Method and system for authentication and compensation

Abstract

The present disclosure describes a method and system for authenticating and dynamically compensating for headphones. The method performs authentication on a user based on a headphone transfer function (HPTF) when the user wears the headphones. The method can also detect whether there is a frequency response deviation between the user's HPTF and the adjusted HPTF. In addition, if there is a frequency response deviation between the user's HPTF and the adjusted HPTF, the method can dynamically compensate for the user's HPTF based on the detected frequency response deviation.

Description

Technical Field

[0001] The present disclosure relates to a method and system for authentication and compensation, and in particular to a method and system for performing biometric authentication and dynamic compensation for headphones based on a headphone transfer function (HPTF). Background Art

[0002] Biometric authentication is used to achieve a seamless user experience for edge devices while providing device security (such as mobile phones and laptops). In order to achieve a better user experience, various technologies have been invented to reduce the intention-to-action time. This intention-to-action time is defined by the moment when the user wants the target device to perform an action to the moment when the edge device completes the execution. Modern recognition technologies (such as image and voice recognition technologies) were then developed to reduce the intention-to-action time. Recent advances in edge computing combined with cloud services have greatly improved the quality of life.

[0003] Facial recognition is based on having a camera mounted on the target device and is mostly achieved by using neural network related techniques to compare pre-registered facial features. Various techniques such as IR-based depth sensors and stereo imaging are then used to improve visual accuracy. These methods are mainly used to prevent malicious people from compromising the system by showing photos of the target. However, these systems tend to be more expensive in terms of power consumption and sensor costs. In addition, in the past two years, mobile devices are trying to get rid of having image sensors on the front to achieve a higher screen-to-body ratio.

[0004] Speech recognition is based on having a microphone capture acoustic input, which is then analyzed to match the real-time streaming of pre-registered commands. Since recognition accuracy is largely coupled to SNR, well-known algorithms such as multiple microphones and noise reduction are used to improve accuracy. Multi-channel and noise reduction techniques are also expensive in terms of power consumption and sensor cost. In addition, speech recognition requires the user to speak keywords, which can sometimes be inconvenient in public places.

[0005] In order to overcome the above inconveniences and the disadvantages of high power consumption and sensor cost, it is necessary to provide an improved authentication technology.

[0006] Furthermore, typically in many headphones, HPTF is measured by using a special ear simulator on a dummy human head. Acoustic engineers adjust the frequency response of the headphones based on the measured HPTF. However, due to individual differences, the HPTF measured by the ear simulator may not be satisfactory. When an end user buys a headphone and listens to music, what he hears may not be the expected sound that the acoustic engineer has tuned. Different listeners hear different sounds in one headphone, even if they wear the headphone perfectly. In addition, even if the headphone has good bass performance, the listener may still hear little bass if the headphone is not worn correctly because there is a lot of air leakage between the headphone and the listener's ear.

[0007] The listener's personal HPTF may involve reflections between the inner surface of the earphone and the eardrum that differ from the measured HPTF, or simply be due to some unwanted air leakage, which may introduce some timbre distortion.

[0008] In order to faithfully play back sound to different listeners through headphones, the HPTF needs to be calibrated and compensated. Therefore, it is necessary to provide an improved technology for adaptively and effectively performing real-time calibration when using headphones after certification. Summary of the invention

[0009] According to one aspect of the present disclosure, a method for authenticating and dynamically compensating for headphones is provided. The method performs authentication on a user based on a headphone transfer function (HPTF) when the user wears the headphones. The method can also detect whether there is a frequency response deviation between the user's HPTF and the adjusted HPTF. In addition, if there is a frequency response deviation between the user's HPTF and the adjusted HPTF, the method can dynamically compensate for the user's HPTF based on the detected frequency response deviation.

[0010] According to another aspect of the present disclosure, a system for authenticating and dynamically compensating for headphones is provided. The system includes a memory and a processor coupled to the memory. The processor is configured to perform authentication on a user based on a headphone transfer function (HPTF) when the user wears the headphones. In addition, the processor is configured to detect whether there is a frequency response deviation between the user's HPTF and the adjusted HPTF. In addition, the processor is configured to: in the event that there is a frequency response deviation between the user's HPTF and the adjusted HPTF, based on the detected frequency response deviation, dynamically compensate the user's HPTF.

[0011] According to yet another aspect of the present disclosure, a computer-readable storage medium including computer-executable instructions is provided. When the computer-executable instructions are executed by a computer, the computer performs the method disclosed herein. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A system configuration of FxLMS according to one or more embodiments of the present disclosure is shown.

[0013] Figure 2 A flow chart of a method for authenticating and dynamically compensating for headphones according to one or more embodiments of the present disclosure is shown.

[0014] Figure 3 A flow chart of a method for constructing an HPTF model and authenticating decisions according to one or more embodiments of the present disclosure is shown.

[0015] Figure 4 A flow chart of a method for real-time authentication of a user based on HPTF according to one or more embodiments of the present disclosure is shown.

[0016] Figure 5 A flow chart of a method for performing dynamic compensation based on HPTF according to one or more embodiments of the present disclosure is shown.

[0017] Figure 6 Exemplary results of an adjusted HPTF curve, a user's HPTF curve, and a corresponding compensation curve are shown.

[0018] Figure 7 A block diagram of HPTF-based motion compensation according to one or more embodiments of the present disclosure is shown.

[0019] Figure 8 The experimental results of the HPTF curve for the user's left ear are shown.

[0020] Fig. 9 The experimental results of the HPTF curve for the user's right ear are shown.

[0021] To aid understanding, the same reference numerals have been used as much as possible to refer to the same elements that are common in the drawings. It is contemplated that elements disclosed in one embodiment may be advantageously used in another embodiment without special indication. The drawings referred to herein should not be understood to be drawn to scale unless otherwise noted. Moreover, the drawings are often simplified and details or parts are omitted for clarity of illustration and explanation. The drawings and discussion are used to explain the principles discussed below, where the same reference numerals represent the same elements. DETAILED DESCRIPTION

[0022] Examples will be provided below for illustration. The description of various examples will be presented for illustrative purposes and is not intended to be exhaustive or limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments.

[0023] The headphone transfer function (HPTF) is defined as the acoustic transfer function of the sound pressure from the speaker of the headphone to the eardrum. In general, different headphones or listeners' personal HPTFs vary significantly because each headphone has its own design features and each listener also has his or her unique ear characteristics. Accordingly, the present disclosure will provide some implementation schemes for HPTF-based applications. For example, in a headphone product, the methods and systems discussed herein can be applied to biometric authentication. After biometric authentication, the present invention will provide a method and system for detecting and calibrating frequency response deviations to obtain the desired sound performance for individual users during the use of the headphone product.

[0024] Active noise canceling (ANC) headphones are based on monitoring the surrounding noise. That is, the headphones use internal and external microphones to capture the ambient sound. Then, using a calibrated playback system to maintain the amplitude of the surrounding noise and invert its phase, a high-precision anti-noise with a tightly coupled feedback loop can be reproduced.

[0025] HPTF involves two parts, the free-field measurement and the impulse response between the pinna plus ear canal and the internal microphone. Since the free-field measurement can be measured in a controlled environment and the manufacturing tolerances can be calibrated in the production line, the only variable left is the microphone-to-pinna plus ear canal response, as depicted as ear reference point (ERP) to ear entry point (EEP). This ERP to EEP transfer function (H 耳 ) between the auricle and the ear canal varies from person to person.

[0026] Figure 1 Schematic diagram of a system configuration of FxLMS according to one or more embodiments of the present disclosure. H can be dynamically calculated using a system identification algorithm (such as FxLMS) 耳 ,

[0027] w(n+1)=w(n)-μe(n)r′(n) (1)

[0028] where μ is the adaptation step size, w(n) is the weight vector at time n, and e(n) = d(n) + w T (n)r(n). e(n) is the residual noise measured by the error microphone, d(n) is the noise to be eliminated, r(n) and r′(n) are obtained from the convolutions r′(n) and r′(n)=h′(n)*x(n), respectively. x(n) is the synthetic reference signal, h(n) and h′(n) are the impulse responses H(f) and H′(f), respectively. H(f) is the transfer function of the secondary path, and H′(f) is an estimate of H(f), which is also regarded as the HPTF. The system configuration of FxLMS can be as follows Figure 1 shown.

[0029] Figure 2 FIG. 1 is a flow chart of a method for authenticating and dynamically compensating for a headset according to one or more embodiments of the present disclosure. Figure 1 As shown, at S201, when the user wears the headphones, authentication is performed on the user based on the headphone transfer function (HPTF). The authentication result can be used to determine whether the user can continue to use the headphones. Then, in order to obtain the desired sound performance of the headphones, adaptive and effective calibration and compensation can be performed in real time. For example, at S202, the frequency response deviation between the user's HPTF and the adjusted HPTF is detected. Then, at S203, dynamic compensation is performed on the user's HPTF based on the detected frequency response deviation. This will be described below. Figure 1 The specific implementation of the method shown.

[0030] HPTF Certification

[0031] In terms of authentication applications, the HPTF discrepancy problem can be transformed into an identification problem, which can be solved using statistical modeling (such as Bayesian methods and neural networks).

[0032] To distinguish the common HPTF of the target user, a statistical model will be used. In this embodiment (not limited to), a Gaussian mixture model (GMM) is constructed based on the measured impulse response. To construct the GMM reference, the free field response in the anechoic chamber is first measured as H 自由场 (f). For each data point, i∈P individuals (which are measured M times (total size is P*M) for training), the transducer-to-microphone transfer function is captured, depicted as H HPTF (f) (i omitted), then H is obtained by 耳 (f):

[0033] H 耳 (f) = H HPTF (f) / H 自由场 (f) (2)

[0034] To improve accuracy, the data can be preprocessed into amplitude data and relative phase data as shown below:

[0035]

[0036] ∠H 耳 (f) = tan -1 [Im(H 耳 (f)) / Re(H 耳 (f))] (4)

[0037] Then, each data point (i) can be regarded as a vector: [amplitude, phase] × [left, right] per sample data, and the data point is measured M times for different accessories on the head of each test subject. Then the global model is trained accordingly according to the GMM model construction process to obtain X~N 全局 (μ,σ).

[0038] HPTF model building and certification decision

[0039] Figure 3 A flow chart of a method for constructing an HPTF model and authenticating decisions according to one or more embodiments of the present disclosure is shown.

[0040] For example, the anechoic free-field transducer-to-microphone transfer function can usually be measured to obtain H 自由场 (f). See Figure 3 At S301, HPTFs from P individuals may be collected during manufacturing, each of which is mounted M times. At S302, a network with X to N HPTFs is formed based on the collected HPTFs. 全局 (μ x ,σ x ) of the global GMM. Then, at S303, the HPTFs from the end users may be collected and mounted M times. Based on the HPTFs collected from the end users, at S304, a GMM with Y to N 局部 (μ Y ,σ Y ). At S305, a runtime loss coefficient is determined by using a predefined loss function, such as minimum mean square error (MMSE).

[0041] To register a new target, use FxLMS in conjunction with the stored H 自由场 (f) H 目标 (f) = H HPTF (f) / H 自由场 (f), and this process for the target user will be repeated M times to create a local model Y~N by the predefined feature distance D 局部 (μ,σ), in this case, the predefined feature distance can be simplified to the distribution minimum mean square error (MMSE), as shown below,

[0042]

[0043] where β 0 …β P are parameter estimates.

[0044] To implement biometric authentication using the model created above, the distance function is calculated as follows: If average(||XY||)>(||Y-μY ||), since the feature distance is larger than the global X~N 全局 (μ x ,σ x ) is closer to local Y~N 局部 (μ Y ,σ Y ), so it can be determined that the device is authenticated. Otherwise, if the feature distance is greater than the local Y~N 局部 (μ Y ,σ Y ) is closer to the global X~N 全局 (μ x ,σ x ), the authentication returns failure as result.

[0045] Runtime HPTF extraction model

[0046] Figure 4 A flow chart of a method for real-time user authentication based on HPTF according to one or more embodiments of the present disclosure is shown. At S401, when the end user uses headphones, an audio stream from a microphone and a transducer may be obtained. Optionally, a check for audio playback and user input may be performed before obtaining the audio stream from the microphone and the transducer. At S402, a transfer function H between the transducer and the microphone may be obtained as described above. 耳 (f) Optionally, at S403, the convergence of the FxLMS algorithm is further checked, and if the FxLMS algorithm converges, the transfer function H is output 耳 (f) At S404, the transfer function is combined with the global X~N 全局 (μ x ,σ x ) and local Y~N 局部 (μ Y ,σ Y ) for comparison. Then, at S405, GMM MMSE-based authentication may be performed based on the comparison. For example, if the feature distance is greater than the global X-N 全局 (μ x ,σ x ) is closer to local Y~N 局部 (μ Y ,σ Y ), the device is authenticated. Otherwise, if the feature distance is greater than the local Y~N 局部 (μ Y ,σ Y ) is closer to the global X~N 全局 (μ x ,σ x ), the authentication process returns failure as result.

[0047] Deviation detection and frequency response calibration

[0048] In order to faithfully play back the sound to different listeners through the headphones and improve the sound experience of use, the HPTF can be calibrated and compensated. Some methods can be used to do this. For example, a method can be used: a microphone is placed in the ear canal of the listener and a one-time calibration is performed, playing a swept frequency signal or other special measurement signal. The method can compensate for the HPTF, but the compensation is only maintained for a short time, because the listener may not wear the headphones in the same position every time, which means that the listener must repeat this calibration every time he wants to use the headphones, otherwise the calibration may be invalid. This article also discloses an improved adaptive and effective method for real-time compensation.

[0049] Considering that listeners may wear headphones with air leakage and different listeners have different HPTFs which are usually very different from the HPTF of a standard phantom head, this paper proposes a method for acoustic engineers to compensate for the difference between the real HPTF and the carefully designed HPTF.

[0050] Figure 5 FIG. 5 is a block diagram of a method for performing dynamic compensation based on HPTF according to one or more embodiments of the present disclosure. At S501, the HPTFH(f) of the listener can be estimated by FxLMS, ​​and at S502, the amplitude response of the HPTFH(f) estimated by FxLMS of the listener is obtained. In addition, the adjusted HPTFH(f) can be obtained from the engineer. 0 The magnitude response of (f). Estimated HPTFH(f) and adjusted HPTFH 0 The magnitude response of (f) can be written as,

[0051] M(f)=|H(f)|,M 0 (f)=|H 0 (f)| (6)

[0052] Where |*| is an absolute value operator. Then, at S503, M(f) is compared with M 0 (f) To determine how much the frequency response deviates when a listener wears the headphones, for example in order to determine how much air is leaking in the low frequency range.

[0053] Then, at S504, dynamic compensation is performed on the user's HPTF curve based on the detected frequency response deviation. For example, a smooth and bounded calibration function F(*) is used to obtain the compensated magnitude M of its difference. c (f)

[0054] M c (f) = F(M 0 (f)-M(f)) (7)

[0055] where F(*) can be some linear or nonlinear function, for example, α and β are two parameters that can be adjusted according to the real system. Figure 6 Examples of an adjusted HPTF curve, a user's HPTF curve, and a corresponding compensation curve are shown.

[0056] Figure 7 FIG. 1 shows a block diagram of a method for performing dynamic compensation based on an HPTF according to one or more embodiments of the present disclosure. Figure 7 As shown, the system for dynamic compensation may include a pre-processing unit 701, a post-processing unit 702, an FxLMS system 703, a real-time calibration unit 704, and a compensation unit 705. For example, when a user wears headphones to listen to music, the music input may first be pre-processed by the pre-processing unit 701, such as through A / D conversion, EQ, adaptive limiter, downmixing, etc. Then, the pre-processed data is input to the compensation unit 705. By using the FxLMS system 703, the transfer function HPTF of the listener may be estimated as described above. In the real-time calibration unit 704, the amplitude response of HPTFH(f) is compared with the adjusted HPTFH from the engineer. 0 (f) is compared with the amplitude response, and then a smooth and bounded calibration function can be used to obtain the compensated amplitude M c (f). Then, the compensated amplitude M c (f) is output to the compensation unit 705 to be used based on the compensated amplitude M c (f) Performing dynamic compensation. The post-processing unit 702 may perform post-processing on the compensated data, such as by using EQ, adaptive limiter, etc.

[0057] In this disclosure, a new way is provided to anonymously detect individual differences between HPTFs across different users. It is then shown how the differences can be exploited for applications such as biometric authentication and headphone health detection based on frequency response deviations. Finally, based on the incremental difference between the detected HPTF and the target curve, dynamic compensation can be performed for the difference and a consistent listening experience will be provided.

[0058] Figure 8 and Fig. 9 The experimental results of the HPTF curve for the left ear and the right ear of the user are shown. For example, 5 users are randomly selected for the experiment, and each user wears the earphone normally to extract the HPTF accordingly. Figure 8 and Fig. 9The mean and variance for each user stacked on top of each other are shown separately for the left and right ears. As shown by the results, there are identifiable differences in the distribution between each person, and this can be depicted as the characteristic distance mentioned in the previous section. This characteristic distance is particularly evident around 500Hz to 2kHz and 5kHz to 15kHz, as these are the pinnae and ear canal differences between the test subjects. Figure 8 It may also indicate that there is some air leakage in the left channel of the headset, as the frequency response below 200Hz varies greatly from user to user.

[0059] Disclosed above is a novel method for interacting with an audible device using a runtime-computed HPTF model. Such actions can be found in consumer devices such as unlocking security devices (e.g., mobile phones) and acoustic personalization (e.g., play / pause, load / store playlists). The same applies to e-commerce and software services. For example, authentication protocols for secure payments (e.g., Google Store) and conferencing software for identity recognition and verification (e.g., WebEx login ID automated meeting setup). The technology disclosed herein is based on the HPTF differences between individuals from both the left and right ears, and provides an alternative means for digital authentication and human-computer interaction. This also extends to methods for determining audible acoustic behavior using statistical analysis.

[0060] The description of various embodiments has been presented for illustrative purposes, but is not intended to be exhaustive or limited to the disclosed embodiments. The terminology used herein is selected to best explain the principles of the embodiments, the practical application or technical improvement of technologies found in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

[0061] In the foregoing, reference is made to the embodiments presented in the present disclosure. However, the scope of the present disclosure is not limited to the specifically described embodiments. On the contrary, any combination of the aforementioned features and elements (whether or not related to different embodiments) is contemplated as implementing and practicing the envisioned embodiments. In addition, although the embodiments disclosed herein may achieve advantages over other possible solutions or over the prior art, whether a particular advantage is achieved by a given embodiment does not limit the scope of the present disclosure. Therefore, the aforementioned aspects, features, embodiments and advantages are merely illustrative and are not considered to be elements or limitations of the appended claims unless expressly stated in the claims.

[0062] Various aspects of the present disclosure may take the following forms: an entirely hardware implementation, an entirely software implementation (including firmware, resident software, microcode, etc.) or an implementation combining software and hardware aspects, each of which may generally be referred to herein as a "circuit", "module", "unit" or "system".

[0063] The present disclosure may be a system, method and / or computer program product. The computer program product may include one (or more) computer-readable storage media having computer-readable program instructions thereon for causing a processor to perform various aspects of the present disclosure.

[0064] A computer-readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. An incomplete list of more specific examples of computer-readable storage media includes the following: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device (such as a punched card or a raised structure in a groove with instructions recorded thereon), and any suitable combination of the foregoing. As used herein, a computer-readable storage medium should not be understood as a transient signal itself, such as a radio wave or other free-propagating electromagnetic wave, an electromagnetic wave propagated through a waveguide or other transmission medium (e.g., a light pulse passing through a fiber optic cable), or an electrical signal transmitted through a wire.

[0065] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to a corresponding computing / processing device via a network (e.g., the Internet, a local area network, a wide area network, and / or a wireless network), or downloaded to an external computer or an external storage device. The network can include copper transmission cables, optical transmission fibers, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers.

[0066] Various aspects of the present disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each box in the flowchart illustration and / or block diagram and the combination of boxes in the flowchart illustration and / or block diagram can be implemented by computer-readable program instructions.

[0067] These computer-readable program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device establish means for implementing the functions / operations specified in one or more blocks of the flowchart and / or block diagram.

[0068] The flow charts and block diagrams in the accompanying drawings illustrate the architecture, functions and operations of possible implementations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram may represent an instruction module, an instruction fragment or an instruction portion, which includes one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions annotated in the box may occur in an order different from the order annotated in the figure. For example, the two boxes shown in succession can actually be executed substantially at the same time, or these boxes can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each box illustrated in the block diagram and / or flow chart, and the combination of boxes illustrated in the block diagram and / or flow chart, can be implemented by a system based on dedicated hardware that performs a specified function or action or performs a combination of dedicated hardware and computer instructions.

[0069] While the foregoing is directed to embodiments of the present disclosure, other and further embodiments of the disclosure may be devised without departing from the basic scope thereof, which is determined by the following claims.

Claims

1. A method for authenticating and dynamically compensating a headset, the method comprising: When the user wears the headset, performing authentication on the user based on a headphone transfer function (HPTF); detecting a frequency response deviation between the user's HPTF and the adjusted HPTF; as well as dynamically compensating the user's HPTF based on the detected frequency response deviation; The performing of the authentication further comprises: Constructing HPTF models and certification decisions; measuring the HPTF for the user; and The user is authenticated based on the measured HPTF, the constructed HPTF model, and the authentication decision.

2. The method according to claim 1, wherein said constructing a HPTF model and certifying a decision further comprises: Collect global HPTF from multiple individuals, repeating a predetermined number of times for each individual; Based on the collected global HPTFs, a global model with global distribution is formed; collecting local HPTF from the user, repeating the predetermined number of times; Based on the collected local HPTFs, forming a local model with a local distribution; and Determine the runtime loss coefficient by using a predefined loss function.

3. The method according to claim 2, wherein the method further comprises: Calculating feature distance based on the global model and the local model; If the feature distance is closer to the local model than the global model, determining that the authentication is successful; as well as If the feature distance is closer to the global model than the local model, it is determined that the authentication has failed. The method according to claim 3 , wherein the global model and the local model are based on a Gaussian mixture model.

5. The method of any one of claims 1 to 4, wherein the method further comprises measuring an anechoic free field transducer to microphone transfer function.

6. The method of claim 1, wherein the detecting the frequency response deviation between the user's HPTF and the adjusted HPTF further comprises: estimating the HPTF of the user by means of an FxLMS algorithm; obtaining a magnitude response of the estimated HPTF of the user; comparing the amplitude response to an adjusted amplitude response; as well as Based on the comparison, the frequency response deviation is determined in real time.

7. A system for authenticating and dynamically compensating a headset, the system comprising: Storage device; as well as a processor coupled to the storage device; The processor is configured to: When the user wears the headset, performing authentication on the user based on a headphone transfer function (HPTF); detecting a frequency response deviation between the user's HPTF and the adjusted HPTF; and dynamically compensating the user's HPTF based on the detected frequency response deviation; The processor is further configured to: Constructing HPTF models and certification decisions; measuring the HPTF for the user; and The user is authenticated based on the measured HPTF, the constructed HPTF, and the authentication decision.

8. The system of claim 7, wherein the processor is further configured to: Collect global HPTF from multiple individuals, repeating a predetermined number of times for each individual; Based on the collected global HPTFs, a global model with global distribution is formed; collecting local HPTF from the user, repeating the predetermined number of times; Based on the collected HPTF, forming a local model having a local distribution; and Determine the runtime loss coefficient by using a predefined loss function.

9. The system of claim 8, wherein the processor is further configured to: Calculating feature distance based on the global model and the local model; If the feature distance is closer to the local model than the global model, determining that the authentication is successful; and If the feature distance is closer to the global model than the local model, it is determined that the authentication has failed.

10. The system of claim 9, wherein the global model and the local model are based on Gaussian mixture models.

11. The system of any one of claims 7 to 10, wherein the processor is further configured to measure an anechoic free field transducer to microphone transfer function.

12. The system of claim 7, wherein the processor is further configured to: estimating the HPTF of the user by means of an FxLMS algorithm; obtaining a magnitude response of the estimated HPTF of the user; comparing the magnitude response to an adjusted magnitude response; and Based on the comparison, the frequency response deviation is determined in real time.

13. A computer-readable storage medium comprising computer-executable instructions, which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 6.

Images