Managing active noise cancellation features

Abstract

A first input signal captured by one or more sensors associated with an ANR earphone is received. A frequency domain representation of the first input signal is computed for a set of discrete frequencies based on a parameter set generated for a digital filter disposed in an ANR signal flow path of the ANR earphone, the parameter set causing a loop gain of the ANR signal flow path to substantially match a target loop gain. Generating the parameter set includes adjusting a response of the digital filter at frequencies (e.g., spanning between 200 Hz and 5 kHz). Adjusting a response of at least 3 second order elemental sections of the digital filter. A second input signal in the ANR signal flow path is processed using the generated parameter set to generate an output signal for driving an electro-acoustic transducer of the ANR earphone.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority and benefit to U.S. Patent Application Serial No. 16 / 857,382, entitled “MANAGING CHARACTERISTICS OFACTIVE NOISE REDUCTION”, filed April 24, 2020, published as U.S. Patent No. 10,937,410. Technical Field

[0003] This disclosure relates to the features of managing active noise cancellation. Background Technology

[0004] The earpiece of a headset or other audio or multimedia device configured for use by a user (such as a single (e.g., left and right) wireless or wired earbud, or the earpiece of a headset or other wearable device) may include circuitry configured based on assumed acoustic conditions that depend on how well the earpiece fits when worn in, on, or around the ear, and the acoustic characteristics of the wearer's ear connected to the headset. For example, for headphones using active noise cancellation (ANR), the actual acoustic conditions associated with a particular fit and a single ear are part of the feedback loop used to provide ANR. To ensure that this feedback loop is stable for any fit that any particular user might experience at any given time, and thus avoids artifacts associated with feedback instability, a trade-off may be made to sacrifice noise cancellation performance for robust stability. Summary of the Invention

[0005] In one aspect, generally speaking, the method includes: receiving a first input signal captured by one or more sensors associated with an active noise-canceling (ANR) headset; calculating a frequency domain representation of the first input signal for a set of discrete frequencies by one or more processing devices; generating, by the one or more processing devices, a set of parameters for a digital filter disposed in an ANR signal flow path of the ANR headset based on the frequency domain representation of the input signal, the set of parameters such that the loop gain of the ANR signal flow path substantially matches a target loop gain, wherein generating the set of parameters includes: adjusting the response of the digital filter at frequencies at least spanning between about 200 Hz and about 5 kHz; adjusting the response of at least three second-order fundamental sections of the digital filter; and processing a second input signal in the ANR signal flow path using the generated set of parameters to generate an output signal for driving an electroacoustic transducer of the ANR headset.

[0006] An aspect may include one or more of the following features.

[0007] The first input signal includes features that vary with different users, and the second input signal includes features that vary less between different users compared to the first input signal.

[0008] The one or more sensors include a feedback microphone for the ANR headphones, and the ANR signal flow path includes a feedback path disposed between the feedback microphone and the electroacoustic transducer.

[0009] For most of the frequency range in which the feedback path has positive loop gain, the change in feedback insertion gain measured by multiple users is less than the change in the physical acoustic response of the ANR headphones measured by the response between the electroacoustic transducer and the feedback microphone used by the multiple users.

[0010] For most of the frequency range in which the feedback path has positive loop gain, the change in the feedback insertion gain is at least 10% smaller than the change in the physical acoustic response of the ANR headphones.

[0011] The average feedback insertion gain, as measured by multiple users, has a high-frequency division of approximately 1.5 kHz or greater.

[0012] Generating the parameter set includes: accessing the nominal parameter set of the digital filter, determining the correction parameter set based on the frequency domain representation of the first input signal, and generating the parameter set as a combination of corresponding parameters in the nominal parameter set and the correction parameter set.

[0013] The nominal parameter set is calculated based on training data that includes responses from multiple ears.

[0014] The nominal parameter set is generated by performing an optimization process that is configured to generate the corresponding ear response.

[0015] Determining the set of correction parameters includes: calculating the loop gain of the nominal parameter set of the digital filter; generating an error vector that includes the deviation between the loop gain and the corresponding target loop gain at different frequencies; and generating the set of correction parameters as the output of the optimization process based on the statistical data of the training data.

[0016] When ANR is activated, the total insertion gain of the ANR headphones is less than -30dB in the frequency range of approximately 1kHz to 2kHz.

[0017] For example, the average active insertion gain measured by multiple users has a high-frequency division of about 2.2 kHz or greater.

[0018] The parameter set is generated within 1 second of receiving the first input signal.

[0019] The method also includes storing the generated set of parameters used to identify or authenticate users.

[0020] A first input signal is captured in response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones. The audio signal includes a broadband signal that includes energy at multiple frequencies in a set of discrete frequencies, and the frequency domain of the first input signal indicates the ear's response to the audio signal.

[0021] The audio signal has a spectrum comprising 10 or more tones concentrated at predetermined frequencies between approximately 45 Hz and 16 kHz.

[0022] These predetermined frequencies include multiple frequencies above 1 kHz, with intervals between these frequencies less than or equal to 1 / 4-octave.

[0023] The audio signal is automatically delivered in response to the detection that the ANR headphones are positioned in, on, or around the user's ear.

[0024] The audio signal is automatically delivered in response to the detection of oscillations in the ANR signal flow path.

[0025] The one or more sensors include a feedforward microphone and a feedback microphone of the ANR headphones, the first input signal includes the ratio of the feedback microphone signal to the feedforward microphone signal, and the ANR signal flow path includes a feedforward path disposed between the feedforward microphone and the electroacoustic transducer.

[0026] The feedforward microphone signal is captured in response to determining that the ambient noise near the ANR headphones is above a threshold.

[0027] In response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones, a feedback microphone signal is captured, the audio signal comprising a broadband signal including energy at multiple frequencies in a set of discrete frequencies.

[0028] In response to determining that the ambient noise near the ANR headphones is higher than a threshold, a feedforward microphone signal is captured, and the following are detected: (i) the absence of an audio signal played through an electroacoustic transducer; and (ii) the absence of user speech.

[0029] The feedforward microphone signal and the feedback microphone signal, or both, are repeatedly captured in units of each of the multiple time intervals.

[0030] The method may also include measuring the seal quality between the ANR headphones and the wearer's ear, and reducing the target loop gain when the seal quality is less than a predetermined threshold.

[0031] On the other hand, generally speaking, the method includes: receiving a first input signal captured by one or more sensors associated with active noise cancellation (ANR) headphones; calculating a frequency domain representation of the first input signal by one or more processing devices; generating a set of parameters for a digital filter disposed in the ANR signal flow path of the ANR headphones based on the frequency domain representation of the input signal by the one or more processing devices, the set of parameters such that the loop gain of the ANR signal flow path substantially matches a target loop gain, wherein the generated set of parameters includes: a first parameter associated with a first frequency in a set of discrete frequencies, the first frequency being less than a high-end gain division frequency (at which the magnitude of the loop gain associated with the ANR signal flow path is equal to one), and a second parameter associated with a second frequency in the set of discrete frequencies, the second frequency being greater than the high-end gain division frequency; and using the generated set of parameters to process a second input signal in the ANR signal flow path to generate an output signal for driving an electroacoustic transducer of the ANR headphones.

[0032] In some specific implementations, the high-gain divider frequency is greater than 1kHz.

[0033] On the other hand, generally speaking, the method includes: in response to sensing that the earpiece of an active noise-canceling (ANR) headphone is positioned in, on, or around the ear: (i) receiving a first input signal captured by one or more sensors associated with the ANR headphone; (ii) calculating a frequency domain representation of the first input signal for a set of discrete frequencies by one or more processing devices; (iii) generating a set of parameters of a digital filter disposed in the ANR signal flow path of the ANR headphone based on the frequency domain representation of the input signal by the one or more processing devices; and (iv) processing a second input signal in the ANR signal flow path using the generated set of parameters to generate an output signal for driving an electroacoustic transducer of the ANR headphone.

[0034] An aspect may include one or more of the following features.

[0035] A first input signal is captured in response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones. The audio signal includes a broadband signal that includes energy at multiple frequencies in a set of discrete frequencies, and the frequency domain of the first input signal indicates the ear's response to the audio signal.

[0036] The audio signal has a spectrum comprising 10 or more tones concentrated at predetermined frequencies between approximately 45 Hz and 16 kHz.

[0037] These predetermined frequencies include at least one frequency below 50 Hz and at least one frequency above 15 kHz.

[0038] These predetermined frequencies include multiple frequencies above 1 kHz, with intervals between these frequencies less than or equal to 1 / 4-octave.

[0039] The audio signal is automatically delivered in response to sensing that the ANR headphones are positioned in, on, or around the user's ear.

[0040] The one or more sensors include a feedback microphone for the ANR headphones, and the ANR signal flow path includes a feedback path disposed between the feedback microphone and the electroacoustic transducer.

[0041] Generating the parameter set includes: accessing the nominal parameter set of the digital filter, determining the correction parameter set based on the frequency domain representation of the first input signal, and generating the parameter set as a combination of corresponding parameters in the nominal parameter set and the correction parameter set.

[0042] The nominal parameter set is calculated based on training data that includes responses from multiple ears.

[0043] The nominal parameter set is generated by performing an optimization process that is configured to generate the corresponding ear response.

[0044] Determining the set of correction parameters includes: calculating the loop gain of the nominal parameter set of the digital filter; generating an error vector that includes the deviation between the loop gain and the corresponding target loop gain at different frequencies; and generating the set of correction parameters as the output of the optimization process based on the statistical data of the training data.

[0045] The method also includes storing the generated set of parameters used to identify or authenticate users.

[0046] Generating the parameter set includes: adjusting the response of the digital filter at frequencies spanning at least between approximately 200 Hz and approximately 5 kHz; and adjusting the response of at least three second-order fundamental sections of the digital filter.

[0047] On the other hand, generally speaking, the method includes: in response to sensing that the ambient noise level near the active noise-canceling (ANR) headphones is higher than a predetermined threshold: (i) receiving a first input signal captured by one or more sensors associated with the ANR headphones; (ii) calculating a frequency domain representation of the first input signal for a set of discrete frequencies by one or more processing devices; (iii) generating a set of parameters of a digital filter disposed in the ANR signal flow path of the ANR headphones by the one or more processing devices based on the frequency domain representation of the input signal; and (iv) processing a second input signal in the ANR signal flow path using the generated set of parameters to generate an output signal for driving an electroacoustic transducer of the ANR headphones.

[0048] An aspect may include one or more of the following features.

[0049] The one or more sensors include a feedforward microphone for the ANR headphones, and the ANR signal flow path includes a feedforward path disposed between the feedforward microphone and the electroacoustic transducer.

[0050] The one or more sensors also include a feedback microphone for the ANR headset, and the first input signal includes the ratio of the feedback microphone signal to the feedforward microphone signal.

[0051] In response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones, a feedback microphone signal is captured, the audio signal comprising a broadband signal including energy at multiple frequencies in a set of discrete frequencies.

[0052] The feedforward microphone signal and the feedback microphone signal, or both, are repeatedly captured in units of each of the multiple time intervals.

[0053] Generating the parameter set includes: accessing the nominal parameter set of the digital filter, determining the correction parameter set based on the frequency domain representation of the first input signal, and generating the parameter set as a combination of corresponding parameters in the nominal parameter set and the correction parameter set.

[0054] The nominal parameter set is calculated based on training data that includes responses from multiple ears.

[0055] The nominal parameter set is generated by performing an optimization process that is configured to generate the corresponding ear response.

[0056] Determining the set of correction parameters includes: calculating the loop gain of the nominal parameter set of the digital filter; generating an error vector that includes the deviation between the loop gain and the corresponding target loop gain at different frequencies; and generating the set of correction parameters as the output of the optimization process based on the statistical data of the training data.

[0057] The method also includes storing the generated set of parameters used to identify or authenticate users.

[0058] Generating the parameter set includes: adjusting the response of the digital filter at frequencies spanning at least between approximately 200 Hz and approximately 5 kHz; and adjusting the response of at least three second-order fundamental sections of the digital filter.

[0059] It may have one or more of the following advantages.

[0060] Systems and programs for customizing compensators for ANR circuitry can use the ear frequency response characterizing a user's specific acoustic condition (e.g., when the earpiece is placed in, on, or around the user's ear). Differences between users (e.g., the shape of the user's ear canal and the acoustic characteristics of the wearer's ear connected to the headphones) and / or variations in earpiece fit can be compensated for by corresponding changes to one or more filters within the ANR circuitry. In some implementations, the customization program may use perturbation techniques to make the calculations more efficient. These perturbation techniques may include linear perturbation techniques that utilize substantially linear adjustments. In other implementations, the customization program may use other techniques, such as machine learning or deep neural networks for customizing compensators for ANR circuitry.

[0061] Due to the performance improvements offered by custom-designed ANRs, various performance factors can be improved. For example, because ANRs do not need to meet certain constraints (e.g., control loop stability) for various ears / fits, the control loop can be designed with predetermined optimized characteristics after customization. An example of a characteristic that can be precisely determined for each ear is canal resonance, as described in more detail below. Furthermore, auditory effects, such as residual obstruction of the wearer's voice, can be reduced due to the increased feedback loop gain and bandwidth achieved through customization for individual ears while maintaining sufficient stability.

[0062] Due to computational efficiency and the minimal computational resources required, custom modules for executing custom programs can be relatively compact. In some implementations, the custom module may be built into a handset or other wearable audio device. The custom module may include the code and data needed to execute the custom program without requiring an online connection to another device (e.g., a phone or cloud infrastructure). For example, a connection might be used to provide firmware updates, but the connection may not need to be active during the custom program.

[0063] In some implementations, the ability to customize the performance of both the feedback compensator and the feedforward compensator separately can also be useful. For example, the feedback compensator can be customized immediately after the wearable audio device is powered on (e.g., in response to detecting that the earpiece has been worn). The feedforward compensator can be customized at a similar time or later, depending on whether there is a sufficient level of ambient noise to perform feedforward customization using signals from a microphone that senses ambient noise. Attached Figure Description

[0064] This disclosure is best understood in conjunction with the accompanying drawings and the following detailed description. It should be emphasized that, in accordance with common practice, the various features in the drawings are not drawn to scale. Instead, for clarity, the dimensions of the various features have been arbitrarily enlarged or reduced.

[0065] Figure 1A This is an illustration of an example of the earpiece in an earphone.

[0066] Figure 1B , Figure 1C and Figure 1D These are illustrations of the earpiece worn in the ear, on the earphone, and around the ear.

[0067] Figure 2 It is a block diagram of a part of a system that includes ANR circuitry.

[0068] Figure 3A and Figure 3B This is a graph showing the amplitude of the example frequency response.

[0069] Figure 3C and Figure 3D These are graphs showing the standard deviations of the amplitude and phase of the exemplary frequency response.

[0070] Figure 4A and Figure 4B These are curves showing the amplitude and phase characteristics of the filter, respectively.

[0071] Figure 4C and Figure 4D These are curves showing the relative amplitude and phase characteristics of the filter, respectively.

[0072] Figure 5A , Figure 5B and Figure 5C This is a graph showing the amplitude and phase of an exemplary feedback loop response.

[0073] Figure 5D , Figure 5E , Figure 5F and Figure 5G This is a graph showing the sensitivity of an exemplary feedback loop.

[0074] Figure 5H and Figure 5I This is a graph illustrating an example of insertion gain comparison.

[0075] Figure 6 This is a flowchart of an exemplary control procedure. Detailed Implementation

[0076] Some circuitry within the earpiece used to regenerate the desired signal (such as music or other acoustic signals) can be customized to the specific acoustic characteristics of a user's ear, derived from the degree of earphone seal to the ear, the detailed shape of the user's ear canal, and the characteristics of the tissues of the ear and eardrum. For example, ANR performance can be customized to use user-specific filter parameters by configuring the ANR circuitry. In some cases, these filter parameters may be stored in or coupled to memory within the earpiece. Some components within the earpiece are used in the customization process, as described in more detail below. References Figure 1A Examples of left earpiece 100L / right earpiece 100R that can be configured to provide customized ANR performance include acoustic drivers 102L (in earpiece 100L) and 102R (in earpiece 100R). The earpiece also includes feedback microphones 104L (in earpiece 100L) and 104R (in earpiece 100R) and feedforward microphones 106L (in earpiece 100L) and 106R (in earpiece 100R). The acoustic drivers 102L / 102R and feedback microphones 104L / 104R are positioned within the respective earpieces 100L / 100R (as indicated by the dashed lines) such that the characteristics of these transducers, their positions, the volume and ports within the earpiece structure, are combined with the geometry and characteristics of the wearer's ear to define the internal acoustic environment formed when the earpiece is worn. Feedforward microphones 106L / 106R are positioned on the outer surface of the corresponding earpieces 100L / 100R, such that these microphones are exposed to the external acoustic environment when the earpieces are worn. In the examples described below, customization procedures are described relative to individual earpieces. In some embodiments, customization procedures are performed independently for each of the left and right earpieces. Alternatively, in other embodiments, if certain assumptions are made regarding the symmetry of the shape of the user's ear and / or the fit of the earpieces in, on, or around the user's ear, some or all of the customization procedures performed in one earpiece can be used to customize other earpieces without needing to repeat the complete customization procedure for other earpieces. For example, a customized set of filter parameters for one earpiece can be used as the default set of filter parameters for another earpiece by transmitting filter parameters between the earpieces via a wired or wireless communication connection between these earpieces.

[0077] Figures 1B to 1D Examples of earpieces positioned in, on, or around the ear are shown (in-ear, on-ear, and around-the-ear adapters are provided). Reference Figure 1B The earpiece 110 is placed in the ear 130, wherein the flexible tip 112 is positioned within the outer portion of the canal 113 of the ear 130, thereby creating a substantially closed acoustic environment within the canal 113. (Reference) Figure 1CThe earpiece 114 is placed on the ear 130, wherein the earpiece 114 forms a cushioned portion that is held against the auricle of the ear 130 to form a substantially sealed acoustic environment for the guide channel 113. Reference Figure 1D The earpiece 120 is placed around the ear 130, with the buffer portion 122 positioned against the portion of the head 140 surrounding the ear 130 to form a substantially sealed acoustic environment for the guide channel 113.

[0078] Figure 2 A block diagram representation 200 of a system positioned in, on, or around the ear in the context of an earpiece is shown. The system includes a controlled system (also called a device) and a portion of a system providing custom control, which in this example includes ANR circuitry including a feedback microphone and a feedforward microphone (also called a device sensor). The system is also situated in an external acoustic environment that provides noise input to the system. In this example, the device corresponds to sound propagating into the ear, represented by the “ear” variable e. The system is able to obtain an approximation of this variable using a feedback microphone placed within the containment / internal acoustic environment formed by the earpiece (from which sound propagates further into the ear canal). This system approximation of the ear variable, controlled using custom feedback, is represented by the “system” variable s. The system is able to obtain a noise sample from the external acoustic environment, just outside the earpiece, represented by the variable n, using a feedforward microphone placed somewhere outside the earpiece. This sample of the external environment outside the earpiece is represented by the “external” variable o. These variables can have quantitative values ​​indicating physical quantities (such as pressure) associated with sound waves, and can be represented as time-dependent signals with values ​​that vary with time, or as frequency-dependent signals with values ​​that vary with frequency. Finally, the system includes two compensation filters: K fb and K ff These two compensation filters collect signals from the feedback microphone and feedforward microphone, respectively, to determine the electrical signal input to the acoustic driver in the earpiece, denoted by the variable d. The following set of equations represents the set of relationships between the various variables in the system.

[0079] d = K fb s+K ff o

[0080] s = G sd d+G sn n

[0081] e = G ed d+G en n

[0082] o = G on n

[0083] The value of G, represented by various subscripts, corresponds to the transfer function of either the microphone (o or s) or the ear (e) (as the first subscript), and the transfer function from either the input (n or d) (as the second subscript). Therefore, the device transfer function corresponds to the value G. sd In some representations, the transfer function can be expressed as a frequency-dependent complex-valued expression using any of a variety of formulas. These formulas are used to represent time-dependent signals (e.g., continuous-time or discrete-time signals) using any of a variety of transforms (e.g., Fourier transform, Laplace transform, discrete Fourier transform, or Z-transform). The value denoted as K corresponds to a compensator, which can be implemented as a digital filter, including a feedback compensator K. fb and feedforward compensator K ff When implemented digitally with low latency (which is important for feedback systems), such filters are often designed as combinations of second-order recursive filters, which are often called "double second-order" (because they are expressed in the Z-domain), and they are implemented with unit delay operators z. -1 The ratio of two quadratic functions in units. Each bisecond order is specified by five parameters, thus determining the two poles and two zero-plus gains, which characterize the frequency response of the bisecond order. In some specific implementations, additional compensators may be included at various locations in the system, such as audio equalizer compensators. Any of these compensators can be customized as part of the customization techniques described herein.

[0084] The driver d and noise n in these equations can be eliminated to produce a pair of relationships representing the ratio of the acoustic signal provided to the ear, measured at the feedback microphone or respectively, relative to the noise:

[0085]

[0086]

[0087] For reference, the open-ear response to noise can be defined as:

[0088]

[0089] The overall performance of the system can be defined as the insertion gain (IG), which in this example represents the ratio of sound at the ear to noise, where the earpiece is in, on, or around the ear and where the ANR circuitry is activated (referred to as an "active system"), divided by the open-ear response, which is... Divide by

[0090]

[0091] The passive insertion gain PIG is defined as the purely passive response to the active system:

[0092]

[0093] These exemplary expressions have been written as transfer functions relative to noise, since noise can be considered as input to the system. In general, there may not be a measure of "noise" in the sense of a diffuse field, but there may be a measure of noise at a single point (e.g., as measured with an omnidirectional reference microphone). For this reason, before and after placing the earpiece in, on, or around the ear, where the system is in active or passive mode respectively, the expressions for IG and PIG can be evaluated as the ratio of energy (without phase) acquired at the microphone located at the point in the system corresponding to the variable e. For example, a small microphone can be suspended along the middle of the ear canal length to measure e.

[0094] Extending this further, the various noise terms can be represented as normalized cross-spectral representations among available microphones, as shown below:

[0095]

[0096] Using these definitions and substituting them into the equations for IG, another, more compact definition of the insertion gain can be expressed as:

[0097]

[0098] We now have the total insertion gain of the active system and the system and two compensators K fb and K ff The acoustically related equations for measurements. These equations can be used to calculate the acoustic properties of a given set of G. sd The optimal feedback compensator K for a set of conditions (defined by one or more ears). fb .

[0099] With regard to these other parameters and the target insertion gain, the optimal feedforward compensator K can also be solved. ff This paper presents a solution for Kfb. In some specific implementations, the insertion gain IG for full ANR (e.g., for maximum noise cancellation) is set to 0, and the insertion gain IG for minimum ANR (e.g., for maximizing the perception of external acoustic environment, including insertion gain variations from only the feedback portion of the system, bypassing PIG and FBIG) is set to 1. The target IG can also be set to some desired response that varies with frequency. Different compensation filters can be configured to achieve the "noise cancellation" (nc) condition or the "perception" (aw) condition, or an intermediate condition within the insertion gain range between 0 and 1. Multiple Kfb... ffFilters can be stored in the headset or computed online, and controls are used to switch between them or combine several filters operating in parallel to achieve the desired effect in the resulting IG. Examples are further described in U.S. Patents 10,096,313 and 10,354,640, the entire contents of which are incorporated herein by reference.

[0100] Considering these definitions, other constraints, and the acoustic response measured against both the driver and noise inputs, various optimization techniques can be used to configure the filter parameter sets of each of these digital filters, thereby achieving feedforward and feedback compensators. For example, measurements can be taken against a large sample of users with different ear characteristics to determine a single filter parameter set for each of these compensators, which can be used for all users and all fits of the earpiece in, on, or around the user's ear. In some specific implementations of such a fixed filter configuration, the filters can be positioned around an average measured G... sd The design aims to deliver some average performance level across all users, where some users perform better than the average noise reduction and some users perform worse. Preferably, in some specific implementations with a fixed filter configuration, additional conditions such as stable feedback behavior may be imposed for all users, which may cause the filter to accommodate a worst-case G. sd The response results in less performance than could be achieved if the design were only used for average values.

[0101] In addition, headphones can reduce G-forces at high frequencies. sd The design of variations is determined, for example, through the interaction between the acoustic design of the headphones and the characteristics of the wearer's ear. This reduced variation simplifies the fixed K... fb The design is suitable for any user's ear, but also results in less cancellation bandwidth. U.S. Patent 9,792,893, incorporated herein by reference in its entirety, describes an earphone design that, due to its close fit to the ear canal, achieves the possibility of high acoustic cancellation bandwidth (as measured in the ear canal, the system variable e in the above mathematical model). To achieve the full performance of such earphones, a custom-compensated filter matched to an individual user's ear can be used. To illustrate this, Figure 3A The figure of G, measured in a set of ears, is shown in a system with a looser connection having nozzles designed to reduce variation (such exemplary systems are described in U.S. Patent 9,792,893). sd Amplitude. Figure 3B G is shown as measured in a comparable set of ears in a more tightly coupled system. sd The amplitude, an example of this more tightly coupled system, is described in more detail in U.S. Patent 9,792,893, which results in high potential elimination. In both cases, Gsd The response has been normalized by gain to approximately adjust for lower-frequency variations caused by interaural changes in the seal and ear canal volume. Larger variations at higher frequencies can be observed in tightly fitted headphones. Figure 3B The following figure shows the amplitude ( Figure 3C ) and phase ( Figure 3D This variation in terms of the standard deviation of the signal; other measures of variation may also be used. Note how the two variation curves essentially diverge starting from approximately 1.5 kHz. Loosely coupled headphones (which in this example have a feedback potential cancellation bandwidth of approximately 1 kHz) can be successfully compensated using a fixed filter for any ear. This is due to the large amount of G at and near the feedback loop gain division frequency. sd The variable, tightly coupled headphones (which in this example have a feedback potential cancellation bandwidth greater than 2.5 kHz) cannot be compensated for with a fixed filter that has a feedback loop bandwidth close to the potential cancellation bandwidth. To achieve practical feedback noise cancellation performance close to the acoustic potential cancellation of these headphones, feedback compensation filters matched to each ear can be used. This disclosure describes practical techniques for implementing such filter customization. However, it should be noted that the described techniques can also be applied to loosely coupled systems.

[0102] The system can be designed to determine a custom filter configuration for each user's ear and / or for each user's every wear (e.g., each time the earpiece is placed in, on, or around the user's ear), thereby achieving improved performance for each user. Given the computational cost required to ensure all performance and stability constraints are met, it may be difficult to execute a complete optimization procedure from scratch for each wear to achieve such a custom filter configuration for a practical, power-constrained wearable system. However, using the techniques described in this document, computations can be performed in an online procedure for each user and each wear event, and these computational resources can be built into the wearable device including the earpiece.

[0103] In this online program, a custom filter parameter set can be generated based on a nominal dataset determined using statistics from training data. For example, the nominal dataset including the nominal filter parameter set can be based on multiple ear frequency responses (G for each subject ear). sd G ed N so and N eo ) and the corresponding filter frequency response (K fb and K ffThe training data is computed. Any of a variety of techniques can be used to compute this nominal set of filter parameters. An example of an analysis that can be performed to generate the nominal dataset is now described. An offline procedure can be used to generate a custom compensator for a single ear and adapt it to that ear using any of a variety of optimization methods. The offline procedure may not need to be as fast as an online procedure. The offline procedure can take the response corresponding to a single wear as input and produce a single set of filter parameters for a compensator used only for that wear, satisfying certain predetermined design constraints related to the acoustic characteristics of the headphones (potential cancellation, volume displacement, etc.) and system performance goals and stability considerations for IG or FBIG. In this way, a large number of wearing events can be treated as input and used to generate a large number of matched compensators as training data, which can show some basic structures that can be utilized. The optimization method used is not important, as long as the system designer has chosen the method that is the best choice for giving a compensator filter for a given wear (single ear acoustic conditions).

[0104] Training data may include real-world ear response data in the form of a measurement transfer function and a normalized cross-spectral representation. For example, real-world ear response data can be defined as the ratio of the input and output Fourier transforms (e.g., Fast Fourier Transform (FFT)) of a time-domain signal recorded by a microphone. The results of the real-world ear response data can be stored as a vector of complex numbers. In this representation, there is no fundamental physical model of the device characteristics (in this example), a combination of user ear characteristics (such as those influenced by the size and shape of the user's ear canal), and the earpiece (which has a specific profile in the frequency response). However, many features of the data may exist that can be considered and influence these responses: such as driver design, microphone response, port design, ear canal geometry, and fit quality. Any of these features can affect the driver's response to the system microphone response G. sd Furthermore, these features can generate identifiable characteristics in the frequency response.

[0105] Various device parameters, such as the poles and zeros that fit the actual ear response data, can be identified within the data, and these parameters can be aggregated when plotted as a function of frequency. Similarly, the poles and zeros corresponding to the second order of the compensator for each individual wear will vary and aggregate, and especially at higher frequencies, the device parameters and compensator parameters can exhibit a roughly inverse functional relationship. For example, the device zeros and compensator poles can be aligned, and the device poles and compensator zeros can be aligned. This can be understood from a control design perspective, as the high-level goal of feedback control design is to reverse the device dynamics during the shaping process to the desired cyclic response. Therefore, training data provides an opportunity to specify the compensator parameters based on the measured device response.

[0106] Some of the specific implementations described in this paper use perturbation analysis to achieve this matching between the feedback compensator response and the device dynamics. Perturbation analysis employs a linearization technique using a system of nonlinear control equations and assumes that solutions close to known nonlinear solutions can be discovered by employing small linear steps or perturbations from the known solutions. In this example, there exists a device model and a matched compensator—both of which can be modeled as a product of nonlinear rational functions, and which are the product of these two functions that define the loop gain of the feedback system.

[0107] While not intending to be bound by theory, the following example of perturbation analysis begins by assuming that the function of interest can be written as a nominal solution plus a small additive deviation (indicated by Δ, for terms close to 0). In this case, we focus on G. sd and K fb Make the following assumptions:

[0108]

[0109]

[0110] Among them, the underlined items (e.g., and ) represents the nominal solution, and the Δ term (e.g., ΔG) sd and ΔK fb This represents a small deviation from the nominal solution. Therefore, the loop gain (complex cyclic response) LG can be defined as:

[0111]

[0112] item Corresponding to the nominal loop gain, where item This corresponds to the contribution to the loop gain deviation caused by variability between different ears / wearing styles. (Item) This corresponds to the contribution to the loop gain deviation caused by the customization of the feedback compensator. (Item) This can be ignored because it is the product of two smaller terms. For terms expanded in this way, this indicates that due to the small change ΔG in the microphone response of a single driver... sd This causes the loop gain of any particular fit to deviate from the nominal value; however, it can also be compensated for by a small change ΔK in the compensator. fb This is to change the loop gain. Therefore, in some specific implementations, the following conditions can be imposed:

[0113] ΔLG=0

[0114] This leads to the following relationship between the nominal parameters and the disturbance parameters:

[0115]

[0116] The following are examples of system parameterization consistent with the assumption of small linear variation, including examples that satisfy the above equations for customization of feedback compensators.

[0117] The above example, based on the linear perturbation representation, is based on the product of two nominal functions with small linear deviations. Alternative representations of perturbation analysis can be expressed using perturbations with multiplicative deviations. For example, the measured driver response G to a microphone... sd The multiplicative bias (indicated by δ, for terms close to 1) can be expressed as the nominal response cascaded with a specific fit Δ factor:

[0118]

[0119] If a nominal compensator is used to measure the loop gain, the measured loop gain is:

[0120]

[0121] Furthermore, the deviation between the measured loop gain and the nominal loop gain can be verified using LG| meas Dividing by the nominal loop gain, as shown below:

[0122]

[0123] At this point, the measured loop gain deviates only from the target because the G value worn by this particular headset... sd Deviation from nominal G sd Essentially the same quantity. Then, the target can be adjusted by modifying the compensator to drive the loop gain back to the nominal target, ensuring that the final loop gain Δ is uniform. This can be achieved by adjusting δK using the multiplicative transfer function. fb This can be achieved by adjusting the compensator, as shown below:

[0124] δLG| final =δLG| meas δK fb =δG sd δK fb ≡1

[0125] This leads to:

[0126]

[0127] Alternatively, when operating on quantities in the logarithmic space, Furthermore, the multiplication deviation can be expressed as based on log... 10(δX) = ΔX, the additive deviation, which can be expressed as:

[0128]

[0129] Therefore, the bias (or correction) of a custom compensator can be substantially reversed (or subtracted in logarithmic space) by the bias introduced by the deformability of the actual ear response. The nominal compensator can be implemented, for example, as a relatively low-order filter (e.g., using about 4 to 7 double second-order filters). Custom compensator K fb According to the nominal compensator Adjustment causes the change in its transfer function to reverse the response of the device to G. sd The following example illustrates the linear perturbation techniques that can be used to calculate these adjustments.

[0130] This example parameterizes the compensator by defining parameters characterizing the stages and zeros of N bisecond-order filters cascaded together (e.g., series multiplication) to form a complete compensated filter. Each of these N bisecond-order filters (labeled BQ1 to BQN) is characterized by two poles (e.g., complex pole pairs) associated with the pole frequencies, and a zero frequency Z. f Two related zeros. A characteristic of the filter may lie in the ratio between these frequencies. and center frequency f c There also exists Q—a factor characterizing the filter shape: extreme Q—a factor P. Q And zero Q-factor Z Q Therefore, each bisecond-order filter can be characterized by a different parameter set BQi (for i = 1 to N), where:

[0131]

[0132] Furthermore, the following expression represents the parameter vector, which is formed from a set of parameters of each of these N bisecond-order filters:

[0133] Γ j =[BQ1,…,BQN] T

[0134] In other specific implementations, the parameters characterizing a given bisecond-order filter can be different. For example, the chosen parameters could be the pole and zero frequencies themselves and their associated Q factors, instead of the four parameters mentioned above, or they could be quadratic coefficients directly used for digital implementation of the bisecond-order filter, and other possibilities. Other filter representations besides bisecond-order can also be used, each representing a frequency response with its own specified parameters.

[0135] Given a nominal parameter vector It generates a nominal compensator We can digitally perturb each parameter ΔΓ j The quantity, and the resulting change in the compensator response:

[0136]

[0137] Custom compensators and nominal compensators can be calculated as functions of the disturbance parameter vector and the nominal compensator, respectively:

[0138] K fb =F(Γ) j )

[0139]

[0140] We now have two compensators, each of which represents a realizable filter with only minor differences, defined by ΔΓ in the basic parameterization. j Limitations. These minute differences are important because they are needed to correct for G. sd Changes in the ear. Figures 4A to 4D This illustrates how a compensator with a single second-order filter stage can be modified by changing a single parameter, which in this example is the center frequency f. c . refer to Figure 4A This illustrates the shape 400 of the absolute amplitude (in logarithmic space) of the nominal compensator, where the filter shape shown on either side changes as the center frequency decreases or increases. Reference Figure 4B The figure shows the shape 402 of the phase (in logarithmic space) of the nominal compensator, wherein the filter shape shown on either side changes as the center frequency decreases or increases. Figure 4C and Figure 4D The relative amplitude and phase characteristics are shown, which are the result of dividing each of these amplitude and phase curves by the nominal amplitude and phase, respectively. Therefore, the flat relative amplitude response shape 404 corresponds to the amplitude response shape 400 divided by itself; and the flat relative phase response 406 corresponds to the phase response shape 402 divided by itself. The relative amplitude and phase responses, and these flat responses, for changing the center frequency relative to the nominal compensator are shown. The difference between either the perturbation filter and the nominal filter (which is a nonlinear function of the variation of specific parameters) can be calculated as:

[0141]

[0142] The aforementioned equation provides a construction for determining the incremental change in a compensator used to compensate for the deviation of a single ear response from the nominal value. However, to achieve this construction, we can compare the desired frequency response change (typically described as the ratio of the Fourier transforms in terms such as amplitude and phase) with a correction ΔΓ for the filter parameters. j Related, it specifies the poles and zero or coefficients of the bisecond-order filter through some parameterization. Filter parameters Γ jj With filter response K fbThe exact relationship is nonlinear. However, the ANR circuitry of the handset can be configured to perform a perturbation calculation that linearizes around small parameter variations to approximate this nonlinear relationship, rather than an exact nonlinear calculation. For example, a vector of partial derivatives of amplitude and phase can be used to calculate a specific frequency f. i Due to the change of a specific parameter ΔΓ j The resulting changes in the compensator response are as follows:

[0143]

[0144] Customization processes can include evaluating complex responses on the right-hand side of the relationship to describe how variations in filter parameters alter the magnitude and phase responses of a given nominal filter response. While these partial derivatives can be evaluated analytically without sacrificing too much accuracy, various approximations of the partial derivative calculations are possible. In this example, the partial derivatives of the compensator response with respect to individual compensator parameters are estimated via first-order finite differences.

[0145] There may exist many parameters, each changing only slightly. Using this linearization, the total change in amplitude at a given frequency can be expressed as the sum of the contributions of all individual changes in the parameters, where it is assumed that the amplitude of the compensator response is expressed in logarithmic space, thus yielding a relationship between additive biases:

[0146]

[0147] A similar relationship exists in the phase. Therefore, we can evaluate the vector at the M frequency point. The changes in amplitude and phase due to variations in N small parameters are as follows:

[0148]

[0149] This is the formula used to calculate the magnitude (in the top row) and phase (in the bottom row) of a linear system using a matrix (called the "influence matrix"). Each row represents the effect of all small variations in the compensator parameters on the response at a single frequency, and each column represents the effect of a single parameter variation across all frequencies in a selected set of frequencies. The equation can be expressed more compactly as:

[0150]

[0151] The custom module can be programmed to apply a solver that calculates small adjustments to the compensator to offset variations in a specific fit. These adjustments can be estimated using a custom audio signal supplied to the earpiece driver and measured by the earpiece microphone used to calculate the ear's frequency response, as described in more detail below. δK fb (and logarithmic space) The above equation indicates that the ideal compensator adjustment can be obtained as the inverse function of the device response change. This equation can be used to derive the relationship between the compensator parameters and the compensator response at a set of discrete frequencies, as shown below (using logarithmic space formulas):

[0152]

[0153] Custom modules can evaluate at the same frequency point set. Any given fitted ΔG at (used to construct the influence matrix) sd Furthermore, the changes in compensator parameters that satisfy this set of equations at all these frequency points can be solved. This can be achieved by reversing the influence matrix, which produces:

[0154]

[0155] In ΔG sd The number of parameters and ΔΓ jj When the number of parameters in the filter changes, the inverse function becomes a pseudo-inverse function, which provides the least-squares optimal solution for the incremental changes of the filter parameters.

[0156] Determining the influence matrix involves a significant amount of computation, as does its pseudo-inverse. However, this only needs to be done once for a given nominal feedback compensation filter and the inverted influence matrix stored in a custom module. The custom module is then able to measure the deviation of the ear response to calculate the necessary compensator adjustment relative to the nominal compensator, driving that particular fit to the target loop gain response with a single matrix multiplication. The process of performing an FFT on the measured signal is efficient, which determines G. sd The nominal change is then multiplied by the vector by a pre-determined and stored inverse effect matrix; this can be done in a processor (such as an ARM chip used in wearable products within a second).

[0157] Figures 5A to 5G The results of the perturbation solution in a custom feedback system are explained for a system with fairly high acoustic potential cancellation (up to about 2 kHz). Figure 5A It shows a fixed feedback compensator K fb A set of training ear / wear loop responses (loop gain and phase) is provided. This fixed feedback compensator is designed to achieve a feedback loop high-frequency gain crossover of approximately 2kHz, with the goal of eliminating the full potential of the headphone acoustics. However, note that the phase of the loop response is close to 0 degrees at the gain crossover, indicating a system with poor feedback stability margin. Figure 5B The training system is shown to customize K for each wear. fb The result. Figure 5BThe circular markers on the frequency axis of the amplitude curve in the diagram define a set of frequencies that constrain the rows of the influence matrix. Note that a loop amplitude division close to 2 kHz is achieved, the amplitude variation range at each frequency is reduced, and the average phase under amplitude division is approximately 45 degrees. This is based on a system with good phase margin (good stability) on the amplitude and phase curves (also known as Bode plots). Figure 5C It shows a fixed K with a modified (i.e., demodulated) configuration. fb To achieve a good stability margin for the same system. However, it should be noted that this applies to K. fb Demodulation sacrifices performance, with an average amplitude crossover of approximately 900Hz. Therefore, for these high-potential-cancelling headphone acoustics, because of the G in the ear... sd The effects of variation (especially at higher ratings) mean that the limitations of fixed feedback compensators can be eliminated.

[0158] Figures 5D to 5F This describes the performance of the same system (observed from its closed-loop performance): feedback loop sensitivity,

[0159]

[0160] Sensitivity is the feedback noise cancellation (feedback insertion gain) measured at the feedback microphone; for a system with sufficiently high potential cancellation, this high potential cancellation approximates the feedback insertion gain (FBIG), as measured in the ear canal. In the sensitivity graph, negative dB values ​​correspond to cancellation, and positive dB values ​​correspond to noise amplification. Values ​​greater than 10 dB to 15 dB indicate that the system is close to oscillation. Figure 5D It shows Figure 5A The stability of the fixed K is relatively poor fb System sensitivity; Note that although the average sensitivity (dashed line) is stable, for many wearers (gray / dotted line), the peak sensitivity ranges from 10dB to 20dB. Figure 5E It shows Figure 5C Good stability of fixed K fb System sensitivity; it should be noted that although the peak sensitivity does not exceed 5 dB for all wearers, the average sensitivity (dashed line) has essentially the same cancellation performance as a system with aggressive average but poor stability (dashed line) – a difference approaching 10 dB at some frequencies. Finally, Figure 5F It shows Figure 5B The sensitivity of the custom Kfb system; it should be noted that the system has good stability (the gray individual wear curve hardly exceeds 5dB), and the average sensitivity (solid black line) has a sensitivity crossover frequency close to 2kHz, and the potential cancellation is generally superior to the well-stable fixed Kfb system. fb System (dashed line).

[0161] One benefit of the increased feedback loop bandwidth achievable from custom-designed high-potential-cancellation headphones is the improvement in the blocking effect, where the amplification of the wearer's voice due to vibrations conducted through the body is coupled into the blocked ear canal. For headphones that are shallowly sealed into the ear canal (at or near the canal opening), blocking is observed at frequencies below approximately 1.5 kHz. For feedback noise cancellation systems, the amplified sound originating from the body's blocking is the noise to be canceled. For headphones with... Figure 5C and Figure 5E The high-potential elimination system of the stable fixed compensator has a feedback loop bandwidth that extends only to 900Hz; this results in a slight amplification of a person's voice when they speak while wearing headphones. For Figure 5B and Figure 5F The custom design shown has a feedback bandwidth extended by more than 1.5kHz, which essentially improves the wearer's voice and therefore gives it a clearer feel when in a "perceived" state.

[0162] For a fixed feedback compensator system, the sensitivity variation in the elimination band at various frequencies (frequency where the loop gain is greater than 0 dB) will essentially be the device response G. sd The change in K. This is evident from the equation for sensitivity, taking into account K. fb It is fixed. Because the sensitivity approximates the feedback insertion gain at the ear, the observable characteristic of headphones implementing the technique described herein is that the variation in both sensitivity and feedback insertion gain in the elimination band is reduced compared to variations in the device's acoustics. Figure 5G This is illustrated in the system shown. Figures 5A to 5F There are different K fb Response. In Figure 5G In the diagram, the dashed lines represent the standard deviation of the device's acoustic performance at various frequencies relative to the wearing experience. The dashed lines represent a fixed K-line with good stability. fb The standard deviation of the system's sensitivity; note that from 30Hz to 500Hz, the change in sensitivity is essentially the same as the change in the device's response. The solid line represents the custom K... fb The standard deviation of the system; it should be noted that the variation in most cancellation bands is half or less compared to the variation in the acoustics of the basic device.

[0163] Although the above example describes the feedback compensator K fb While customization is possible, a similar approach can be used to determine a perturbation-based custom feedforward compensator K. ff (For elimination or perception modes). The equations for IG given above can be solved. Given a target IG such as 0 (elimination) or 1 (perception), for K... ff According to K fbThis is achieved through various acoustic responses measurable at the microphones in the headphones and the subject's ear canal. The latter is possible in the laboratory as part of obtaining the training dataset. The resulting K... ff The solution is N so / G sd The product of terms includes factors related to the feedback system and response to the system microphone and ear microphone signals. The latter term can be averaged over the training data. Therefore, as long as the perturbation method modifies K according to the nominal response... fb Customized G sd To achieve a wider bandwidth with more consistent (fewer variations) and better execution of the feedback loop response, the same approach (using a pseudo-inverse function of the influence matrix determined from the training dataset through a computationally intensive and rigorous offline process) can be used to modify K according to the nominal response. ff Thus, N can be customized. so / G sd The changes result in a wider bandwidth and better execution of total insertion gain (a combination of passive, feedback, and feedforward modes).

[0164] Using the techniques described in this paper to customize the feedback compensator can result in active insertion gain, combining the effects of both the feedback and feedforward systems, where the bandwidth sink exceeds 2kHz, such as... Figure 5H As shown in the diagram. When combined with additional bandwidth derived from a custom feedforward compensator, the combined active insertion gain bandwidth can exceed 2kHz, as also... Figure 5H As shown. A drawback of active noise-cancelling headphones is that they initially have an active insertion gain crossover (the frequency at which 0dB is canceled) below the frequency at which the passive insertion gain plateaus, thus creating a "hole" in the total insertion gain at mid-frequency. This is not the case for additional bandwidth derived from custom designs. Therefore, as... Figure 5I As shown, at these intermediate frequencies, a total insertion gain of over 30 dB can be important for reducing broadband noise and distracting speech.

[0165] In some implementations, feedforward customization for a given earphone / wear is performed after feedback customization for that earphone / wear, and the results of the feedback customization for that ear / wear are used. This is desirable because feedback customization provides a more consistent system basis for the feedforward system. Alternatively, the results of previous feedback customization for previous earphones / wear for the same user can be used for feedforward customization for a given ear / wear.

[0166] After a suitable nominal dataset, including nominal functions and parameter values, has been calculated through an offline design process, the nominal dataset is loaded into the earpiece's memory or another part of the wearable device to which the earpiece can be accessed. A relatively small amount of memory can be used to store the nominal dataset, which may include functions and parameters evaluated at a relatively small number of discrete frequencies, as well as an inverted effect matrix. Optionally, to enable operation before any customization occurs or when customization capabilities are disabled, the memory may also store default filter parameters for feedback and / or feedforward filters that may differ from the nominal parameters. For example, while the nominal parameters will be adjusted to ensure that they satisfy various constraints (e.g., stability constraints) for a given fit, in most cases, default parameters may be selected to ensure that they satisfy those constraints that may occur for a given user for any of the various potential fits.

[0167] Figure 6 A flowchart of an exemplary control program 600 is shown, which the customization module uses to determine when to execute a customization program for customizing a feedback compensator, a feedforward compensator, or both. After the earpiece is powered on (e.g., when the wearable device is powered on), the control program 600 is in a wear-sensing state 602, in which the customization module can sense that the earpiece is being worn by sensing that it is placed in, on, or around the ear, thus preparing for fitting measurement. This sensing may be performed, for example, using one or more sensors (e.g., skin touch sensors, proximity sensors, optical sensors, motion sensors, acoustic sensors, and / or pressure sensors). The control program 600 enables the customization module to measure the acoustic characteristics of the earpiece (which has been placed in the ear during wear) in a single ear 604 by playing a customized audio signal via the earpiece driver and recording the response signal sensed at the feedback microphone of the ANR circuit, which is then used to trigger the customization of the feedback compensator. The customized tone may be output independently in each earpiece (e.g., the right and left earpieces), and the playback of the tone may be synchronized so that they play substantially simultaneously. In some cases, custom tone can also be used to confirm that the user has a sufficient quality fit or seal between the earpiece and his or her ear in order to proceed with customization.

[0168] A customized audio signal can be designed as a relatively short acknowledgment sound played through the audio driver of each earpiece of the wearable device. This acknowledgment sound serves as an indication to the user that the earpiece of the wearable device has been worn as intended. To provide an appropriate measurement of the acoustic environment formed when the earpiece is worn, the spectrum of the customized audio signal can be shaped to include a sufficient amount of energy to be used by the feedback customization process at a predetermined set of frequencies, selected based on the acoustic characteristics of headphones in a typical ear (e.g., frequencies that characterize resonance and their maximum and minimum values). The user may not necessarily know that a measurement will be performed, but simply hearing the acknowledgment sound can be considered a normal part of the experience of wearing the wearable device. For example, the acknowledgment sound could be the “startup” tone heard by the user when first wearing and powering the wearable device.

[0169] In the example of the custom audio signal, the duration of the signal can be relatively short (e.g., less than one second, or between about one-tenth of a second and about half a second), and the spectrum of the signal can include peaks at frequencies focused on harmonics corresponding to a fundamental low-frequency tone focused on the fundamental frequency. Therefore, this fundamental frequency can be chosen to correspond to the lowest frequency in a set of frequencies used by the custom program (e.g., 46.875 Hz). The next tone in the spectrum can be focused on frequencies that are higher harmonics (i.e., integer multiples of the fundamental frequency), with the spacing between these higher harmonics increasing approximately linearly for the first few harmonics, and then gradually increasing in multiple orders, but not necessarily monotonically (e.g., multiples of 2, 4, 6, 8, 12, 16, 18, corresponding to frequencies of 93.75 Hz, 187.5 Hz, 281.25 Hz, 375 Hz, 562.5 Hz, 750 Hz, 843.75 Hz). As the frequency increases, the energy level of each tone may decrease (e.g., gradually decreasing relative to a logarithmic magnitude scale), but not necessarily monotonically. Higher frequency tones in the spectrum (e.g., tones above 1 kHz) may occur near approximate multiples of the fundamental frequency, but may not be as precise as at lower frequencies. For example, due to relaxed constraints at higher frequencies, there may be some flexibility regarding the exact value of the center frequency of the tone relative to the exact value of the high-frequency harmonics of the fundamental frequency. The order between higher frequencies may also increase non-linearly (e.g., exponentially, or according to the logarithm of the frequency), but not necessarily through a constant function (e.g., high-frequency tones may be concentrated at 1031.3 Hz, 1218.8 Hz, 1500 Hz, 1781.3 Hz, 2156.3 Hz, 2531.3 Hz, 3000 Hz, 3562.5 Hz, 4218.8 Hz, 5062.5 Hz, 6000 Hz, 7125 Hz, 8531.3 Hz, 10125 Hz, 12000 Hz, 14250 Hz, 16969 Hz). In some implementations, there may be a preferred spacing between higher frequencies (e.g., a quarter-octave spacing may be used). Alternatively, at higher frequencies, low-amplitude sinusoidal sweeps or band-limited bursts of pink noise may be used. For example, a high-frequency spectrum with band-limited frequencies greater than about 1 kHz and a relatively wide bandwidth (rather than a single tone with a peak at the selected frequency) above 1 kHz may be used.

[0170] While a custom audio signal is played through the driver of each earpiece, a feedback microphone in each earpiece is used to receive a response signal, a sensed version of the custom audio signal, influenced by acoustic features resulting from the combination of the earpiece with the size, shape, and tissue characteristics of a single ear canal. For each earpiece, a sample of the received time-domain response signal can be stored in memory as a measure of these features. The customization module then uses the measured actual ear response data to perform a feedback customization procedure 606, as described in more detail below. After the feedback compensator has been customized, the control procedure 600 enters a noise sensing state 608. The customization module monitors the sound level of noise sensed by the feedforward microphone to determine whether to initiate the customization of the feedforward compensator. If the sound level is low (i.e., below a predetermined threshold), the control procedure 600 remains in the noise sensing state 608. If the external sound level is not high enough, there may not be enough information in any recorded signal. Furthermore, if the external sound level is not high, customized feedforward performance may be less necessary. If the sound level is high (i.e., above a predetermined threshold), control procedure 600 enables the customization module to record 610 noise, such as noise present in the external acoustic environment via the feedforward microphone of the ANR circuit, and noise present in the internal acoustic environment via the feedback microphone of the ANR circuit. In some examples, the customization of the feedforward compensator may not occur until the system detects an audio signal not being played through the electroacoustic transducer in the earpiece and / or the user is not speaking, except to wait until the external sound level is sufficiently high. The customization module may store samples of the two recorded signals for a given earpiece in the earpiece's memory. The duration of the recorded signals may be relatively short (e.g., less than one second or about half a second) or may be averaged over longer time intervals in various time or frequency domain manners to improve measurement quality. When recording a signal sensed at the microphone, for open-loop measurements, there is no signal played through the earpiece driver, or for closed-loop measurements, a predetermined signal is played through the driver. In addition to detecting the ambient sound level (as part of the decision-making process during noise sensing), the noise sensing state 608 also checks the signal levels at both the feedback microphone and the feedforward microphone, and also checks the accelerometer in the earphone to determine if the wearer is speaking. Noise recording 610 is best avoided when the wearer is speaking because the blocking effect results in a high signal level at the feedback microphone, and therefore does not characterize the sound N entering the ear through and via the earphone. so The transmission and recording achieve the desired accuracy. Whether the wearer's speaking state needs to be considered also depends on the noise level, the acoustic design of the headphones, and the feedback operation status during recording.

[0171] In examples where external sound levels are insufficient to trigger a customized feedforward compensator, the user can be instructed to generate noise in an environment where he or she can generate noise. For example, the user can be instructed to generate audio from an external device such as a telephone, home speaker, portable speaker, or home theater system. This audio may contain background noise with spectral content sufficient to customize the feedforward compensator.

[0172] After recording the noise signal, the customization module uses the recorded noise signal to perform feedforward customization calculation 612; this may also include incorporating the previously measured and calculated factor (G) from the feedback customization. sd and K fb This is taken into account. However, in some cases, the step of customizing the feedforward compensator may not be performed. In this exemplary embodiment, control program 600 checks whether the measure of the relative change between the resulting custom feedforward compensator parameters and the currently loaded feedforward compensator parameters is sufficiently large by comparing the measure with a predetermined threshold 614. If the measure is higher than the threshold, control program 600 enables the customization module to perform the feedforward customization procedure 616 using the result of the feedforward customization calculation 612. If the measure is not higher than the threshold, control program 600 does not change the currently used feedforward compensator parameters. This ensures that the user does not unnecessarily experience changes in ANR performance. Alternatively, the customization module may accumulate the results of noise recording and related data, such as N as a function of time and even multiple headphone wear. so / G sd The different forms of measurement results. The earpiece can then analyze the statistical data of these measurement results in various ways (including averaging) to determine the significance of K. ff The continuously improving estimates are optimal for the wearer.

[0173] After determining whether to apply any triggered feedforward customization, control program 600 enters wear-sensing state 618, in which the customization module can sense that the earpiece has been removed by sensing that the earpiece is no longer placed in, on, or around the ear. This sensing can be performed, for example, using one or more sensors (e.g., skin touch sensors, proximity sensors, motion sensors, acoustic sensors, and / or pressure sensors). If the earpiece has not been removed, control program 600 remains in wear-sensing state 618. When the earpiece is worn, control program 600 returns to wear-sensing state 602 to perform new customization for the new user and / or new fit. In some cases, if the user removes the earpiece for less than a threshold amount of time (e.g., a few seconds), new customization may not be triggered. For feedforward customization, the earpiece may optionally trigger new customization when the user is in an environment where feedforward customization can be improved by automatically triggering customization or prompting the user to perform customization.

[0174] This exemplary control program 600 is merely one example of a customized technique for initiating feedback compensators and / or feedforward compensators. In an alternative example, in addition to G obtained from 604... sd Aside from the additional step of comparing the stored value determined in previous measurements, the procedure shown in 600 can be followed, and if the value sufficiently matches the previously stored value (acoustic “earprint”), the previously determined compensation filter can be used instead, thus eliminating the need for additional measurements and filter customization. In a second alternative, an associated app or voice prompt issued by the customization module can guide the handset owner (after the product has been purchased) to perform a series of measurements to obtain a compensation filter for that user; these filters are then stored for use in all subsequent use sessions. In this second alternative, the initiation of measurements can be manually triggered, and additionally, the “earprint” can also be used to trigger the use of the stored compensation filter. In any of these examples, if the product is in use and oscillations in the feedback system are detected by some means, the system can very quickly switch to an open-loop mode of operation and trigger measurements. Other alternative sequences of steps for customizing the system are possible, as are various combinations of the above alternatives.

[0175] Different implementations of the customized program may perform different steps and / or different calculations depending on whether the feedback compensator or the feedforward compensator is customized.

[0176] In some implementations, other forms of input can be used to trigger custom procedures or other adjustments to the loop gain or other characteristics of the ANR circuit. For example, adjustments may be made in response to the detection of an outbreak of instability in the feedback loop or in response to the detection of a significant pressure change, which could be an indication of a significant change in the earpiece fit. As another example, the target loop gain may be reduced when a worse seal than typical or expected is detected.

[0177] Different implementations of the customization process may involve different steps and / or different calculations depending on whether the wearable audio device has a headset configured to be worn in, on, or around the ear. For example, for a customization process for on-ear or around-ear fits, there may be relatively more focus on modifying the compensator at lower frequencies due to leakage associated with poor fits in on-ear or around-ear fits (which may primarily affect relatively low frequencies). Alternatively, for a customization process for in-ear fits, there may be additional focus on modifying the compensator at higher frequencies due to fit variations resulting from tighter fits with different ear canal sizes and / or shapes (which may primarily affect relatively high frequencies). In some implementations, for any of the in-ear, on-ear, or around-ear fits, the compensator may be customized over a relatively wide frequency range (e.g., 20 Hz to 10 kHz), extending above and below the gain crossover frequency of the feedback loop. For example, customization of the compensator can modify one or more parameters associated with one or more frequencies below the high-side gain divider frequency (where the magnitude of the loop gain associated with the ANR signal path (i.e., the feedback path or feedforward path) is approximately equal to one), and modify one or more parameters associated with one or more frequencies above the high-side gain divider frequency. Customization can also allow the gain divider frequency to be relatively high, thereby producing a stable feedback loop over a wide frequency range. For example, in the case of customization, the low-side gain divider frequency can be about 20 Hz, and the high-side gain divider frequency can exceed 1 kHz (e.g., about 2 kHz or about 3 kHz). Without customization, the high-side gain divider frequency can be intentionally limited to below about 800 Hz or 700 Hz to ensure stability for various users and / or adaptations.

[0178] In some specific implementations, the number of parameters of the customized compensator is relatively large. For example, for a feedback compensator implemented using cascaded bisecond-order filters, there may be three, four or more bisecond-order filters, resulting in 12 or 16 or more parameters in the parameter vector (assuming each bisecond-order filter is characterized by at least 4 parameters), thereby achieving a significant level of customization of the ANR.

[0179] Wearable devices can also be configured to use customized information, such as filter parameters obtained from a customization process, for a variety of purposes. For example, because the expected feedback filter parameters are different for different users and relatively consistent for a particular user, the customized feedback filter information can be used to identify or authenticate the user if the user wears the device with the earpiece in a specific manner. The feedback filter parameters may be related to the device G. sdThe nominal deviation can be used as a reference or for calculating or finding an identification code. While measurements from a single earpiece can be used, combining parameters from the left and right ears (which are not identical) increases the uniqueness of this "earprint." G of the left / right earprint combination. sd The filter parameters can also be combined with other information (e.g., the formal structure of the wearer's voice when speaking or saying their name) to further enhance the uniqueness of user identification. In response to a specific identification code, the audio characteristics of the wearable device can be tuned (e.g., for specific equalizer settings, or for preloading specific filters or altering some other mode of headphone operation). The identification code can also be used by means such as via a Bluetooth link to uniquely identify the user to unlock other systems (such as the user's computer, servers) and to unlock doors and vehicles.

[0180] The described custom procedure is computationally feasible because it is sufficient to store the inverted (or pseudo-inverted) influence matrix in addition to the response measurement results. The calculation of determining the nominal filter and the inverted influence matrix is ​​performed offline, and this calculation may involve time-consuming and computationally intensive methods. Alternatives to this method are possible. In one alternative, measurements performed by a linear perturbation method can be performed out of the box, manually triggered by the user and guided by an app or voice prompts. These measurements can be uploaded to a server with standard accessories and optimization tools (such as those available in Mathworks' Signal Processing Toolbox) to determine the compensators to adjust the measured acoustics to achieve the target performance. These filters can then be downloaded from the server and stored in the product for later use. This process can be performed by multiple people sharing headphones, where each of those filters determined by server-based calculations is stored for selection based on earprints measured while wearing the headphones. A second alternative abandons the linearization of the relationship between filter parameters and changes in amplitude and phase used in the perturbation method. Instead, a nominal compensation filter K is determined. fb and K ff(The system designers believe these are optimal for the headphones), and the parameters defining those filters can be changed, with the corresponding changes in amplitude and phase determined beyond the accuracy of a linear approximation. A multidimensional nonlinear surface can then be adapted to relate the nominal amplitude and phase changes (as independent variables) and the changes in filter parameters (as dependent variables). The equations describing this surface can then be stored in a custom module for customizing the filter with each wear. A third alternative, given a nominally compensated filter best suited for the headphones and a large training dataset (consisting of different filter parameters and corresponding changes in filter responses (amplitude and phase)), is used to train a deep neural network (DNN) to predict changes in filter parameters from changes in response. Once trained, the DNN can be implemented in a custom module to determine the custom filter from the response for a given wear measurement. In recent years, DNNs have shown great utility in modeling systems with previously difficult-to-determine mathematical solutions. The advantage of applying DNNs to this problem is that the dataset (filter changes and corresponding response changes) required to train the DNN can be arbitrarily large and spans deviations in filter parameters that are more difficult to handle than linear perturbation methods can handle.

[0181] While the examples described herein include a single feedback microphone and a single feedforward microphone for each earpiece, additional feedback microphones and / or feedforward microphones may be used in other examples. ANR circuitry may be included in the earpiece (e.g., for wireless earbuds) and / or in a wired control module (e.g., for wired earbuds), or in a remote module communicating with one or both earpieces (e.g., via a wired or wireless link). Any or all ANR circuitry may be implemented using dedicated hardware modules and / or a processor configured to perform software stored on a non-transitory computer-readable medium (for performing any calculations of the ANR circuitry), and the circuitry may be configured as described, for example, in U.S. Patent Publication 2013 / 0315412 and U.S. Patent Publication 2016 / 0267899, each of which is incorporated herein by reference.

[0182] While this disclosure has been described in conjunction with certain examples, it should be understood that this disclosure is not limited to the disclosed examples, but rather is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims, which will be given the broadest interpretation to cover all such modifications and equivalent structures as permitted under the law.

Claims

1. A method for managing features of active noise cancellation, comprising: Receive a first input signal captured by one or more sensors associated with the active noise-canceling (ANR) headphones; The frequency domain representation of the first input signal for a set of discrete frequencies is calculated by one or more processing devices; The one or more processing devices generate a set of parameters for a digital filter disposed in the ANR signal flow path of the ANR headphones based on the frequency domain representation of the input signal, the set of parameters such that the loop gain of the ANR signal flow path substantially matches the target loop gain, wherein generating the set of parameters includes: The response of the digital filter is adjusted at frequencies spanning at least between 200 Hz and 5 kHz; and Adjusting the response of at least three second-order fundamental sections of the digital filter; and The generated parameter set is used to process the second input signal in the ANR signal flow path to generate an output signal for driving the electroacoustic transducer of the ANR headphones.

2. The method of claim 1, wherein the first input signal includes features that vary with different users, and the second input signal includes features that vary less between different users compared to the first input signal.

3. The method of claim 1, wherein the one or more sensors include a feedback microphone of the ANR headphones, and the ANR signal flow path includes a feedback path disposed between the feedback microphone and the electroacoustic transducer.

4. The method of claim 3, wherein for most of the frequency range in which the feedback path has a positive loop gain, the variation in feedback insertion gain measured by the plurality of users is less than the variation in the physical acoustic response of the ANR headphones measured by the response between the feedback microphone and the electroacoustic transducer for the plurality of users.

5. The method of claim 4, wherein for most of the frequency range in which the feedback path has a positive loop gain, the change in the feedback insertion gain is at least 10% smaller than the change in the physical acoustic response of the ANR headphones.

6. The method of claim 3, wherein the average feedback insertion gain measured by multiple users has a high-frequency division of greater than or equal to 1.5 kHz.

7. The method of claim 1, wherein generating the parameter set comprises: Access the nominal parameter set of the digital filter. The set of correction parameters is determined based on the frequency domain representation of the first input signal, and The parameter set is generated as a combination of the corresponding parameters in the nominal parameter set and the calibration parameter set.

8. The method of claim 7, wherein the nominal parameter set is calculated based on training data including multiple ear responses.

9. The method of claim 8, wherein the nominal parameter set is generated by performing an optimization process configured to generate a corresponding ear response.

10. The method of claim 9, wherein determining the set of correction parameters comprises: Calculate the loop gain of the nominal parameter set of the digital filter; Generate an error vector, which includes the deviation between the loop gain and the corresponding target loop gain at different frequencies; as well as The set of correction parameters is generated based on the statistical data of the training data and serves as the output of the optimization process.

11. The method of claim 1, wherein when ANR is activated, the total insertion gain of the ANR headphones is less than -30 dB in the frequency range of 1 kHz to 2 kHz.

12. The method of claim 1, wherein the average active insertion gain measured by a plurality of users has a high-frequency division of greater than or equal to 2.2 kHz.

13. The method of claim 1, wherein the parameter set is generated within 1 second of receiving the first input signal.

14. The method of claim 1, further comprising storing a generated set of parameters for identifying or authenticating users.

15. The method of claim 1, wherein: The first input signal is captured in response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones. The audio signal comprises a broadband signal, which includes energy at multiple frequencies from the set of discrete frequencies. The frequency domain representation of the first input signal indicates the ear's response to the audio signal.

16. The method of claim 15, wherein the audio signal has a spectrum comprising 10 or more tones concentrated at predetermined frequencies between 45 Hz and 16 kHz.

17. The method of claim 16, wherein the predetermined frequency comprises a plurality of frequencies higher than 1 kHz, the spacing between the plurality of frequencies being less than or equal to 1 / 4-octave.

18. The method of claim 15, wherein the audio signal is automatically delivered in response to detecting that the ANR headphones are positioned in, on, or around the user's ear.

19. The method of claim 15, wherein the audio signal is automatically delivered in response to the detection of oscillations in the ANR signal flow path.

20. The method according to claim 1, wherein: The one or more sensors include the feedforward microphone and the feedback microphone of the ANR headphones. The first input signal includes the ratio of the feedback microphone signal to the feedforward microphone signal, and The ANR signal flow path includes a feedforward path disposed between the feedforward microphone and the electroacoustic transducer.

21. The method of claim 20, wherein the feedforward microphone signal is captured in response to determining that ambient noise near the ANR headphones is higher than a threshold.

22. The method of claim 21, wherein the feedback microphone signal is captured in response to the delivery of an audio signal via the electroacoustic transducer of the ANR headphones, the audio signal comprising a broadband signal, the broadband signal comprising energy at a plurality of frequencies in the set of discrete frequencies.

23. The method of claim 20, wherein the feedforward microphone signal is captured in response to determining that ambient noise near the ANR headphones is higher than a threshold, and the following are detected: (i) a lack of audio signal played through the electroacoustic transducer; and (ii) a lack of user speech.

24. The method of claim 20, wherein one or both of the feedforward microphone signal and the feedback microphone signal are repeatedly captured in units of each of a plurality of time intervals.

25. The method according to claim 1, further comprising: The seal quality between the ANR headphones and the wearer's ear is measured, and the target loop gain is reduced when the seal quality is less than a predetermined threshold.

26. A method for managing features of active noise cancellation, comprising: Receive a first input signal captured by one or more sensors associated with the active noise-canceling (ANR) headphones; The frequency domain representation of the first input signal is calculated by one or more processing devices; The one or more processing devices generate a set of parameters for a digital filter disposed in the ANR signal flow path of the ANR headphones based on the frequency domain representation of the input signal, the set of parameters such that the loop gain of the ANR signal flow path substantially matches the target loop gain, wherein the generated set of parameters includes: A first parameter associated with a first frequency in a set of discrete frequencies, the first frequency being less than the high-end gain division frequency, at which the magnitude of the loop gain associated with the ANR signal flow path is equal to 1, and A second parameter associated with a second frequency in the set of discrete frequencies, wherein the second frequency is greater than the high-end gain divider frequency; as well as The generated parameter set is used to process the second input signal in the ANR signal flow path to generate an output signal for driving the electroacoustic transducer of the ANR headphones.

27. The method of claim 26, wherein the high-end gain divider frequency is greater than 1 kHz.

Images