Respiratory monitoring on TWS headsets

Abstract

The invention relates to respiratory monitoring on TWS headsets. A wearable electronic device detects respiration of a user based on bone conduction of acoustic waves. The wearable electronic device includes an inertial sensor unit. The inertial sensor unit generates sensor data based on bone conduction of the sound. The inertial sensor unit generates frequency domain data based on sensor data. The inertial sensor unit detects the respiration of the user by performing a classification process based on the frequency domain data.

Description

Technical Field

[0001] The present disclosure relates to human respiration detection based on bone conduction of sound detected by an inertial MEMS sensor. Background Art

[0002] In many medical applications, it is beneficial to reliably detect an individual's respiration. Typically, human respiration detection (along with other biometric signals) is implemented in an ad-hoc device, which is usually positioned near the body part generating the signal (such as a wearable bracelet, etc.). The user typically wears the device only when continuous measurement is required.

[0003] In one possible solution, a human respiration detection device uses a microphone to detect respiration. However, microphones have electrical noise, are vulnerable to external interference, and are generally very expensive in terms of processing and energy resources. It may be difficult to efficiently detect human respiration using a microphone.

[0004] All topics discussed in the background art section are not necessarily prior art and should not be considered prior art merely because of their discussion in the background art section. Along these lines, any recognition of problems in the prior art discussed in the background art section or associated with such topics should not be considered prior art unless explicitly stated as such. Instead, the discussion of any topic in the background art section should be considered part of the inventor's approach to solving a particular problem, which itself may be creative. Summary of the Invention

[0005] Embodiments of the present disclosure detect human respiration via bone conduction of sound using a low-power MEMS inertial sensor. A sensor unit including the MEMS sensor can be embedded in common wearable electronic devices (such as headphones, headsets, smart glasses, or other types of electronic devices in contact with the user's body). The sensor unit processes inertial sensor data from the inertial sensor and generates a spectral representation of the sensor data. The sensor unit extracts spectral features based on bone conduction of sound and identifies periods of human respiration based on the spectral features.

[0006] In one embodiment, a method includes: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user; generating frequency domain data based on the sensor data using the inertial sensor unit; and detecting the user's respiration by performing a classification process based on the frequency domain data using the inertial sensor unit.

[0007] In one embodiment, the method includes: calculating spectral energy, spectral centroid frequency, and spectral spread from frequency-domain data based on sensor data; and detecting a user's respiration by using an inertial sensor unit to perform a classification process based on the spectral energy, the spectral centroid frequency, and the spectral spread.

[0008] In one embodiment, a wearable electronic device includes a first sensor unit, the first sensor unit including: an inertial sensor configured to generate first sensor data based on bone conduction of sound; and a control circuit. The control circuit is configured to generate frequency-domain data based on the first sensor data and generate respiration detection data indicating a user's respiration based on spectral energy, spectral centroid frequency, and spectral spread.

[0009] In one embodiment, a method includes: generating sensor data based on bone conduction of sound by using an inertial sensor unit of an electronic device worn by a user; and performing an axis fusion process on the sensor data to fuse multiple axes into the sensor data. The method includes: generating a plurality of windows from the sensor data; calculating a first feature and a second feature for each window by using the inertial sensor unit; and detecting a user's respiration by using the inertial sensor unit based on the first feature and the second feature of the plurality of windows. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 is a block diagram of a system according to one embodiment, the system including a wearable electronic device that detects a user's respiration based on bone conduction of sound.

[0011] Figure 2 is according to one embodiment Figure 1 of a functional block diagram of a control circuit of an inertial sensor unit of a wearable electronic device.

[0012] Figures 3A to 3C includes a diagram associated with a frequency-domain representation of inertial sensor data according to one embodiment.

[0013] Figure 4 is a flowchart of a process for detecting human respiration according to one embodiment.

[0014] Figure 5 is an illustration of a user with a wearable electronic device according to one embodiment.

[0015] Figure 6 is an illustration of a user with a wearable electronic device according to one embodiment.

[0016] Figure 7 is a flowchart of a method for detecting human respiration according to one embodiment.

[0017] Figure 8 FIG. 1 is a flowchart of a method for detecting human respiration according to one embodiment. DETAILED DESCRIPTION

[0018] In the following description, certain specific details are set forth in order to provide a thorough understanding of the various disclosed embodiments. However, one of ordinary skill in the art will recognize that embodiments may be practiced without one or more of these specific details or with other methods, components, materials, and the like. In other instances, well-known systems, components, and circuitry associated with integrated circuits have not been shown or described in detail to avoid unnecessarily obscuring the description of the embodiments.

[0019] Unless the context requires otherwise, throughout the specification and the appended claims, the word "comprise" and variations thereof (such as "comprises" and "comprising") are to be construed in an open, inclusive sense, i.e., "including but not limited to." Further, unless the context clearly indicates otherwise, the terms "first," "second," and similar ordinal indicators are to be construed as interchangeable.

[0020] Throughout this specification, references to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" throughout this specification are not necessarily all referring to the same embodiment. Moreover, in one or more embodiments, the particular features, structures, or characteristics may be combined in any suitable manner.

[0021] As used in this specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a," "an," and "the" include plural referents. It should also be noted that, unless the context clearly indicates otherwise, the term "or" is generally employed in its broadest sense, i.e., meaning "and / or."

[0022] Figure 1 FIG. 2 is a block diagram of a system 100 for detecting human respiration according to one embodiment. System 100 includes a wearable electronic device 101. Wearable electronic device 101 includes a sensor unit 102. As will be elaborated in more detail below, the components of wearable electronic device 101 cooperate to detect the respiration of a user wearing wearable electronic device 101.

[0023] The wearable electronic device 101 is a device that can be coupled to or worn on a personal body. When used by an individual, the wearable electronic device 101 can be in continuous contact with the individual's skin. The wearable electronic device 101 can include wireless earphones, headphones, smart glasses, or other types of wearable electronic devices.

[0024] When a person speaks, the sound emitted by the user is generated as sound waves corresponding to the vibrations generated in the vocal cords. The sound waves can continue to propagate through the air. When the ears of other people sense the vibrations of the sound waves transmitted through the air, they can hear what the person is saying.

[0025] Sound waves are also conducted through media other than air. Sound waves are also conducted through liquids and solid materials. Human bones conduct sound waves in a unique way. Therefore, when a person speaks, the vibrations are conducted through human bones.

[0026] Human breathing also generates sound waves. One way to detect human breathing is to place a microphone at a position where it can sense the sound waves propagating through the air. However, human breathing is relatively quiet, and detecting breathing may require a highly sensitive microphone. In addition, background noise may interfere with the ability of the microphone to detect human breathing.

[0027] The wearable electronic device 101 detects the user's breathing based on the bone conduction of the sound waves generated by the user's breathing. Since the electronic device 101 is in contact with the user's skin, the electronic device 101 can reliably sense the bone conduction of the sound generated by the user's breathing. The sound waves generated by breathing typically can have a frequency range between 150 Hz and 450 Hz, which visually appears as (refer to Figures 3A to 3C ) a noise floor cloud. Frequencies below 150 Hz can correspond to external interferences such as human movement and may be ignored when detecting breathing via bone conduction.

[0028] In one embodiment, the wearable electronic device includes a sensor unit 102. The sensor unit 102 includes an inertial sensor 104. The inertial sensor 104 senses the vibrations generated by the bone conduction of the sound of the user's breathing. The inertial sensor 104 can include a microelectromechanical system (MEMS) sensor.

[0029] In one embodiment, the inertial sensor 104 includes an accelerometer. The accelerometer can be a triaxial accelerometer that senses the acceleration of each of three mutually orthogonal sensing axes. The sensing axes can correspond to the X-axis, Y-axis, and Z-axis. The inertial sensor 104 can also include a gyroscope that senses the rotation about the three mutually orthogonal sensing axes.

[0030] The sensor unit 102 includes a control circuit 106 coupled to an inertial sensor 104. The control circuit 106 may include processing resources, memory resources, and communication resources. As will be elaborated in more detail below, the control circuit 106 is capable of detecting a user's respiration based on bone conduction of sound sensed by the inertial sensor 104.

[0031] In one embodiment, the inertial sensor 104 is implemented in a first integrated circuit die. The control circuit 106 is implemented in a second integrated circuit die directly coupled to the first integrated circuit die. In one embodiment, the inertial sensor 104 and the control circuit 106 are implemented in a single integrated circuit die as a system-on-chip. The control circuit 106 may correspond to an application specific integrated circuit (ASIC).

[0032] The inertial sensor 104 initially generates an analog sensor signal based on bone conduction of sound waves. Vibration may be sensed capacitively, piezoelectrically, or in another suitable manner. The inertial sensor 104 generates an analog sensor signal based on the sensing. The analog sensor signal is converted into digital sensor data by an analog-to-digital converter (ADC) in the inertial sensor 104 or in the control circuit 106. The ADC can be configured to generate a selected number of samples of sensor data per second.

[0033] In one embodiment, the control circuit 106 receives the sensor data (or generates the sensor data from the sensor signal) and processes the sensor data to detect a user's respiration. The control circuit 106 may include signal processing circuitry to perform a plurality of processing steps to format or condition the sensor data to detect a user's respiration.

[0034] In one embodiment, the control circuit 106 generates frequency domain data from the sensor data. The sensor data is typically in the time domain (i.e., sequential samples). The control circuit system 106 generates the frequency domain data by transforming the sensor data from the time domain to the frequency domain. Thus, as used herein, the frequency domain data corresponds to the sensor data that has been transformed to the frequency domain. The frequency domain data may include spectral data indicative of the frequencies present in a group of samples of the sensor data.

[0035] In one embodiment, the control circuit system 106 performs a discrete Fourier transform (DFT) on the sensor data to convert the sensor data into frequency domain data. Alternatively, the control circuit system 106 is capable of performing other types of transforms or conversions to generate the frequency domain data. For example, the control circuit system can utilize other types of Fourier transforms, wavelet transforms, or other types of transforms to generate the frequency domain data. In one embodiment, the control circuit system 106 performs a sliding discrete Fourier transform (SDFT). The frequency domain data may be referred to as "spectral data".

[0036] In one embodiment, control circuitry 106 generates a plurality of spectral features from frequency-domain data. Control circuitry 106 is then able to analyze the spectral features in order to detect a user's respiration. Analyzing the spectral features may include performing a classification process based on the spectral features.

[0037] In one embodiment, control circuitry 106 generates a spectral energy value X for each set of samples based on the spectral data E . The spectral energy value corresponds to a spectral feature. In one embodiment, control circuitry 106 generates the spectral energy value as follows:

[0038]

[0039] where Xkr is the real component of the Fourier component with index k, Xki is the imaginary component of the Fourier component with index k, bin_start corresponds to the start frequency interval, and bin_stop corresponds to the final frequency interval. A spectral energy value Xe (not present in the above formula but present in the following formula) is generated for each of the n intervals. In an example of bone conduction of sound, the first frequency interval may be about 150 Hz and the final frequency interval may be about 450 Hz, although other frequency ranges can be utilized without departing from the scope of the present disclosure.

[0040] In one embodiment, control circuitry 106 generates a spectral centroid frequency (SCF) value for each set of samples based on the spectral data. The spectral centroid frequency value corresponds to a spectral feature. The spectral centroid frequency value can correspond to the center of mass of the spectrum of a set of samples. In one embodiment, control circuitry 106 generates the spectral centroid frequency value as follows:

[0041]

[0042] In one embodiment, control circuitry 106 generates a spectral spread value (SSP) for each set of samples based on the spectral data. The spectral spread value corresponds to a spectral feature. The spectral spread can correspond to the frequency range present within a particular set of samples. In one embodiment, the spectral spread can be generated as follows:

[0043]

[0044] where d f k corresponds to the frequency corresponding to interval k.

[0045] In one embodiment, the control circuit 106 detects respiration by performing a classification process or a detection process based on features generated from spectral data. For example, the respiration detection process can detect the respiration of a sample group by comparing one or more of the features generated from the spectral data with one or more thresholds. The classification process can correspond to a respiration detection algorithm.

[0046] In one embodiment, the control circuit 106 includes an analysis model trained using a machine learning process to detect human respiration. A supervised machine learning process can be used to train the analysis model to detect human respiration based on one or more of raw sensor data, spectral data, spectral features, or time-domain features. Thus, the analysis model can be trained based on a training set collected from a user of the wearable electronic device 100, based on a training set collected from other users of the wearable electronic device 100, or based on other types of training sets. The analysis model can output a classification indicating whether the sample group indicates human respiration.

[0047] In one embodiment, human respiration is detected based on multiple sets of samples of sensor data. Each group can correspond to a frame or window of samples of the sensor data. The respiration detection process can output a pre-classification for each window or frame of the sensor data. An overall classification of respiration can be generated for multiple windows or frames of the sensor data. For example, respiration detection can correspond to detecting whether the wearer is inhaling. Inhalation can occur over a large number of frames or windows of the sensor data. The respiration detection process can generate an inhalation classification for multiple consecutive windows or frames. This can correspond to detecting the time range during which the user inhales. The pre-classification from each window or frame can be used to determine whether a set of consecutive frames represents inhalation. A similar type of classification can be performed for exhalation instead of inhalation. Without departing from the scope of the present disclosure, various other types of classification processes can be utilized.

[0048] In one embodiment, the window or frame is further processed by a window function. The window function can correspond to a Hann window function. The Hann window function starts and ends at zero values, with a center value of one. Without departing from the scope of the present disclosure, other types of window functions can be applied to further process the window or frame.

[0049] In one embodiment, the wearable electronic device 101 includes a wireless transceiver 108. The wireless transceiver 108 can receive respiration detection data from the sensor unit 102 and can transmit the respiration detection data to the remote electronic device 103. The wireless transceiver can operate according to one or more wireless protocols, including Bluetooth, Wi-Fi, or other suitable wireless communication protocols.

[0050] In one embodiment, the remote electronic device 103 receives respiratory detection data from the wearable electronic device 101. The remote electronic device 103 is capable of displaying the read data to the user. The remote electronic device 103 may include one or more applications or circuits that process the respiratory data in order to perform one or more health monitoring-related functions. The remote electronic device 103 may include a smart phone, a smart watch, a laptop computer, a tablet computer, or other types of electronic devices.

[0051] The remote electronic device 103 may include a wireless transceiver 109 that is capable of receiving respiratory detection data from the wearable electronic device 101. The wireless transceiver 109 is also capable of transmitting data to the wearable electronic device 101. The wireless transceiver is capable of operating according to one or more wireless protocols, including Bluetooth, Wi-Fi, or other suitable wireless communication protocols.

[0052] Figure 2 is a block diagram of a control circuit 106 according to one embodiment. The control circuit 106 includes a preprocessing circuit 110, a spectrum generator 112, a feature generator 114, and a classifier 116. Figure 2 The control circuit 106 of Figure 1 is an example of the control circuit 106 of

[0053] The preprocessing circuit 110 receives sensor data from the inertial sensor 104. The preprocessing circuit 110 includes an axis fusion module 118. The axis fusion module 118 performs an axis fusion process on the sensor data. In embodiments where the control circuit 106 detects human respiration based on spectral features, the data associated with any particular axis of the inertial sensor 104 may be less beneficial than the total magnitude of the signals of all the combined axes. Thus, in one embodiment, the axis fusion module 118 generates the absolute value of the magnitude of each sample of the sensor data by taking the square root of the sum of the squares of the sensor data on each axis. This preserves the frequency information while simplifying the data set. The axis fusion module may perform classical norm or magnitude calculations to mix the vibration contributions from all axes. In another embodiment, the axis fusion module 118 can be configured to select only the values generated by a particular axis and ignore the values generated by the other axes.

[0054] In one embodiment, the preprocessing circuit 110 includes a low-pass filter module 120. The low-pass filter module 120 performs a low-pass filtering process on the fused sensor data (i.e., the sensor data modified as by the axis fusion module 118). The low-pass filtering process can correspond to a CIC filter that also decimates the sensor data. Without departing from the scope of the present disclosure, the preprocessing circuit 110 may include other circuitry and perform other functions. In one example, the low-pass filter may have a bandwidth of 1 kHz.

[0055] In one embodiment, the spectrum generator 112 includes a discrete Fourier transform module 122. The discrete Fourier transform (DFT) module 122 receives the filtered sensor data from the low-pass filter 120 of the preprocessing circuit 110. The DFT module 122 generates spectrum data from the sensor data by performing a discrete Fourier transform on the sensor data. Without departing from the scope of the present disclosure, other types of processes can be utilized to generate spectrum data from the sensor data. The spectrum data can be output based on the number of samples in each window.

[0056] The feature generator 114 receives the spectrum data from the DFT module 122. In one embodiment, the feature generator 114 includes a spectral centroid frequency module 124. The spectral centroid frequency module 124 generates a spectral centroid frequency, as previously described with respect to Figure 1 In one embodiment, the feature generator 114 includes a spectral spread module 126. The spectral spread module 126 generates a spectral spread value, as previously described with respect to Figure 1 In one embodiment, the feature generator 114 includes a spectral energy module 128. The spectral energy module 128 generates a spectral energy value, as previously described with respect to Figure 1 Without departing from the scope of the present disclosure, the feature generator 114 may include other circuitry or generate other features in addition to those described herein.

[0057] In one embodiment, the classifier 116 includes one or more classification process modules 130. The one or more classification process modules are capable of performing one or more classification processes or respiration detection processes based on the feature data generated by the feature generator 114.

[0058] In one embodiment, the classification process includes generating a value based on the spectral energy X E , the spectral centroid frequency SCF, and the spectral spread SSP, and comparing the value with a threshold. In one embodiment, respiration is detected by generating an output value (out) and comparing the output value with a threshold (th) as follows:

[0059]

[0060] In this example, if the output value is greater than the threshold, then respiration is detected. This metric is proportional to the presence of the noise floor cloud. The threshold can be used to distinguish true respiration activity. Without departing from the scope of the present disclosure, other types of mathematical formulas can be utilized.

[0061] In one embodiment, the classification process includes using multiple thresholds. Specifically, the classification process can compare the spectral energy with a first threshold (th1), compare the spectral centroid frequency with a second threshold (th2), and compare the spectral spread with a third threshold th3 in the following manner:

[0062] out = (X E (n) > th1) && (SCF(n) <= th2) && (SSP(n) <= th3)

[0063] In this case, if the spectral energy is greater than the first threshold, the spectral centroid frequency is less than or equal to the second threshold, and the spectral spread is less than or equal to the third threshold, then breathing is detected. This process can correspond to a binary tree process. The results are combined to satisfy the boolean values for breathing activity detection.

[0064] In one embodiment, the classification process includes an analysis model (such as a neural network) trained using a machine learning process to detect and classify breathing based on spectral features. In one embodiment, the analysis model receives the spectral energy, spectral centroid frequency, and spectral spread, and detects breathing based on these features. Without departing from the scope of the present disclosure, other types of features and other types of analysis models can be used to detect breathing. Neural network inference can be trained to mix the three spectral features to detect breathing activity. Fully connected convolutional or recurrent neural network layers can be used to consider the timing envelope. The neural network can include a dense layer that can aggregate the outputs of previous layers and generate a single breath detection or prediction.

[0065] Figure 3A FIG. 300 is a diagram illustrating the relationship between spectral data and time according to one embodiment. The x-axis corresponds to time, and the y-axis corresponds to frequency. In one example, the frequency is between 100 Hz and 800 Hz, although other frequency ranges can be used without departing from the scope of the present disclosure.

[0066] Figure 3B is according to one embodiment of Figure 3A FIG. 300, but with markings highlighting the spectral features. Marking 306 corresponds to the spectral energy X E corresponds. Marking 304 corresponds to the spectral spread SSP, and marking 302 corresponds to the spectral centroid frequency SCF.

[0067] According to one embodiment, Figure 3C corresponds to FIG. 300 and includes a breathing indication diagram 310 positioned below FIG. 300. Diagram 310 includes pulses indicating the periods of time during which breathing is detected. The pulses can correspond to the inhalation phase during which the user of the wearable electronic device 100 is inhaling. Four inhalation periods are as Figure 3Cas shown. The user inhales between time t 1 and t 2 and between time t 3 and t 4 and between time t 5 and t 6 and between time t 7 and t 8 . Thus, there are four inhalation periods in Figure 3C .

[0068] Figure 4 is a block diagram of a control circuit 106 according to an embodiment. The control circuit 106 is an example of the control circuit 106 of Figure 1 . The control circuit 106 receives raw sensor signals, which include accelerometer signals aX, aY, and aZ for three axes (which can indicate bone conduction of sound) and gyroscope signals gX, gY, and gZ for three axes. In some cases, only acceleration data can be used.

[0069] The feature extraction module 402 performs feature extraction of the sensor signals. The feature extraction can include digitally filtering the sensor data using a digital filter 408. The feature extraction includes generating time-domain features from the sensor data using a time-domain feature module 410. The time-domain features can include peak-to-peak value, root mean square (RMS), mean value, variance, energy, maximum value, minimum value, zero crossing rate (ZCR), or other types of time-domain features. The feature extraction includes using a sliding discrete Fourier transform (SDFT) module 412 to generate a frequency-domain signal and frequency-domain features. This can include performing a Fourier transform and generating spectral energy, spectral centroid frequency, and spectral spread data as previously described.

[0070] The control circuit 106 includes a neural network 404. The neural network 404 is capable of receiving one or more of raw sensor data, time-domain features, frequency-domain features, and full-spectrum data from the feature extraction module 402. The neural network then analyzes various data and features in order to generate a pre-classification. For each frame or window, the pre-classification can indicate whether breathing is detected. Without departing from the scope of the present disclosure, other types of analysis models other than neural networks can be used. As Figure 4 shown, in one embodiment, the feature extraction is bypassed, and the sensor data is directly provided to the neural network.

[0071] The control circuit 106 includes a meta-classifier 406 that receives pre-classification data from the neural network 404. The meta-classifier can include multiple conditions, states, and commands. The meta-classifier is capable of generating a classification (i.e., detected breathing) based on the pre-classification provided by the neural network and based on the data provided by the feature extraction. As Figure 4As shown, it is possible to bypass the neural network 404 and directly provide the feature data from feature extraction to the meta-classifier. In addition, it is also possible to bypass the meta-classifier 406 such that the neural network provides classification.

[0072] In Figure 4 the example of, the meta-classifier 406 includes three states, three commands, and six conditions, based on which classification and prediction can be performed. However, without departing from the scope of the present disclosure, other numbers of states, commands, and conditions can be utilized. Without departing from the scope of the present disclosure, various other types of processes and configurations can be utilized.

[0073] Figure 5 is an illustration of an individual wearing the earphone 501 and holding the smartphone 503 according to one embodiment. The earphone 501 is Figure 1 an example of the wearable electronic device 101 of Figure 1 The earphone 501 is in contact with the user's skin. The smartphone 503 is Figure 5 an example of the remote electronic device 103 of

[0074] In one embodiment, the sensor unit 102 including the inertial sensor 104 and the control circuit 106 is included in any one of the earphones 501. As previously described, the earphone 501 detects the user's breathing based on bone conduction of sound. In one embodiment, the second earphone 501 can also generate sensor data and provide the sensor data to the first earphone 501. The first earphone 501 can then detect the breathing based on the sensor data from both the first earphone 501 and the second earphone 501.

[0075] The earphone 501 is capable of providing the breathing detection data to the smartphone 503. The smartphone 503 can utilize the breathing detection data in one or more applications or circuits of the smartphone 503. The smartphone 503 can display the breathing detection data to the user. The smartphone 503 can generate health or wellness data based on the breathing detection data. The smartphone 503 can generate one or more reports, graphs, images, or other data that can be displayed or provided to the user or can be provided to other systems.

[0076] Figure 6 is an illustration of an individual wearing the smart glasses 601 and holding the smartphone 503 according to one embodiment. The smart glasses 601 are Figure 1 an example of the wearable electronic device 101 of Figure 1 The smart glasses 601 are in contact with the user's skin. The smartphone 601 is

[0077] In one embodiment, a sensor unit 102 including an inertial sensor 104 and a control circuit 106 is included in smart glasses 601. As previously described, the smart glasses 601 detect a user's respiration based on bone conduction of sound.

[0078] In one embodiment, the smart glasses 601 are capable of displaying respiration detection data to the user. The smart glasses 601 may utilize the respiration detection data in one or more applications or circuits.

[0079] The smart glasses 601 are capable of providing the respiration detection data to a smart phone 503. The smart phone 503 may utilize the respiration detection data in one or more applications or circuits of the smart phone 503. The smart phone 503 may display the respiration detection data to the user. The smart phone 503 may generate health or wellness data based on the respiration detection data. The smart phone 503 may generate one or more reports, graphs, images, or other data that can be displayed or provided to the user or to other systems.

[0080] Figure 7 is a flowchart of a method for detecting respiration using a wearable electronic device. Method 700 can utilize the components, processes, and systems described with respect to Figures 1 to 6 At 702, method 700 includes: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user. At 704, the method includes: generating frequency domain data based on the sensor data using the inertial sensor unit. At 706, the method includes: detecting the user's respiration by performing a classification process based on the frequency domain data using the inertial sensor unit.

[0081] Figure 8 is a flowchart of a method for detecting respiration using a wearable electronic device. Method 800 can utilize the components, processes, and systems described with respect to Figures 1 to 6 At 802, method 800 includes: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user. At 804, method 800 includes: performing an axis fusion process on the sensor data to fuse multiple axes into the sensor data. At 806, method 800 includes: generating multiple windows from the sensor data. At 808, method 800 includes: calculating a first feature and a second feature for each window using the inertial sensor unit. At 810, method 800 includes: detecting the user's respiration based on the first and second features of the multiple windows using the inertial sensor unit.

[0082] In one embodiment, a method includes: using an inertial sensor unit of an electronic device worn by a user to generate sensor data based on bone conduction of sound; using the inertial sensor unit to generate frequency domain data based on the sensor data; and using the inertial sensor unit to detect the user's respiration by performing a classification process based on the frequency domain data.

[0083] In one embodiment, the method includes: calculating spectral energy, spectral centroid frequency, and spectral spread from the frequency domain data based on the sensor data; and using the inertial sensor unit to detect the user's respiration by performing a classification process based on the spectral energy, the spectral centroid frequency, and the spectral spread.

[0084] In one embodiment, the method includes: performing axis fusion on the sensor data using norm calculation before generating the frequency domain data. In one embodiment, the method includes: performing low-pass filtering and decimation after performing axis fusion and before generating the frequency domain data.

[0085] In one embodiment, the method includes: generating a plurality of windows from the sensor data; and generating the frequency domain data by performing a sliding discrete Fourier transform on each window.

[0086] In one embodiment, calculating the spectral energy, the spectral centroid frequency, and the spectral spread includes calculating the spectral energy, spectral centroid frequency, and spectral spread of each window.

[0087] In one embodiment, the classification process includes: classifying each window from a set of windows; and detecting respiration of the set of windows based on the classification of each window of the set.

[0088] In one embodiment, the classification process includes: generating a value by dividing the spectral energy by the product of the spectral centroid frequency and the spectral spread; and comparing the value with a threshold.

[0089] In one embodiment, the classification algorithm includes: comparing the spectral energy with a first threshold; comparing the spectral centroid frequency with a second threshold; and comparing the spectral spread with a third threshold.

[0090] In one embodiment, the method includes: outputting respiration detection data from the inertial sensor unit based on detecting respiration.

[0091] In one embodiment, the classification process includes: passing the spectral energy, the spectral centroid frequency, and the spectral spread to an analysis model trained using a machine learning process; and detecting respiration based on the classification of the analysis model.

[0092] In one embodiment, the method includes: outputting the respiration detection data from the wearable electronic device to a remote electronic device.

[0093] In one embodiment, a wearable electronic device includes a first sensor unit, the first sensor unit including: an inertial sensor configured to generate first sensor data based on bone conduction of sound; and a control circuit. The control circuit is configured to generate frequency domain data based on the first sensor data and generate respiration detection data indicating the user's respiration based on spectral energy, spectral centroid frequency, and spectral spread.

[0094] In one embodiment, the control circuit is configured to generate spectral energy, spectral centroid frequency, and spectral spread from the frequency domain data based on the first sensor data, and generate respiration detection data based on the spectral energy, the spectral centroid frequency, and the spectral spread.

[0095] In one embodiment, the control circuit includes an analysis model trained using a machine learning process to detect respiration based on spectral energy, spectral centroid frequency, and spectral spread.

[0096] In one embodiment, the electronic device includes a first earphone, the first earphone including the sensor unit.

[0097] In one embodiment, the electronic device includes a second earphone, the second earphone including a second sensor unit configured to provide second sensor data to the first sensor unit. The control circuit is configured to generate respiration detection data based on the first sensor data and the second sensor data.

[0098] In one embodiment, a method includes: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user; and performing an axis fusion process on the sensor data to fuse multiple axes into the sensor data. The method includes: generating a plurality of windows from the sensor data; calculating a first feature and a second feature for each window using the inertial sensor unit; and detecting the user's respiration based on the first feature and the second feature of the plurality of windows using the inertial sensor unit.

[0099] In one embodiment, the method includes: calculating a third feature for each window using the inertial sensor unit; and detecting the user's respiration based on the first feature, the second feature, and the third feature of the plurality of windows using the inertial sensor unit.

[0100] In one embodiment, the first feature is spectral energy, the second feature is spectral centroid frequency, and the third feature is spectral spread.

[0101] The various embodiments described above can be combined to provide additional embodiments. In light of the foregoing detailed description, these and other changes can be made to the embodiments. Generally, in the following claims, the terms used should not be construed as limiting the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments and the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the present disclosure.

Claims

1. A method comprising: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user; generating frequency domain data based on the sensor data using the inertial sensor unit; as well as The user's breathing is detected by performing a classification process based on the frequency domain data using the inertial sensor unit.

2. The method according to claim 1, comprising: calculating spectral energy, spectral centroid frequency, and spectral spread from the frequency domain data based on the sensor data; as well as The user's breathing is detected using the inertial sensor unit by performing a classification process based on the spectrum energy, the spectrum centroid frequency, and the spectrum spread. The method of claim 2 , comprising performing axis fusion on the sensor data prior to generating the frequency domain data. 4 . The method of claim 3 , comprising performing low pass filtering and decimation after performing axis fusion and before generating the frequency domain data.

5. The method according to claim 2, comprising: generating a plurality of windows from the sensor data; as well as The frequency domain data is generated by performing a sliding discrete Fourier transform on each window. 6 . The method according to claim 5 , wherein calculating the spectral energy, the spectral centroid frequency, and the spectral extension comprises calculating the spectral energy, the spectral centroid frequency, and the spectral extension for each window.

7. The method according to claim 6, wherein the classification process comprises: performing a classification for each window from a set of said windows; as well as Respiration of the set of windows is detected based on the classification of each window of the set.

8. The method according to claim 2, wherein the classification process comprises: generating a value by dividing the spectrum energy by the product of the spectrum centroid frequency and the spectrum spread; as well as The value is compared to a threshold value.

9. The method of claim 2, wherein the classification algorithm comprises: comparing the spectrum energy with a first threshold; comparing the spectrum centroid frequency with a second threshold; as well as The spectrum spread is compared with a third threshold.

10. The method of claim 2, comprising outputting respiration detection data from the inertial sensor unit based on detecting the respiration.

11. The method of claim 2, wherein the classification algorithm comprises: passing the spectrum energy, the spectrum centroid frequency, and the spectrum spread to an analysis model trained using a machine learning process; as well as Respiration is detected based on the classification of the analysis model.

12. The method of claim 11, comprising outputting the breathing detection data from the wearable electronic device to a remote electronic device.

13. A wearable electronic device comprising: The first sensor unit comprises: an inertial sensor configured to generate first sensor data based on bone conduction of sound; The control circuit is configured as follows: generating frequency domain data based on the first sensor data; Breathing detection data is generated based on the spectrum energy, the spectrum centroid frequency, and the spectrum extension, the breathing detection data indicating the breathing of the user.

14. A wearable electronic device according to claim 13, wherein the control circuit is configured to generate spectral energy, spectral centroid frequency and spectral extension from the frequency domain data based on the first sensor data, and generate the breathing detection data based on the spectral energy, the spectral centroid frequency and the spectral extension.

15. The electronic device of claim 13, wherein the control circuit comprises an analysis model trained using a machine learning process to detect respiration based on the spectrum energy, the spectrum centroid frequency, and the spectrum spread.

16. The electronic device according to claim 15, comprising a first earphone, the first earphone comprising the sensor unit.

17. The electronic device according to claim 16 comprises a second earphone, the second earphone comprising a second sensor unit, the second sensor unit being configured to provide second sensor data to the first sensor unit, wherein the control circuit is configured to generate the breathing detection data based on the first sensor data and the second sensor data.

18. A method comprising: generating sensor data based on bone conduction of sound using an inertial sensor unit of an electronic device worn by a user; performing an axis fusion process on the sensor data to fuse multiple axes into the sensor data; generating a plurality of windows from the sensor data; calculating a first feature and a second feature for each window using the inertial sensor unit; as well as The user's breathing is detected based on the first feature and the second feature of a plurality of windows using the inertial sensor unit.

19. The method according to claim 18, comprising: calculating a third feature for each window using the inertial sensor unit; as well as The user's breathing is detected based on the first feature, the second feature, and the third feature of a plurality of windows using the inertial sensor unit.

20. The method of claim 19, wherein the first characteristic is spectral energy, the second characteristic is spectral centroid frequency, and the third characteristic is spectral spread.

Images