and equipment to allow for improved fitting and tuning of hearing device parameters.

[0006] A fitting agent for a hearing device system is disclosed, the hearing device system comprising a hearing device worn by a hearing device user, wherein the fitting agent comprises one or more processors configured to: initialize a user model and an environment model, the user model comprising a plurality of user preference functions and associated user response distributions; obtain environment data indicating a current environment; optionally determine a first initial environment probability of a first environment and / or a second initial environment probability of a second environment based on the environment data and / or the environment model, and obtain a test setting comprising a primary test setting and a secondary test setting of the hearing device based on the first initial environment probability and / or the second initial environment probability; present the test setting to the hearing device user; obtain user input of a preferred test setting indicating a preference for the primary test setting or the secondary test setting; and optionally update the user model based on the preferred test setting and the environment data to provide an updated user model.

[0007] The present disclosure takes into account any differences in user preferences in different environments and provides efficient and memory-efficient user preference model updates and / or efficient and memory-efficient environment model updates. In particular, the need to store historical and specific user feedback (previous primary and secondary test settings and their preferences) can be eliminated, which in turn reduces memory requirements.

[0008] Furthermore, the context-dependent user preference function may provide a simpler user preference function, thereby allowing one or more hearing device parameters of a hearing device to be configured in a simple and effective manner. Furthermore, the present disclosure provides an improved listening experience for the user by improving the modeling of user preferred hearing parameter settings in different environments, thereby leading to the application of optimized settings in the hearing device, which in turn allows an improved user experience.

[0009] Furthermore, the present disclosure provides an efficient automatic search for optimal hearing device parameters by incorporating a personalized environment model into the learning cycle. A fitting agent, device and method are provided which allows learning the environment in an efficient and minimally obtrusive manner by authorizing the user to make direct decisions and have a direct impact on the fitting and / or tuning process.

[0010] Furthermore, the present disclosure allows that hearing device parameters may be configured (such as fitting and / or tuning) during normal operating conditions and / or with little user input / interaction, thereby providing a simple and smooth user experience of the hearing device. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and other features and advantages of the present invention will become apparent to those skilled in the art through the following detailed description of exemplary embodiments of the present invention with reference to the accompanying drawings, in which:

[0012] Figure 1 Schematic diagram showing a fitting agent with a preference learning system that perceives a scene / environment,

[0013] Figure 2 Schematically illustrating a hearing system according to the present disclosure,

[0014] Figure 3 Describe the interaction between the user and the fitting agent, and

[0015] Figure 4 is a flow chart of an exemplary method according to the present disclosure.

[0016] Reference numerals list

[0017] 1 Hearing Device System

[0018] 2 Hearing devices

[0019] 4 Accessories

[0020] 5 Server Equipment

[0021] 6 Transceiver Module

[0022] 8 Antenna

[0023] 10 Transceiver

[0024] 11 Wireless connection between hearing aids and accessory devices

[0025] 11A Wireless connection between accessory equipment and server equipment

[0026] 12 First Microphone

[0027] 14 First microphone input signal

[0028] 16 Processors

[0029] 18 Electrical output signal

[0030] 20 User Interface

[0031] 22 Receiver

[0032] 24 User interface for accessory devices

[0033] 26 Touch Display

[0034] 27 Fitting Agent

[0035] 27A Fitting Agent Part I

[0036] 27B Part 2 of the Fitting Agent

[0037] 27C Part III of the Fitting Agency

[0038] 28 (virtual) start button

[0039] 30 Control signals indicating primary and secondary test settings

[0040] 32 Primary Virtual Button

[0041] 34 Secondary virtual buttons

[0042] 38 Control signal indicating hearing device parameters of preferred test settings

[0043] 40 users

[0044] 41 Environment

[0045] 42 nth trial, test setup and

[0046] 44 User response to the nth trial

[0047] 46 n+1th trial, test setup and

[0048] 50 Environmental Models

[0049] 52 Preference Learning System

[0050] 100 Method for updating a user model and / or an environment model of a hearing device user and / or determining hearing device parameters of a hearing device

[0051] S102 Initialize user model

[0052] S104 Obtaining environmental data

[0053] S106 Get Test Settings

[0054] S106A Get the Primary Test Setup

[0055] S106B Get Secondary Test Setup

[0056] S108 presenting the test setup to the hearing device user

[0057] S108A outputs primary test signal

[0058] S108B outputs secondary test signal

[0059] S110 Get user input

[0060] S112 Update user model and / or environment model

[0061] S112A Update multiple user preference functions of the user model, such as a subset or all of the user preference functions

[0062] S112B Update multiple (such as all) predictors / user response models of the user model

[0063] S112C Determine a posteriori of parameters of a user preference function of a user model based on a posteriori of prior parameters, a preferred test setting, a non-preferred test setting, and environmental data

[0064] S114 determines whether the stop criteria are met

[0065] S116 Update hearing device parameters

[0066] ENV_D Environmental Data DETAILED DESCRIPTION

[0067] Various exemplary embodiments and details are described below with reference to the accompanying drawings when relevant. It should be noted that the drawings may or may not be drawn to scale, and elements of similar structure or function are represented by similar reference numerals in all drawings. It should also be noted that the drawings are intended only to facilitate the description of the embodiments. They are not intended to serve as an exhaustive description of the present invention or as a limitation on the scope of the present invention. In addition, the illustrated embodiments do not need to have all aspects or advantages shown. The aspects or advantages described in conjunction with a particular embodiment are not necessarily limited to that embodiment, and may be practiced in any other embodiment, even if not so described or not so explicitly described.

[0068] A fitting agent is disclosed. The fitting agent or at least a first part thereof may, for example, be implemented as an application in an accessory device such as an electronic device. The accessory device comprises an interface, a processor and a memory. The accessory device may, for example, be or comprise a mobile phone such as a smartphone, a smartwatch, a dedicated device, a computer such as a laptop or PC, or a tablet computer. The fitting agent or at least a second part thereof may be implemented in a server device. The fitting agent or at least a third part thereof may be implemented in a hearing device.

[0069] The present disclosure relates to a hearing device system, a fitting agent, an accessory device of the hearing device system and a hearing device and related methods. The accessory device forms an accessory device of the hearing device. The accessory device is typically paired with or wirelessly coupled to the hearing device. The hearing device may be, for example, a hearing aid of the behind-the-ear (BTE) type, in-the-ear (ITE) type, in-the-canal (ITC) type, receiver-in-the-canal (RIC) type, receiver-in-the-ear (RITE) type or microphone-in-the-ear (MaRIE) type. The hearing device may be a wearable device, such as a pair of earplugs or headphones. Typically, the hearing device system is owned and controlled by a hearing device user.

[0070] The hearing device system may comprise a server device and / or a fitting device. The fitting device is controlled by the distributor and is configured to determine configuration data, such as fitting parameters. The server device may be controlled by the hearing device manufacturer.

[0071] The fitting agent may be a fitting agent for a hearing device system comprising a hearing device worn by a hearing device user.

[0072] The fitting agent includes one or more processors. The one or more processors are configured to initialize a user model and an environment model. The environment model is also denoted as an environment model.

[0073] The user model includes a plurality of user preference functions and associated user response distributions. The user response distributions are optionally state-dependent and can be expressed as a single distribution, where, depending on the state, the distribution has a corresponding state-dependent form.

[0074] There is a probability mapping between each user preference function and the environment distribution function of the environment model.

[0075] The user model optionally represents one or more probabilistic descriptions of user responses when comparing two sets of hearing device parameter settings. The components of the user model include one or more (such as multiple) user preference functions, and one or more (such as multiple) distributions of user responses to the presented parameter selections, also represented as user response distributions. The user model includes one or more distributions of parameters of the corresponding user preference functions. In other words, each user preference function has an associated distribution of hearing device parameters. In other words, the user model may include a first user preference function and an associated first user response distribution, wherein the first user preference function is optionally associated with a first environment. The user model may include a second user preference function and an associated second user response distribution, wherein the second user preference function is optionally associated with a second environment. The user model may include a third user preference function and an associated third user response distribution, wherein the third user preference function is optionally associated with a third environment. The user model may include K user preference functions and associated user response distributions, wherein the kth user preference function k=1, 2, ..., K is optionally associated with the kth environment.

[0076] The vector of hearing device parameters is optionally defined on an M-dimensional continuous compact surface. In particular, the hearing device parameter x is optionally defined on an M-dimensional hypercube, i.e. x∈[0,1] M In one or more exemplary fitting agents / methods, the hearing device parameters may be normalized by their physical ranges. The fitting agent / method is configured to find optimized / improved values ​​of the hearing device parameters for a particular user, also denoted as θ. The number M of hearing device parameters may be 1 and / or less than 100, such as in the range from 10 to 50. The number M of hearing device parameters may be greater than 20, such as in the range from 25 to 75.

[0077] Usually, the user preference function is unknown. User preference function f(x, θ, Λ, k) = f(x; θ k , Λ k ), k = 1, 2, ..., K, where K is the number of user preference functions, which may be, for example, hearing device parameters x∈[0, 1] with known form but unknown shape M This shape is optionally determined by fitting or tuning parameters θ k ∈[0, 1] M and the scaling matrix Λ k The scaling matrix Λ k It can be a positive definite scaling matrix Λ k The scaling matrix can be a diagonal matrix Λ k=diag([λ1,λ2,...,λ M )]), m=1,2,…,M。

[0078] The user preference function or at least one or more of the user preference functions of the K user preference functions may be unimodal.

[0079] In one or more exemplary fitting agents / methods, a distribution prior may be applied to Λ k For example, a lognormal prior can be applied to Λ k Each element of Alternatively, a gamma distribution prior can be applied to each element of Λ, such that λ m ~Gamma(α m , β m ).

[0080] Scaling matrix Λ k It does not need to be a diagonal matrix. The scaling matrix Λ k Can be selected as Λ k = L′*L, where L is a lower triangular matrix (also called Λ k Cholesky decomposition of ). The Gaussian prior can be applied to each element of L, such as

[0081] The k-th environment maximizes the independent variable θ k can be a transformation of another variable, for example, it can be constrained by the following prior assumptions

[0082]

[0083] in is the cumulative density function of a probability distribution such as the standard normal distribution, and is a sample from another probability distribution.

[0084] In one or more example fitting agents / methods, maximizing the independent variable θ k It can be obtained from the following about k Expressed as The transformation assumptions and the corresponding prior distribution of the transformed variables are constrained:

[0085] in in is the cumulative density function of the standard normal distribution, and is a vector with mean μ k and the covariance matrix ∑ k The mean and covariance values ​​can be learned from the user responses.

[0086] In one or more example fitting agents / methods, the scaling matrix may be a transformed scaling matrix, such as by applying a logarithmic transformation to elements of the scaling matrix.

[0087] The user preference function can be expressed as f or f k (x;θ k , Λ k ), where each user preference function f k (x;θ k , Λ k ) is associated with an environment or state k in the K environment / user preference functions. User preference function f k (x;θ k , Λ k ) can be given by the following formula:

[0088] f k (x;θ k , Λ k )=-((x-θ k ) T Λk ( x-θ k )) v , (1)

[0089] where x is optionally in the hypercube [0, 1] M The M-dimensional vector in , and represents the (M) hearing device parameters of the device in the k-th environment, θ k Yes k The maximization independent variable, Λ k is a positive definite M×M scaling matrix characterizing the user's sensitivity to changes in hearing device parameters, M is an integer, and v is a real-valued exponent. The real-valued exponent v may be in the range from 0.01 to 0.99. The real-valued exponent v may be less than 0.5, such as in the range from 0.01 to 0.45. The real-valued exponent v may be greater than 0.5, such as in the range from 0.55 to 0.99. In one or more exemplary fitting agents / methods, the real-valued exponent v may be in the range from 0.25 to 0.75, such as 0.5.

[0090] User preference function f k (x;θ k , Λ k ) can be given by the following formula:

[0091] f(x;θ k , Λ k )=exp(-(x-θ k ) T Λ k (x-θ k )).

[0092] Variable θ k ,Λ k It is also expressed as the user preference function parameter, where θ k is the optimal value of the user preference function corresponding to the optimal user-specific HA parameters, and is a diagonal matrix indicating user sensitivity, i.e., the ability to discriminate parameter changes. Note that other unimodal user preference functions are also possible.

[0093] The fitting agent / the one or more processors of the fitting agent are configured to update the user model, such as one or more user preference functions f based on the hearing device parameters of the preferred test setup and the environmental data. k (x;θ k , Λ k ) and / or one or more user response distributions P u， In other words, the user response distribution P u， User responses in one or more environments can be modeled. The user response distribution can be a weighted user response distribution of all or a subset of K environments, for example, where the weights are based on the environment probabilities ENVP_k, k=1, ..., K and the user preference function f k (x;θ k , Λ k )’s parameters.

[0094] The fitting agent is configured to update the user model based on the preferred test settings, the hearing device parameters of the non-preferred test settings and the optional environmental data. In particular, the fitting agent can be configured to update the user model based on (r, x_ref, x_alt, s). Updating the user model may include updating the user preference function or at least its parameters and / or updating the parameter distribution associated with the user preference function based on the environmental data and one or more of the user inputs of the primary test settings, the secondary test settings and the preferred test settings. In other words, the parameters of the user preference function and / or the associated parameter distribution can be updated based on the test results of the preferred test settings including the primary test settings, the secondary test settings, the primary test settings and the secondary test settings, and the environmental data. Updating the user model may include updating (one or more) user response distributions or at least its parameters based on the environmental data and one or more of the user inputs of the primary test settings, the secondary test settings and the preferred test settings. In other words, the parameters of the user preference function can be updated based on the test results of the preferred test settings including the primary test settings, the secondary test settings, the primary test settings and the secondary test settings, and the environmental data. Updating the user model may include updating the user response distribution(s) or at least parameters thereof based on one or more environmental probabilities, optionally including the first environmental probability and the second environmental probability. Thus, updating the user model may include updating the user preference function parameters θ based on one or more of the environmental data and the user input of the primary test setup, the secondary test setup, and the preferred test setup. k and / or k In other words, the parameter distribution associated with the user preference function may be updated based on the test results of the primary test setting, the secondary test setting, the preferred test setting of the primary test setting and the secondary test setting, and the environmental data. Updating the user model, such as updating the user preference function parameter θ k and / or k The parameter distribution of (one or more) may include updating the user preference function parameter θ based on one or more environment probabilities, optionally including a first environment probability and a second environment probability. k and / or k The parameter(s) of the distribution.

[0095] In one or more exemplary fitting agents and / or methods, updating the user model may be based on Bayesian inference. Updating the user model may include updating one or more of the parameters of the user preference function and / or the user response distribution(s) and / or parameter distributions associated with the user preference function. Updating the user model may include determining one or more posteriori for the parameters of the user preference function(s).

[0096] Updating the user model may include, for example, determining one or more user preference functions, such as a subset of user preference functions or all of the parameter posteriori based on environmental data, previous parameter posteriori (such as the last or current parameter posteriori), preferred test settings, and non-preferred test settings.

[0097] Therefore, the user preference function for the kth environment or state can be given as:

[0098]

[0099] The tuning parameter corresponds to the function θ k , which is the optimal position of the user's hearing device parameters; and corresponds to the optimal value around this optimal value Λ k The distribution of , which respectively characterizes the user's sensitivity to changes in hearing device parameters. The tuning parameters are user-specific and need to be learned in order to provide an optimal or improved user experience. This form of user preference function has attractive properties, which allows to quickly narrow down the search space.

[0100] A user model such as one or more user preference functions is optionally based on a first assumption and / or a second assumption. The first assumption may be that the user preference function is a unimodal preference function representing the user's preferences. The second assumption may be that the user may have uncertain preferences, which may be represented by an additive random variable. The value function u may be defined k (·) in order to model the uncertainty of user preferences. The value function u k (·) can be defined as

[0101] u k (x) = f k (x;θ k ,Λ k )+ε k , (5)

[0102] where the user uncertainty error ε is optionally assumed k has a Gaussian distribution, such as a standard Gaussian distribution, e.g.

[0103]

[0104] The variance of a Gaussian distribution may be real valued, for example in the range from 0 to 1 or greater than 1.

[0105] The user's experiment defines a pair of test settings, including a primary test setting and a secondary test setting. The user obtains user responses by pairwise comparison of the test settings from the experiment, which is defined by a pair of test settings {x_ref, x_alt}, where the primary test setting x_ref and the secondary test setting x_alt are the so-called reference and alternative parameter suggestions, respectively. In other words, the execution is also denoted as D nTests T_n, where each test T_n includes or is set by a primary test Secondary Test Setup and preferred / selected test settings n Definition: The index n-1 refers to the previous trial T_n-1.

[0106] In a user system, the user's response to a test depends on the user's preference. This relationship can be modeled via a comparative judgment model as follows:

[0107] P(r|x ref ,x alt ,f,z=k)=

[0108] Φ(f(x alt θ k ,Λ k )-f(x ref θ k ,Λ k )) r ·Φ(f(x ref θ k ,Λ k )-f(x alt θ k ,Λ k )) 1-r (7)

[0110] where Φ(·) is the cdf of the standard Gaussian distribution, x ref and x alt are two sets of hearing device parameters that the user compares, and z∈{1, 2, ..., K} is the environment / state of the user's environment / environment identifier. It is assumed that the user has some uncertainty about his / her choice, which is modeled by additive random noise with a standard Gaussian distribution.

[0111] In one or more example fitting agents, assuming there are K possible environments / states of the environment, a Gaussian mixture model (GMM) is used to model the environment data / observed scene s:

[0112]

[0113] where π k is the mixing probability of the environment, and μ s，k and∑ s，k are the mean and covariance matrices of the Gaussian distribution corresponding to the kth environment.

[0114] The environment probabilities ENVP_k of K environments, k=1, ..., K, can be based on the following:

[0115]

[0116] where s t is the signal / scene observed at time step t, and z t is the environment at this time step.

[0117] A complete hearing device system is optionally defined as a series of distributed

[0118] By a series of The generation distribution of the fitting agent defined will be the next test The generative process of is described as a weighted combination of proposals of possible environments. Generate a prediction distribution Characterize a next test Define preliminary test setup / reference recommendations in a deterministic manner based on user responses in previous trials Secondary Test Setup / Alternative Suggestions is generated a posteriori from the user preferences, since from an optimization point of view the fitting agent aims to reach the best state as quickly as possible and therefore to provide an option that is better than the reference suggestion with high probability. Obviously, this option will correspond to the user preference or the fitting agent's estimate of it.

[0119] Finally, these distributions The fitting agent strategy defined for sequential trial design in a hearing device system may be to generate a series of trials t=1, 2, ..., T based on the interaction history with the user in some environment according to the following rule, where I(·; ·) represents mutual information:

[0120]

[0121] In order to generate the desired trials, the fitting agent must first evaluate the user model. Therefore, the fitting agent must sequentially assign the prediction distribution Q associated with the user model t , so that for all system interactions, Q t The difference between and the true distribution of user responses is minimized, and this difference is given by the normalized weighted KL divergence after T interactions:

[0122]

[0123] With some weighting function w(θ, Λ). For the hearing device system defined above, this normalized KL divergence can be given by the following expression:

[0124]

[0125] in is the distribution that minimizes the normalized weighted KL divergence, and the weighted distribution corresponds to the posterior distribution of the parameters of the user preference function.

[0126] It can be seen that the metric in (12) can be written as the sum of the mutual information at each interaction step and can characterize the fitting proxy uncertainty about the user system. Therefore, the fitting proxy goal of generating informative trials (test settings) is transformed into finding trials / test settings that minimize the remaining system uncertainty. This metric also allows for efficient monitoring of system performance.

[0127] Finally, after the fitting agent strategy has been defined, the posterior representation of the user(s) preferences must be learned This step is also part of the fitting agent strategy, since the posterior of the user preferences is also the weighted distribution used in generating the predicted user response distribution. In one or more example fitting agents, a GMM is used to estimate the environment state, and then optionally Bayesian variational inference is used to additionally infer the user's preferences for hearing device parameters for the K environments.

[0128] In one or more example fitting agents, obtaining environmental data includes obtaining audio data, and optionally determining environmental data based on the audio data and / or including the audio data in the environmental data. In other words, one or more processors of the fitting agent may be configured to obtain audio data, and determine environmental data or at least one or more environmental parameters based on the audio data. The audio data may include first audio data representing or indicating audio recorded by one or more microphones of a hearing device of a user. The audio data may include second audio data representing or indicating audio recorded by one or more microphones of one or more accessory devices of the user. The audio data may include third audio data representing or indicating audio wirelessly transmitted to the hearing device of the user. For example, the fitting agent may be configured to classify the environment based on the hearing device audio, and set one or more environmental identifiers and / or environmental probabilities of the environmental data accordingly.

[0129] In one or more example fitting agents, obtaining environment data includes, for example, receiving user input indicating an environment via a user interface of the accessory device. Thus, the user can select and indicate a current environment via the accessory device, for example, from a list of environments presented on a touch display of the accessory device.

[0130] In one or more example fitting agents, one or more processors are configured to obtain a first environment probability of a first environment and / or a second environment probability of a second environment based on the environment data, and wherein obtaining a test setting and / or updating a user model is optionally based on the first environment probability and / or the second environment probability. In other words, the one or more processors of the fitting agent may be configured to obtain (such as one or more of determining, estimating, receiving, and retrieving) a first environment probability of a first environment ENV_1, also denoted as ENVP_1, and / or a second environment probability ENVP_2 of a second environment ENV_2, and optionally obtain a test setting and / or update a user model, such as a first user preference function and / or a second user preference function, based on the first environment probability and / or the second environment probability. The environment data optionally includes an environment probability ENVP_k for each environment ENV_k of the K environments or associated with each environment ENV_k, wherein the environment probability ENVP_k is the probability that the current environment is the corresponding environment ENV_k. For example, ENVP_2=0.5 may indicate that the probability that the current environment is the second environment ENV_2 is 0.5. For example, ENVP_2=1.0 may indicate that the current environment is the second environment ENV_2.

[0131] In other words, a fitting agent, such as one or more processors of a fitting agent, may be configured to obtain (such as one or more of determining, estimating, receiving, and retrieving) one or more environmental probabilities (such as K environmental probabilities), for example based on environmental data (such as one or more environmental parameters), the environmental probabilities indicating the probability that the current environment is the corresponding environment.

[0132] In one or more exemplary fitting agents, the ambient probability is given by Equation 8A.

[0133] The first environment probability may indicate a probability that the current environment is a first environment. The first environment may be a first type of environment, such as an environment characterized by low SNR and high sound level / power, such as a cocktail party or a concert.

[0134] The second environment probability may indicate a probability that the current environment is the second environment. The second environment may be an environment of a second type, such as an environment characterized by a low SNR and a medium sound level / power.

[0135] In one or more example fitting agents, the plurality of user preference functions includes K user preference functions, where K is 2, 3, or greater than 3. The number K of user preference functions may be in the range from 5 to 25, such as from 5 to 10, for example 7. In one or more example fitting agents, updating the user model includes selecting a user preference function to be updated from the K user preference functions based on the environmental data, and updating a posteriori of parameters of the user preference function to be updated based on the test setting and the user input of the preferred test setting.

[0136] In one or more example fitting agents, updating the user model includes selecting multiple user preference functions (such as 2, 3, 4 or more) to be updated from K user preference functions based on the environmental data, and updating the posterior or posterior distribution of parameters of the multiple user preference functions to be updated based on the test setting and the user input of the preferred test setting. In one or more example fitting agents, updating the user model includes updating K user preference functions and parameter distributions associated therewith, or updating user preference functions with environmental probabilities greater than 0 and parameter distributions associated therewith.

[0137] In one or more example fitting agents, obtaining the environment data includes, for example, determining an environment identifier using a Gaussian mixture model and / or obtaining one or more environment probabilities, such as K environment probabilities. Updating the user model may be based on the environment identifier and / or one or more environment probabilities, such as environment probabilities ENV_k, k=1, 2, ..., K. For example, updating the user model based on the environment identifier may include selecting a user preference function and associated parameter distribution for updating based on or according to the environment identifier, and selecting the user preference function and associated parameter distribution for updating based on the preferred test settings and optionally the non-preferred test settings.

[0138] In one or more example fitting agents, the fitting agent is configured to determine whether a test criterion based on environmental data is met, wherein the fitting agent is configured to perform, based on a determination that the test criterion is met: obtaining a test setup including a primary test setup and a secondary test setup for a hearing device; presenting the test setup to a hearing device user; obtaining user input of a preferred test setup indicating a preference for the primary test setup or the secondary test setup; and updating a user model based on hearing device parameters of the preferred test setup and environmental data. Thus, the test criterion may be used to determine when user input is required to improve the accuracy of a user preference function. Thus, efficient and point-targeted training of a user preference function is provided, such that, for example, user input is prompted only when required, thereby reducing and / or optimizing the number of user interactions.

[0139] In one or more example fitting agents, the test criteria is based on an uncertainty metric that indicates uncertainty of the user preferences in the current environment. For example, if it is determined that the user preference function associated with the current environment is not accurate enough, such as having a high degree of uncertainty, such as reflected in an uncertainty metric that is greater than a threshold, then the test criteria may be met or at least partially met.

[0140] In one or more example fitting agents, obtaining a test setup including a primary test setup and a secondary test setup for a hearing device includes determining the secondary test setup based on the environmental data, eg, as described above with respect to equations (10)-(12).

[0141] In one or more example fitting agents, the fitting agent is configured to determine and update hearing device parameters of the hearing device based on an updated user model and / or environmental data, such as environmental probabilities of the environmental data.

[0142] The environment model can be a Gaussian mixture model, for example with K predefined maximum clusters / states / environments. K is pre-set, but some clusters may be empty. In other words, a cluster of environments can represent an environment.

[0143] The environment model may be a non-parametric Bayesian mixture model, such as a Dirichlet process Gaussian mixture model, also denoted DP-GMM. A non-parametric Bayesian mixture model models an unknown infinite number of environment clusters.

[0144] In one or more examples, updating the environment model can include increasing the number of clusters, such as increasing the number of populated or non-empty clusters of the environment model.

[0145] In one or more examples, updating the environment model can include reducing the number of clusters, such as reducing the number of populated or non-empty clusters of the environment model. In other words, updating the environment model can include increasing the number of empty clusters.

[0146] A fitting agent, such as one or more processors of the fitting agent, is configured to obtain environmental data indicative of the current environment. The environmental data may include a signal vector or signal matrix represented as s of one or more input signals, and / or other scene information or any parameters determined based thereon, such as signal-to-noise ratio (SNR) and power.

[0147] For one or more input signals, an environment such as an acoustic environment may be characterized by one or more of a first environmental parameter such as an input signal level or input power and a second environmental parameter such as an SNR. In other words, the environmental data may include a (first) SNR and / or a (first) power of the (first) input signal. Therefore, acquiring environmental data indicating the current environment may include acquiring (such as determining or receiving) one or more environmental parameters optionally including a (first) SNR and / or a (first) power of the (first) input signal.

[0148] The fitting agent, such as one or more processors of the fitting agent, is configured to determine a first initial environment probability for the first environment and / or a second initial environment probability for the second environment, for example based on the environment data and the environment model.

[0149] The fitting agent, such as one or more processors of the fitting agent, is configured to obtain (such as one or more of determining, receiving and retrieving) a test setup comprising a primary test setup and a secondary test setup for the hearing device, e.g. based on the first initial ambient probability and / or the second initial ambient probability.

[0150] The primary test setup is also denoted as reference suggestion and the secondary test setup is also denoted as alternative suggestion. A test setup is also referred to as a trial. Each trial denoted by {θref, θalt} comprises a reference suggestion θref and an alternative suggestion θalt.

[0151] The fitting agent is configured to obtain the primary test settings of the hearing device, also denoted as θref, x_ref or x ref The primary test setup is a vector comprising M hearing device parameters of the hearing device. The hearing device parameters may include one or more of filter coefficients, compressor settings, gains or other parameters related to the operation or signal processing of the hearing device. The primary test setup may be based on and / or dependent on the current environment, such as based on the first initial environment probability and / or the second initial environment probability. In other words, the primary test setup may be based on and / or dependent on environment data.

[0152] The primary test settings may be based on and / or dependent on one or more environment probabilities, including a first environment probability ENVP_1 indicating that the current environment is the first environment and / or a second environment probability ENVP_2 indicating that the current environment is the second environment. The primary test settings may be a weighted combination of test settings for environments ENV_k, k=1, 2, ..., K (with the environment probabilities as weights ENVP_k).

[0153] The fitting agent is configured to obtain the secondary test settings of the hearing device, also denoted as θalt, x_alt or x alt The secondary test setting x_ref is a vector comprising M hearing device parameters of the hearing device. The hearing device parameters may include one or more of filter coefficients, compressor settings, gains or other parameters related to the operation or signal processing of the hearing device. The secondary test setting may be based on and / or dependent on the current environment, such as based on the first initial environment probability and / or the second initial environment probability. In other words, the secondary test setting may be based on and / or dependent on the environment data.

[0154] The secondary test settings may be based on and / or dependent on one or more environment probabilities, including a first environment probability ENVP_1 indicating that the current environment is the first environment and / or a second environment probability ENVP_2 indicating that the current environment is the second environment. The secondary test settings may be a weighted combination of test settings for environments ENV_k, k=1, 2, ..., K (with the environment probabilities as weights ENVP_k).

[0155] A fitting agent, such as one or more processors of a fitting agent, is configured to present a test setup to a hearing device user. Presenting the test setup to the hearing device user includes presenting a primary test setup and / or a secondary test setup to the user. Presenting the primary test setup and the secondary test setup to the user optionally includes outputting a primary test signal according to the primary test setup. The primary test signal may be an audio signal. The primary test signal may be output via a speaker or a receiver of the hearing device. Presenting the primary test setup and the secondary test setup to the user optionally includes generating a primary test signal according to the primary test setup in an accessory device, and streaming the primary test signal from the accessory device to the hearing device. Presenting the primary test setup and the secondary test setup to the user optionally includes transmitting a control signal indicating the primary test signal / primary test setup from the accessory device to the hearing device. The control signal may include the primary test setup. Presenting the primary test setup and the secondary test setup to the user may include generating a primary test signal according to a control signal in the hearing device, such as a primary test setup based on the control signal.

[0156] The fitting agent, such as one or more processors of the fitting agent, is configured to obtain (such as receive and / or detect) user input indicating a preference for a primary test setup or a secondary test setup. In other words, the user gives a binary response r to indicate his / her preferred HA parameter set among the two suggestions θref and θalt.

[0157] A non-preferred test setting is a primary test setting or a secondary test setting that is not selected as a preferred test setting. Therefore, a hearing device and / or (one or more) accessory devices that implement a fitting agent or at least a part of a fitting agent may include one or more user interfaces for obtaining (such as receiving and / or detecting) user input. For example, a hearing device may include a user interface to receive user input. The user interface of the hearing device may include one or more buttons, an accelerometer and / or a voice control unit. The accessory device may include a user interface. The user interface of the accessory device may include a touch-sensitive surface (e.g., a touch display) and / or one or more buttons. The user interface of the accessory device may include a voice control unit. The user interface of the hearing device may include one or more physical sliders, knobs and / or buttons. The user interface of the accessory device may include one or more physical or virtual (on-screen) sliders, knobs and / or buttons.

[0158] The fitting agent is configured to obtain (such as detect) user input of a preferred / selected test setting indicating a preference for a primary test setting or a secondary test setting. In the fitting agent, detecting the user input of the preferred test setting indicating a preference for the primary test setting or the secondary test setting may include prompting the user for the user input, for example by a beep signal or a voice signal from the hearing device and / or a visual, tactile and / or audio prompt from an accessory device. Detecting the user input may be performed on the hearing device, for example by the user activating a button and / or an accelerometer in the hearing device (e.g., a single or double tap on the hearing device housing). Detecting the user input may be performed on the accessory device, for example by the user selecting a user interface element representing the preferred test setting, for example on a touch-sensitive display of the user's accessory device.

[0159] A fitting agent, such as one or more processors of a fitting agent, is configured to update the user model, for example based on preferred test settings and environmental data to provide an updated user model.

[0160] The fitting agent, such as one or more processors of the fitting agent, may be configured to determine hearing device parameters based on the updated user model and output the hearing device parameters, such as transmit the hearing device parameters to the hearing device.

[0161] The fitting agent may be implemented in one or more electronic devices such as a hearing device and / or (one or more) accessory devices. The fitting agent is an electronic fitting agent. The fitting agent or at least a portion thereof may be implemented in an accessory device or multiple accessory devices, such as in a smartphone optionally combined with a smartwatch. The fitting agent may be used for / part of one or more of optimization, determination, fitting, tuning and modeling, such as determining hearing device parameters of a hearing device. Implementing the fitting agent in (one or more) accessory devices (such as a smartphone optionally combined with a smartwatch) may help provide smoother user input and user experience. In addition, from the perspective of the hearing device, implementing a portion of the fitting agent in (one or more) accessory devices may help provide a more energy-efficient approach. The fitting agent or at least a portion thereof may be implemented in a server device and / or a fitting device.

[0162] The present disclosure relates to a fitting agent for a hearing device, and in particular to a fitting agent for one or more of optimizing, determining, fitting and tuning hearing device parameters of a hearing device.

[0163] In one or more examples, a fitting agent for a hearing device system is disclosed, the hearing device system including a hearing device worn by a hearing device user, wherein the fitting agent includes one or more processors configured to initialize a user model and an environment model, the user model including one or more user preference functions and associated user response distributions; obtain environment data indicating a current environment; determine a first initial environment probability for a first environment and a second initial environment probability for a second environment based on the environment data and the environment model, obtain a test setting including a primary test setting and a secondary test setting for the hearing device based on the first initial environment probability and the second initial environment probability; present the test setting to the hearing device user; obtain user input of a preferred test setting indicating a preference for the primary test setting or the secondary test setting; and update the user model based on the preferred test setting and the environment data to provide an updated user model. The updated user model can be based on the trial / test setting.

[0164] In one or more examples, the one or more processors are configured to estimate a personalized environment probability for a current environment based on an updated user model.

[0165] In one or more examples, the personalized environment probability is a probability of one or more user preference clusters in the domain of hearing device parameters. In other words, the personalized environment probability can include one or more probabilities, each of which is a posterior probability of a user preference cluster. We also refer to these probabilities as reliabilities or soft information.

[0166] In one or more examples, one or more processors are configured to update the environment model to provide an updated environment model, for example based on the personalized environment probability and the environment data. The updating of the environment model can be performed before, simultaneously with, and / or after updating the user model to provide an updated environment model. Updating the environment model can include outputting, such as transmitting and / or storing the updated environment model. The updated environment model can be transmitted to a server device and stored therein, for example for later use and updating to the environment model. The updated environment model can be stored in an accessory device, for example for later use and updating to the environment model, for example, stored in a memory of the accessory device.

[0167] In one or more examples, updating the environment model includes updating an environment distribution function of the environment model based on the personalized environment probability and the environment data. In other words, the environment model may include the environment distribution function.

[0168] In one or more examples, the environment model is a Gaussian mixture model, also denoted GMM. The environment model may be a non-parametric Bayesian mixture model, such as a Dirichlet process Gaussian mixture model, also denoted DP-GMM.

[0169] For example, given a signal vector st Representing the environmental data and the GMM model of the environment, the reliability information about the current state of the environment can be obtained as follows:

[0170]

[0171] in represents the state of the environment, μ s，k and∑ s，k represents the parameters (mean and covariance) of the signal distribution of the kth cluster, and π t，k represents the environmental mixing probability.

[0172] In this disclosure, c t 、m k , W k Represents the state of the environment and the parameters of the signal distribution. Here W k is the precision matrix, which corresponds to the inverse of the covariance matrix. The purpose of the notation will be clear from the scenario. Index 0 of a parameter also indicates a hyperparameter.

[0173] Depending on the scenario, when we refer to reliability or soft information, we refer to the prior or posterior probabilities of clusters in the signal / environment domain or the user preference / parameter domain. The prior probabilities in the user preference domain are the posterior probabilities from the signal / environment domain; the posterior information in the user preference / parameter domain is the personalized environment probability.

[0174] This reliability / reliability information is used to map from the environment space to the parameter space, where the user preferences depend on the environment. In the parameter space, this reliability information is treated as a priori information about the best preference cluster. In other words, this reliability information is the mixing coefficient of the GMM corresponding to the parameter space.

[0175] After user-agent interaction during trials in this environment, i.e., after presenting test setups and obtaining user input on preferred test setups, the fitting agent updates the user model parameter posterior given the kth environment, e.g., using hypothesis density filtering (ADF). Since the approximation family is Gaussian, the full parameter posterior is also a GMM.

[0176] Therefore, based on the approximate parameter posterior q t (φ t |z t ), the posterior of the preference cluster of the optimal parameter settings obtained after the agent-user interaction is It can be obtained as follows:

[0177]

[0178] One or more processors can use environmental data from the parameter space (the observed signal s t) and reliability information The environment model is updated based on the personalized environment probabilities and the environment data to provide an updated environment model. Thus, a mapping in the reverse direction from the user preference parameter space to the environment space is obtained.

[0179] In one or more examples, updating the environment model is based on an expectation maximization algorithm, such as an online EM. In other words, an online or recursive EM can be used to update the environment models of the various components, for example:

[0180]

[0181] where 0<ν<1 is the forgetting factor.

[0182] Other variations of online EM are possible. For example, one or more EMs with time-dependent forgetting factors and / or regularization can be implemented.

[0183] Note that with some modifications, for example by introducing a Dirichlet prior on the number of clusters, i.e., K,π t+1，k , can also be used to remove a cluster.

[0184] In one or more examples, obtaining the environmental data includes obtaining location data, such as from a GPS unit of the accessory device, the location data indicating a user location, and determining the environmental data based on the location data.

[0185] In one or more examples, obtaining the environmental data includes obtaining audio data indicating sounds in a current environment, and determining the environmental data based on the audio data.

[0186] In one or more examples, obtaining environmental data includes obtaining scene data indicating the surrounding environment and / or activity of the user, and determining the environmental data based on the scene data. The scene data may include one or more of calendar data, activity data, or motion data, and one or more processors may be configured to determine the environmental data based on one or more of the calendar data, activity data, and motion data. The activity data may indicate the user's activities, such as watching TV, attending meetings, listening to concerts, eating dinner, etc. The scene data may indicate the scene in which the user is located, such as indicating the user's orientation, position, movement, temperature, pulse, or other data related to the environment. For example, the scene data may include orientation data, such as GPS coordinates, and / or movement data, such as accelerometer data. The scene data may include calendar data, and the environmental data may be based on the calendar data. The scene data may include, for example, sensor data from one or more sensors of an accessory device and / or from one or more sensors of a hearing device. The scene data may include hearing device data transmitted from a hearing device, such as one or more program identifiers, one or more operating parameters, and / or one or more operating mode identifiers of a hearing device.

[0187] In one or more example fitting agents, obtaining environment data includes, for example, receiving user input indicating an environment via a user interface of the accessory device. Thus, the user can select and indicate a current environment via the accessory device, for example, from a list of environments presented on a touch display of the accessory device.

[0188] In one or more examples, initializing the environment model includes obtaining a user profile and initializing the environment model based on the user profile.

[0189] In one or more examples, the user profile includes one or more of age, gender, hearing loss level, and activity level, and wherein initializing the environment model is based on one or more of age, gender, hearing loss level, and activity level.

[0190] In one or more examples, initializing the environment model includes obtaining an environment profile and initializing the environment model based on the environment profile. Obtaining the environment profile may include selecting, determining, and / or retrieving a default or initial environment. Obtaining the environment profile may include determining the environment profile based on audio data from one or more microphones and / or user input (such as via a questionnaire).

[0191] The fitting agent may be configured to use one or more of, implement, store, infer, and employ the updated user model as the user model in the fitting agent.

[0192] The fitting agent may be configured to one or more of use, implement, store, infer, and employ the updated environment model as the environment model in the fitting agent.

[0193] The present disclosure provides a scene / environment-aware preference learning system that can learn user preferences in different environments. In Section A below, an example preference learning system for a single environment is disclosed. In Section B, an example preference learning system for multiple environments is disclosed. Section C describes an example environment model, and Section D provides a model summary.

[0194] A. Preference Learning System

[0195] The preference learning system is modeled as two subsystems interacting with each other, namely the fitting agent and the user. The parameters of the fitting agent represent its internal state. In one interaction cycle, the fitting agent proposes trials to the user based on its internal state and the interaction history. Each trial, represented by {θref, θalt}, contains a so-called reference and an alternative suggestion. The user provides a binary response r to indicate the set of hearing device parameters he / she prefers between the two suggestions. More precisely, the user's response to a trial {θref, θalt} can be defined as

[0196]

[0197] where θref>θalt means that the user prefers the reference suggestion over the alternative suggestion; and where R denotes a random variable corresponding to the user response and r denotes its realization. By interacting with the user, the fitting agent (also denoted as agent) aims to estimate the user's optimal preference given the user's response in as few trials as possible. In order to define the user response model and define an efficient interaction, two assumptions about the user's preferences are made.

[0198] First, there is an optimal user preference and it is described by a unimodal preference function f(x; θ, Λ) given by

[0199]

[0200] where x represents the hearing device tuning parameters, θ∈[0,1] D denotes the optimal tuning parameters corresponding to the user's preference, and Λ = diag([λ1, ..., λ D ]), λ d ∈R + , d = 1, 2, ..., D represents the user's sensitivity to parameter changes.

[0201] Note that for a hearing device also denoted as a HA with D tuning parameters, usually different parameters have different scales, so in order to ensure that θ∈[0,1] D , each parameter is normalized to [0, 1]. Second, the user can be uncertain about his / her preference for one of the two recommendations, which is represented by an additive random variable ε ≈ N(0, 1).

[0202] The user response model is then given by the binomial probit model:

[0203] P(r|θ ref ,θ alt ,θ,Λ)

[0204] = Ber(r|Φ(f(θ alt ;θ,Λ)-f(θ ref ; θ, Λ))), where Φ(·) is the cumulative distribution function (CDF) of the standard normal distribution.

[0205] The true values ​​of the parameters θ and Λ of the user preference function (2) are unknown. To learn these parameters, the agent proposes to conduct experiments and infer θ and Λ based on the information obtained from the user.

[0206] Observe that the tuning and sensitivity parameters θ, Λ are bounded. For modeling convenience, this constraint is relaxed by introducing the following transformation:

[0207] φ=[Φ -1 (θ), lnv] T ,

[0208] where Φ -1 (·) is the standard normal CDF and This transformation ensures that the user preferences remain in the hypercube [0, 1] during the learning process. D , while the user sensitivity remains in the positive real half-space.

[0209] The agent has two components. One component corresponds to the internal state given previous observations and trials. The generation distribution P of the trajectory under t a The other component is the predicted user response distribution P used by the agent to model user responses. t u In summary, the preference learning system model consists of two subsystems {P t a , P t u} That is, the combination of agent and user is defined:

[0210]

[0211] For t = 1, 2, ..., where P t a Given a series of previous responses r 1:t-1 =(r1,...,rt-1 ) and test {θref, θalt} 1:t-1 Internal state The generated proposal distribution under P t u is given r and {θref, θalt} 1:t-1 The predicted PMF of the user's response to a new trial {θref, θalt} under this internal state.

[0212] B. State-Dependent Preference Learning Systems

[0213] To extend the preference learning system to be able to learn multiple user preferences, we assume that user preferences depend on the acoustic environment. We use environment states to represent different acoustic environments. Using the environment states, we define the state-dependent user preference function as

[0214] where θk and Λk are the parameters corresponding to states k and θ, and Λ is the set of all parameters corresponding to all states. Therefore, at time t, the state-dependent user response model becomes

[0215] where c t ∈N is the state of the environment at time t, and θref, θalt are the reference and alternative suggestions for this state. Using these two formulas, the state-dependent preference learning system can be expressed as

[0216]

[0217] in It is all states A set of k = 1, 2, ..., and To complete the model, assume The prior normal distribution of

[0218] C. Environmental Model

[0219] The HA tuning parameters are used to enhance the acoustic signals of the users to provide them with good sound quality. From this perspective, the state of the environment should be inferred from the acoustic signals generated in this environment. Therefore, an acoustic signal model is implemented to supply environmental data to the preference learning system. The inferred environmental state serves as state c in section B above. t .

[0220] The HA will record the acoustic signal from the environment and then convert the signal into a signal feature s at time step t. tThe environment model can be defined in this feature space. In addition, a correspondence between the signal feature space and the user preferences is assumed, i.e., for each cluster in the feature space, there is a corresponding user preference. Therefore, the environment state c t Shared by the acoustic signal model and the agent.

[0221] Use Gaussian mixture model (GMM) to characterize the signal t In this model, each component represents a signal cluster, and each k-th component is modeled by a Gaussian distribution with mean vector m k and the precision matrix W k , to represent different signal clusters. Therefore, GMM can be defined as

[0222]

[0223] where m=[m1,m2,...,m ∞ ] and W=[W1,...,W ∞ ]. For all parameters m of the mixture components k and W k , k = 1, 2, …, K, assuming the same Normal-Wiechter distribution as prior:

[0224]

[0225] where m0, λ0, v0, W0 are the four parameters of the Normal-Wiechter distribution. Observe that GMM has a predefined number of clusters. However, the number of relevant clusters for a user is typically unknown and can change over time. To address this issue, we propose to use DP as a prior for the mixture components. Given a mixture probability vector π, the cluster labels c t Follows categorical distribution

[0226] p(c t |π)=Cat(c t |π).

[0227] This mixed probability π describes the probability of each state of the environment and follows the GEM distribution

[0228] p(π) = GEM(π|α), where GEM distribution is the limiting distribution of Dirichlet distribution

[0229]

[0230] Named after Griths, Engen, and McCloskey. Its sample contains an infinite number of elements. Here α is the concentration parameter.

[0231] D. Model Summary

[0232] A preference learning system for perceiving a scene can be defined as outlined in Sections B and C, optionally with a t The DP prior for is defined together with the priors for the signal model and the agent. In summary, the preference learning system for perceiving scenes can be defined as the following generative model:

[0233] π~GEM(α)

[0234] c t |π~Cat(π)

[0235]

[0236] Note that the description and characteristics of the fitting agent functionality also apply to the method and vice versa.

[0237] Figure 1 Schematic diagram of a fitting agent with a preference learning system for perceiving a scene. A signal s is received by a hearing device 2 and a fitting agent 27, e.g. implemented in an accessory device, which passes the environment data to an environment model 50. The environment model 50 infers the current environment state c for a preference learning system 52. Based on the environment 41 / state as well as previous trials and user responses, the preference learning system 52 including the user model proposes two sets of tuning parameters θ (first primary test setting) and θ′ (first secondary test setting) for the hearing device. The hearing device 2 processes the signals observed by the microphone(s) of the hearing device 2 using these tuning parameters to generate audio signals y and y′, respectively, which are then presented to a user 40 for evaluation via a receiver of the hearing device. The user 40 indicates a preference for the first primary test setting or the first secondary test setting via a user input represented by r.

[0238] Figure 2 The present invention provides an overview of a hearing system with a fitting agent according to the present disclosure. The hearing system comprises a hearing device 2, an accessory device 4 and an optional server device / fitting device 5. The hearing device 2 comprises a server device for communicating with the accessory device 4 and an optional contralateral hearing device ( Figure 21 and 12. The hearing device 2 includes a transceiver module 6 for (wireless) communication (not shown). The transceiver module 6 comprises an antenna 8 and a transceiver 10 and is configured to receive and / or transmit wireless signals to the accessory device 4 via a wireless connection 11. The accessory device 4 is configured to receive and / or transmit wireless signals to the server device / fitting device 5 via a wireless connection 11A. The hearing device 2 comprises a set of one or more microphones, including a first microphone 12 for providing a first microphone input signal 14; a processor 16 for processing an input signal including the first microphone input signal 14 according to one or more hearing device parameters and providing an electrical output signal 18 based on the input signal; an optional user interface 20, which is connected to the processor 16; and a receiver 22 for converting the electrical output signal 18 into an audio output signal.

[0239] The accessory device 4 is a smartphone and includes a user interface 24 including a touch display 26, a processor (not shown) and a memory (not shown).

[0240] In the hearing device system 1 , the fitting agent 27 is an application installed in the memory of the accessory device 4 .

[0241] The fitting agent 27 is a fitting agent for updating a user model and / or an environment model of a hearing device user and / or performing one or more of optimization, determination, fitting, tuning and modeling of hearing device parameters of the hearing device.

[0242] The fitting agent 27 includes one or more processors, such as one or more processors of the accessory device and the hearing device, which are configured to initialize a user model and an environment model, the user model including a plurality of user preference functions and associated user response distributions, wherein each user preference function f(x; θ k ,Λ k ) is associated with a kth environment (k=1, 2, ..., K); obtain environmental data indicating the current environment; determine a first initial environmental probability for a first environment and a second initial environmental probability for a second environment using the environment and based on the environmental data and an environmental model, and obtain a test setup including a primary test setup and a secondary test setup for a hearing device based on the first initial environmental probability and the second initial environmental probability; present the test setup to a hearing device user, for example via a wireless connection 11; obtain / detect a user input r of a preferred test setup indicating a preference for a primary test setup x_ref or a secondary test setup x_alt; and update a user model and / or an environmental model, such as a user preference function f(x; θ, Λ) and / or a user response model (user response distribution and parameter distribution), for example based on hearing device parameters of the preferred test setup (such as one or more, for example all, of r, x_ref and x_alt) and the environmental data.

[0243] The fitting agent 27 implemented in the accessory device 4 is configured to update a user model and / or an environment model, such as a plurality of user preference functions of the user model, based on the primary test settings, the secondary test settings, the preferred test settings, and the environment data (such as the initial environment probability of the environment). In other words, the fitting agent is configured to update the user model based on the hearing device parameters of the preferred test settings and the environment data. Updating the user model may include updating a user preference model and a user response model based on the primary test settings, the secondary test settings, the preferred test settings, and the environment data. The fitting agent 27 / accessory device 4 may be configured to transmit the primary test settings, the secondary test settings, and the preferred test settings to a server device 5, which updates the user model and transmits the updated user model to the fitting agent 27 / accessory device 4. Thus, the fitting agent 27 / accessory device 4 may be configured to receive the updated model from the server device 5. In other words, the fitting agent 27 may be distributed on the accessory device 4 and one or more of the hearing device 2 and the server device 5. Hence, the fitting agent 27 may comprise a first part 27A implemented in the accessory device 4 , an optional second part 27B implemented in the server device and an optional third part 27C implemented in the hearing device 2 , such as in the processor 16 .

[0244] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, updating the user model includes determining a posteriori the parameters of the user preference function, for example based on previous parameter posteriori (such as the last or current parameter posteriori), preferred test settings, and non-preferred test settings. Determining a posteriori the parameters of the user preference function may include determining or updating θ k is the kth maximization variable of f and / or determines or updates Λ k Indicating or characterizing the user's sensitivity to changes in hearing device parameters in the k-th environment. Determining the parameter posterior optionally comprises applying a sequential estimation in the fitting agent 27, 27A, 27B, 27C. Thus, the fitting agent 27, 27A, 27B, 27C is optionally configured to determine the parameter posterior based only on the previous user model and (r, x_ref, x_alt, s).

[0245] The one or more processors of the fitting agent 27 are configured to estimate a personalized environment probability of the current environment based on the updated user model, wherein the personalized environment probability is a probability of one or more user preference clusters in the domain of hearing device parameters, and the one or more processors are configured to update the environment model based on the personalized environment probability and the environment data to provide an updated environment model, as described above. The updating of the environment model for providing the updated environment model is performed before, simultaneously with, and / or after the updating of the user model.

[0246] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, obtaining the environmental data includes obtaining audio data and determining the environmental data based on the audio data. For example, the microphone 12 of the hearing device 2 may provide audio data representative of the sound received by the microphone 12, wherein the audio data may be transmitted to the accessory device 4, wherein the fitting agent 27, 27A of the accessory device 4 is configured to determine the environmental data based on the audio data from the hearing device, and / or wherein the fitting agent 27C of the hearing device 2 is configured to determine the environmental data or at least a portion of the environmental data based on the audio data from the microphone 12. The environmental data determined by the fitting agent 27C of the hearing device 2 is transmitted to the fitting agent 27A of the accessory device 4 for further processing.

[0247] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, obtaining the environment data includes, for example, obtaining scene data from one or more sensors of the hearing device 2 and / or the accessory device 4, and determining the environment data based on the scene data. Determining the environment data may be based on the audio data and the scene data. In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, obtaining the scene data includes obtaining the scene data from one or more applications (such as a calendar application) of the accessory device 4.

[0248] In one or more example fitting agents including the fitting agents 27, 27A, 27B, 27C, the environment data includes a first environment probability of a first environment and / or a second environment probability of a second environment, and wherein updating the user model is based on the first probability and / or the second probability.

[0249] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, the plurality of user preference functions includes K user preference functions, where K is greater than 3, such as 5, 6, 7, 8 or more, and where updating the user model includes updating the posteriori of parameters of the K user preference functions (or a subset thereof) based on a test setup, user input of a preferred test setup, and optionally one or more of environmental data and environmental probabilities.

[0250] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, the fitting agents are configured to use a Gaussian mixture model to obtain information such as determining one or more environment probabilities, such as environment probabilities and / or environment identifiers for each environment ENV_k, k=1, 2, ..., K, and wherein updating the user model is optionally based on the environment identifiers and / or environment probabilities.

[0251] In one or more example fitting agents including fitting agents 27, 27A, 27B, 27C, the fitting agent is configured to determine whether a test criterion based on environmental data is met. The fitting agent 27, 27A, 27B, 27C may be configured to perform, in accordance with the determination that the test criterion is met: obtaining a test setup including a primary test setup and a secondary test setup for a hearing device; presenting the test setup to a hearing device user; obtaining user input of a preferred test setup indicating a preference for the primary test setup or the secondary test setup; and updating a user model based on hearing device parameters of the preferred test setup and the environmental data. The test criterion may be based on an uncertainty measure indicating uncertainty of a user preference in a current environment.

[0252] In one or more example fitting agents including the fitting agent 27 , 27A, 27B, 27C, obtaining the test settings including the primary test settings and the secondary test settings of the hearing device includes determining the secondary test settings based on the environmental data.

[0253] In one or more example fitting agents including the fitting agent 27 , 27A, 27B, 27C, the fitting agent is configured to determine and update hearing device parameters of the hearing device based on the updated user model and optionally environmental data.

[0254] In implementations including the accessory device 4 , the fitting agent 27 / accessory device 4 may be configured to send a control signal 30 to the hearing device 2 , the control signal 30 indicating the primary test settings and the secondary test settings, thereby enabling the hearing device 2 to output a test signal accordingly.

[0255] The hearing device 2 (processor 16 ) is optionally configured to output a primary test signal according to the primary test setup via the receiver 22 , and to output a secondary test signal according to the secondary test setup via the receiver 22 .

[0256] The fitting agent 27, 27A, 27C (the hearing device 2 (processor 16) and / or the accessory device 4) is configured to detect a user input of a preferred test setting indicating a preference for a primary test setting or a secondary test setting, for example by detecting a user input on the user interface 20 or by detecting a user selection of one of a primary virtual button 32 (primary test setting is preferred) and a secondary virtual button 34 (secondary test setting is preferred) on the user interface 24 of the accessory device 4.

[0257] It should be noted that the fitting agent 27, 27A, 27B, 27C may be configured to detect a user input of a preferred test setting indicating a preference for the primary test setting or the secondary test setting, for example by receiving a wireless input signal from a secondary accessory device such as a smart watch comprising a user interface. Thus, a more convenient user input is provided, which in turn increases the user-friendliness of the fitting agent.

[0258] The user may initiate the updating of the user model and / or environment model by pressing the start button 28 on the user interface 24 of the accessory device. In other words, the fitting agent may detect a user input indicating a start and execute the method / update of the user model based on the detection of the user input indicating a start.

[0259] In an implementation comprising an accessory device 4 , the accessory device 4 may be configured to send a control signal 38 to the hearing device 2 indicating hearing device parameters of a preferred test setup, thereby enabling the hearing device to update and apply the preferred hearing device parameters in the hearing device.

[0260] Figure 3 The interaction between the user 40 and the fitting agent 27 in the environment 41 is illustrated. The fitting agent 27 obtains environment data ENV_D indicative of the current environment 41, for example based on audio data and / or scene data from one or more hearing devices worn by the user 40, from the user providing user input to the fitting agent, from accessory device microphones, sensors and / or applications.

[0261] For each interaction or trial indexed by index n, the fitting agent generates a primary test set and secondary test setup And presenting 42 test settings to the user and For example, by controlling the hearing device to output a primary test signal and a secondary test signal that are respectively in accordance with and indicative of the primary test setup and the secondary test setup. The user evaluates the two test setups and and the fitting agent 27 receives and detects the user's response 44, r n Indicates the preferred test settings for the primary test settings and the secondary test settings. The fitting agent determines the preferred test settings based on the environmental data, n , and The fitting agent determines the primary test setting for the n+1th trial based on the preferred test setting for the nth trial and the environmental probability for the n+1th trial. And generate the n+1th trial Determine the secondary test setup And present the user with 46 test settings and

[0262] Figure 4is a flowchart of an exemplary method according to the present disclosure. Method 100 is a method of updating a user model and / or an environment model of a hearing device user and / or optimizing, determining, fitting, tuning and modeling hearing device parameters of a hearing device (such as determining), wherein the method comprises initializing S102 an environment model and a user model, the user model comprising a plurality of user preference functions and associated user response distributions, wherein each user preference function is associated with an environment; obtaining S104 environmental data indicating a current environment; obtaining S106 a test setting comprising a primary test setting S106A and a secondary test setting S106B of a hearing device; presenting S108 the test setting to a hearing device user, comprising outputting S108A a primary test signal according to or based on the primary test setting and outputting S108B a secondary test signal according to or based on the secondary test setting; obtaining S110 a user input of a preferred test setting indicating a preference for the primary test setting or the secondary test setting; and updating S112 the user model and / or the environment model based on the hearing device parameters of the preferred test setting and the non-preferred test setting and the environment data. Updating S112 the user model optionally includes updating S112A a plurality of user preference functions of the user model, such as a subset or all of the user preference functions. Updating S112 the user model optionally includes updating S112B a plurality (such as all) of predictors / user response models of the user model.

[0263] In method 100 , updating S112 the user model and / or the environment model may include determining S112C a posteriori of parameters of a user preference function of the user model based on previous or latest parameter posteriori, preferred test settings, non-preferred test settings, and environment data.

[0264] The method 100 optionally comprises determining S114 whether a stopping criterion is met, and updating S116 hearing device parameters of the hearing device according to the updated model, e.g. as described above with respect to the fitting agent. If it is determined that the user model has not been sufficiently updated (stopping criterion is not met), i.e. further trials and user input to a test setup are desired or required for accurate and precise modeling of the environment and / or user preference functions of the user model, the method 100 optionally proceeds to S104.

[0265] The use of the terms "first," "second," "third," and "fourth," "primary," "secondary," "tertiary," etc. does not imply any particular order, but are included to identify the various elements. Furthermore, the use of the terms "first," "second," "third," and "fourth," "primary," "secondary," "tertiary," etc. does not indicate any order or importance, but rather the terms "first," "second," "third," and "fourth," "primary," "secondary," "tertiary," etc. are used to distinguish one element from another. Note that the words "first," "second," "third," and "fourth," "primary," "secondary," "tertiary," etc. are used herein and elsewhere for labeling purposes only and are not intended to indicate any particular spatial or temporal order.

[0266] The memory may be one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, a random access memory (RAM), or other suitable devices. In a typical arrangement, the memory may include a non-volatile memory for long-term data storage and a volatile memory used as a system memory for the processor. The memory may exchange data with the processor via a data bus. The memory may be considered a non-transitory computer-readable medium.

[0267] The memory may be configured to store information in a portion of the memory.

[0268] Furthermore, the marking of a first element does not imply the presence of a second element and vice versa.

[0269] It will be appreciated that the accompanying drawings include some modules or operations illustrated with solid lines and some modules or operations illustrated with dotted lines. The modules or operations included in the solid lines are the modules or operations included in the most extensive example embodiments. The modules or operations included in the dotted lines are the following example embodiments, which may be included in the modules or operations of the solid line example embodiments, or are a part thereof, or are other modules or operations that may be adopted in addition thereto. It will be appreciated that these operations do not need to be performed in the order presented. In addition, it will be appreciated that not all operations need to be performed. The exemplary operations may be performed in any order and in any combination.

[0270] It should be noted that the word "comprising" does not necessarily exclude the presence of other elements or steps than those listed.

[0271] It should be noted that the word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

[0272] It should also be noted that any reference signs do not limit the scope of the claims, that exemplary embodiments may be implemented at least in part by both hardware and software, and that multiple "means," "units," or "devices" may be represented by the same item of hardware.

[0273] The various exemplary methods, devices, and systems described in the present disclosure are described in the general context of method steps, which in one aspect can be implemented by a computer program product embodied in a computer-readable medium, including computer executable instructions, such as program code, executed by a computer in a network environment. Computer-readable media may include removable and non-removable storage devices, including but not limited to read-only memory (ROM), random access memory (RAM), compact disk (CD), digital versatile disk (DVD), etc. In general, program modules may include routines, programs, objects, components, data structures, etc. that perform specified tasks or implement specific abstract data types. Computer executable instructions, associated data structures, and program modules represent examples of program codes for performing the steps of the methods disclosed herein. A specific sequence of such executable instructions or associated data structures represents an example of corresponding actions for implementing the functions described in such steps or processes.

[0274] Although features have been shown and described, it should be understood that they are not intended to limit the claimed invention, and it will be apparent to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the claimed invention. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense. The claimed invention is intended to cover all alternatives, modifications, and equivalents.

Claims

1. A fitting agent for a hearing device system, the hearing device system comprising a hearing device worn by a hearing device user, wherein the fitting agent comprises one or more processors, the processors being configured to: Initializing a user model and an environment model, wherein the user model includes a plurality of user preference functions and associated user response distributions; Obtaining environmental data indicating the current environment; determining a first initial environment probability for a first environment and a second initial environment probability for a second environment based on the environment data and the environment model, obtaining a test setup comprising a primary test setup and a secondary test setup for the hearing device based on the first initial environment probability and the second initial environment probability; presenting the test setup to the hearing device user; obtaining user input of a preferred test setup indicating a preference for the primary test setup or the secondary test setup; as well as The user model is updated based on the preferred test settings and the environmental data to provide an updated user model.

2. The fitting agent of claim 1, wherein the one or more processors are configured to estimate a personalized environment probability for the current environment based on the updated user model. 3 . The fitting agent of claim 2 , wherein the personalized environment probability is a probability of one or more user preference clusters in a domain of hearing device parameters.

4. The fitting agent according to any one of claims 2-3, wherein the one or more processors are configured to update the environment model based on the personalized environment probability and the environment data to provide an updated environment model. 5 . The fitting agent of claim 4 , wherein updating the environment model comprises updating an environment distribution function based on the personalized environment probability and the environment data.

6. The fitting agent according to any one of claims 1-5, wherein the environment model is a Gaussian mixture model.

7. The fitting agent of claim 6, wherein updating the environment model is based on an expectation-maximization algorithm.

8. The fitting agent according to any one of claims 1 to 7, wherein obtaining environmental data comprises: Position data indicative of a user's position is obtained and the environmental data is determined based on the position data.

9. The fitting agent according to any one of claims 1 to 8, wherein obtaining environmental data comprises: Audio data indicative of sounds in the current environment is acquired and the environment data is determined based on the audio data.

10. The fitting agent according to any one of claims 1 to 9, wherein obtaining environmental data comprises: Context data indicative of a surrounding environment and / or an activity of the user is acquired and the environmental data is determined based on the context data.

11. The fitting agent according to any one of claims 1 to 10, wherein initializing the environment model comprises: A user profile is obtained and the environment model is initialized based on the user profile.

12. The fitting agent of claim 11, wherein the user profile comprises one or more of age, gender, hearing loss level, and activity level, and wherein initializing the environment model is based on one or more of age, gender, hearing loss level, and activity level.

13. The fitting agent according to any one of claims 1 to 12, wherein initializing the environment model comprises: An environmental profile is obtained and the environmental model is initialized based on the environmental profile.

Images