Heart rate estimator and headset system

Abstract

A heart rate estimation device includes a feature point extraction unit that extracts a waveform feature point from a waveform representing a pressure fluctuation in a closed space; a pulse waveform generation unit that generates a pulse waveform constituting a pulse beat at the same timing as the waveform feature point; a frequency analysis unit that performs frequency analysis of the pulse waveform; and a heart rate estimation unit that estimates a heart rate by analyzing a frequency feature point obtained from a result of the frequency analysis.

Description

[Technical field]

[0001] The technique of the present disclosure relates to a heart rate estimation device and a headset system. STATE OF THE ART

[0002] There are already attempts to collect biological information from the human head using a headset. Among the biological information, it is particularly desirable to collect information with high accuracy regarding head movements, core body temperature, and heart rate.

[0003] In heart rate estimation processing, it is important to eliminate erroneous detections due to noise generated by body movements. Therefore, an optical heart rate estimation method is known in which multiple light sources and light sensors are used to observe the same point from multiple directions, and noise due to body movements is removed (see, for example, U.S. Patent No. 8,998,815). SUMMARY OF THE INVENTION Problem to be solved by the invention

[0004] However, in a case where heart rate is estimated from pressure fluctuations in a closed space, no noise reduction effect can be expected even if multiple pressure sensors or microphones are used, since pressure fluctuations in the same space are essentially uniform.

[0005] Therefore, it is necessary to estimate heart rate by analyzing pressure fluctuations detected by a pressure sensor or microphone in a noisy environment. However, during pressure fluctuations in a confined space, the fluctuations due to body movement are often significantly larger than the expansion and contraction of blood vessels due to blood flow.

[0006] Thus, even when performing a general frequency analysis (spectral analysis) of pressure fluctuations in a closed space, the characteristics due to body movements, artifacts, and the like become dominant, and there arises a problem that cyclic feature points caused by the expansion and contraction of a blood vessel are obscured.

[0007] An object of the technique of the present disclosure is to provide a heart rate estimation device and a headset system capable of accurately estimating a heart rate from pressure fluctuations in an enclosed space. Means to solve the problem

[0008] One aspect of the present disclosure is a heart rate estimation device configured to include: a feature point extraction unit that extracts a waveform feature point from a waveform representing a pressure fluctuation in a closed space; a pulse waveform generation unit that generates a pulse waveform constituting a pulse beat (pulse curve, pulse amplitude) at the same timing as the waveform feature point; a frequency analysis unit that performs frequency analysis of the pulse waveform; and a heart rate estimation unit that estimates a heart rate by analyzing a cyclic feature point obtained from a result of the frequency analysis.

[0009] One aspect of the present disclosure is a headset system configured to include the heart rate estimation device described above, and a headset comprising: a hollow housing worn on an ear of a user; a driver for acoustic signal output provided inside the housing; an optical detection unit provided inside the housing on the back of the driver; and an output unit that outputs to the heart rate estimation device a signal output from the optical detection unit as a waveform representing a pressure fluctuation in a closed space. Effect of the invention

[0010] As explained above, according to the heart rate estimation device and the headset system of the technique of the present disclosure, the heart rate can be accurately estimated from pressure fluctuations in a closed space. BRIEF DESCRIPTION OF THE DRAWINGS Fig. Figure 1A is a diagram showing an example of a waveform of a signal output from a microphone. Fig. Figure 1B is a diagram showing an example of waveform feature points extracted from a waveform of a signal output from a microphone. Fig. Figure 1C is a graph showing an example of identical timing to waveform feature points. Fig. Figure 1D is a diagram showing a method for generating a pulse waveform. Fig. Figure 1E is a diagram showing an example of a pulse waveform. Fig. Figure 1F is a graph showing an example of a frequency analysis result of a pulse waveform. Fig. Figure 1G is a graph showing an example of cyclic feature points obtained from a frequency analysis result. Fig. Figure 2A is a graph showing an example of waveform feature points extracted from a waveform of pressure fluctuations in an enclosed space during a resting phase. Fig. Figure 2B is a graph showing an example of cyclic feature points obtained from a frequency analysis result. Fig. Figure 3A is a graph showing an example of waveform feature points extracted from a waveform of pressure fluctuations in an enclosed space during a body movement phase. Fig. Figure 3B is a graph showing an example of cyclic feature points obtained from a frequency analysis result. Fig. Figure 4A is a graph showing an example of a result of frequency analysis using a discrete Fourier transform. Fig. Figure 4B is a graph showing an example of a result of frequency analysis using a fast Fourier transform. Fig. 5 is a cross-sectional view showing an overall configuration of a headset according to a first exemplary embodiment of the technique of the present disclosure. Fig. 6 is a block diagram showing a configuration of an information processing unit according to a first exemplary embodiment of the technique of the present disclosure. Fig. 7 is a diagram showing a method of extracting a zero crossing as a waveform feature point. Fig. Figure 8 is a graph showing an example of preliminary peaks and preliminary troughs captured in a signal waveform. Fig. Figure 9 is a graph showing an example of preliminary peaks and preliminary troughs after removing the preliminary peaks and preliminary troughs that are believed to have been erroneously recorded. Fig. 10 is a flowchart showing the content of heart rate estimation processing by an information processing unit according to a first exemplary embodiment of the technique of the present disclosure. Fig. 11 is a cross-sectional view showing an overall configuration of a headset according to a second exemplary embodiment of the technique of the present disclosure. Fig. 12 is a cross-sectional view showing an overall configuration of a headset according to an example. DESCRIPTION OF THE INVENTION

[0011] Hereinafter, exemplary embodiments of the technique of the present disclosure will be explained in more detail with reference to the drawings. [First Exemplary Embodiment] <Overview of the First Exemplary Embodiment of the Technique of the Present Disclosure>

[0012] When processing to estimate heart rate, it is important to eliminate erroneous detection of the heartbeat due to noise generated by body movements, artifacts, or the like.

[0013] In the present exemplary embodiment, frequency analysis is not directly performed on a waveform of pressure fluctuations in a closed space, and waveform feature points that are characteristic points of the waveform are extracted, and a pulse waveform is generated in which the pulse beats occur at the same time points at which the waveform feature points are generated. Frequency analysis is performed on the pulse waveform, and the heart rate is estimated by analyzing cyclic feature points that are characteristic points of the frequency analysis.

[0014] More precisely, a waveform of the pressure fluctuation in the enclosed space, as in Fig. 1A is acquired, and waveform feature points are plotted as shown in Fig. 1B (see the x-marks in Fig. 1B). As in Fig. 1C and Fig. 1D, a pulse waveform is generated in which the pulse beats are formed at the same time as the waveform feature points. For a pulse waveform as shown in Fig. 1E, a frequency analysis is performed and a Fig. 1F is obtained. The heart rate is determined from the values marked by the arrows in Fig. 1G specified cyclic feature points.

[0015] A waveform feature point caused by a large pressure change in the confined space due to body movement or the like and a waveform feature point caused by a small pressure change in the confined space due to the expansion and contraction of a blood vessel are converted to the same fluctuation level, and frequency analysis is performed. Thus, even in cases where they would be hidden in the frequency characteristics of body movement using conventional methods, it is easy to distinguish cyclic feature points that have occurred due to the expansion and contraction of a blood vessel, thereby enabling accurate heart rate estimation even in conditions where body movement occurs.

[0016] Here shows Fig. 2A shows an example of waveform feature points extracted from a waveform of pressure fluctuations in a closed space during a resting phase. In this example, the pulse frequency is estimated to be 64 ppm (1.07 Hz) based on the waveform spacing during filling and emptying of the blood vessel. At this time, in the present exemplary embodiment, a result of the frequency analysis as shown in Fig. 2B, and a tip (see arrow mark in Fig. 2B) appears clearly in a band calculated based on the waveform spacing and is extracted as a cyclic feature point. It is understood that this cyclic feature point corresponds to 64 ppm (1.07 Hz).

[0017] Furthermore, Fig. 3A shows an example of waveform feature points extracted from a waveform of pressure fluctuations in a closed space during a phase of body movement. In this example, the pulse frequency is estimated to be 68 ppm (1.13 Hz) based on the waveform spacing during filling and emptying of the blood vessel. At this time, in the present exemplary embodiment, a result of the frequency analysis as shown in Fig. 3B, and a tip (see arrow mark in Fig. 3B) appears clearly in a band calculated based on the waveform spacing and is extracted as a cyclic feature point. It is understood that this cyclic feature point corresponds to 68 ppm (1.13 Hz).

[0018] Furthermore, the amplitude of the pulse waveform can be changed using any value or fixed. For frequency analysis, the fast Fourier transform can be used, but the discrete Fourier transform (DFT) is preferred.

[0019] Since it is sufficient to extract a high-intensity frequency for heart rate estimation, the discrete Fourier transform for obtaining the frequency characteristic Ff of the pulse wave X(n) can be simplified as follows, omitting the operations for normalizing the output signal, such as calculating the square root of the root mean square. F(f)=(∑n=0N−1X(n)⋅cos(2πnffs))2+(∑n=0N−1X(n)⋅sin(2πnffs))2

[0020] In equation (1), N indicates the number of samples of the pulse wave X(n), fs indicates the sampling frequency of X(n), and f indicates the frequency to be analyzed.

[0021] Note that since X(n) is a pulse wave, no arithmetic operations need to be performed because X(n) is zero except for the pulse point n where a pulse beat is present. Furthermore, since in cases where the amplitude is a fixed value, the sine / cosine value is obtained only by table reference, the intensity can be obtained by performing an N-ALU / MUX operation (complex processing of table reference, multiplication, and addition) for each frequency f. Furthermore, it is possible to perform a calculation limited to only the frequency f required to estimate the heart rate. This allows for a significant reduction in computational complexity compared to performing a fast Fourier transform (computational complexity: Nlog2N).

[0022] An example of a case where equation (1) is applied to the pulse waveform of Fig. 1E is applied is shown below. Sampling frequency f s = 100Hz Scan length N = 2048 Number of pulse points: 45 points Analysis frequency: 0.5 ≦ f ≦ 3.0 (equivalent to 30 ppm to 180 ppm)

[0023] The effort for calculating the frequency characteristic Xf in the pulse waveform of Fig. 1E, described above, was as follows.

[0024] The table memory capacity is 2048 × 16 bits. The number of ALU / MUX operations is 4500 (= 45 (number of pulse points) x 50 (number of analysis bands) x 2 (real number part + imaginary number part). Where 45 is the number of pulse points, 50 is the number of analysis bands, and 2 is the pair of a real number part and an imaginary number part.

[0025] The effort required when using the fast Fourier transform is as follows.

[0026] The table memory capacity is 2048 × 16 bits. The number of ALU / MUX operations is 40960 (= 2048 × log2(2048) × 2).

[0027] In this way, the advantageous effects of the present exemplary embodiment can be confirmed in comparison with the effort required when performing a fast Fourier transform.

[0028] The Fig. 4A and Fig. 4B show the obtained frequency characteristics. Fig. 4A is a graph showing an example of a result of frequency analysis using a discrete Fourier transform, illustrating an example in which the analysis frequency f is 0.5 Hz to 3.0 Hz. Fig. Figure 4B is a graph showing an example of a frequency analysis result using a fast Fourier transform.

[0029] Although the result of frequency analysis using discrete Fourier transform is different from the result of frequency analysis using fast Fourier transform on the Y-axis scale, it can be confirmed that there is no problem because the shapes are consistent even if the shape is simplified by discrete Fourier transform. <Konfiguration des Headset-Systems der ersten beispielhaften Ausführungsform der Technik der vorliegenden Offenbarung>

[0030] A headset system according to a first exemplary embodiment of the technique of the present disclosure includes a headset 100 configured in Fig. 5 is shown.

[0031] As in Fig. As shown in Figure 5, the headset 100 includes a hollow housing 10 that is worn on a user's ear and houses various functional components. The headset 100 includes a tubular external ear canal insertion portion 12, which is a portion of the housing 10 and has a hollow portion provided at a portion on the ear canal side of the housing 10 when worn on a user's ear.

[0032] Furthermore, the headset 100 includes a driver 14 for outputting sound signals, which is provided within the housing 10.

[0033] Further, the headset 100 is provided with a microphone 16 so that signals transmitted into the hollow portion of the external auditory canal insertion portion 12 can be collected, with a reproduction unit 20 that outputs sound signals from the driver 14, and with a communication unit 23 that transmits an output signal of the microphone 16 to an information processing unit 50.

[0034] The playback unit 20 and the communication unit 23 are mounted on a main board (not shown) located within the housing 10.

[0035] When no sound signal is output from the driver 14, the signal output from the microphone 16 is transmitted from the communication unit 23 to the information processing unit 50. When no sound signal is output from the driver 14, the signal output from the microphone 16 is a waveform representing a pressure fluctuation in the enclosed space formed by the hollow portion of the housing 10 and the user's external auditory canal, and is a waveform that, for example, Fig. 1A, which was described above.

[0036] A headset system according to the first exemplary embodiment of the technique of the present disclosure includes the Fig. 6. Note that the information processing unit 50 is an example of a heart rate estimation device.

[0037] The information processing unit 50 is configured by a portable terminal, a computer terminal, or the like, or the information processing unit 50 is integrated into a headset system. Here, the portable terminals include smartphone terminals, mobile phones, and PDA (Personal Digital Assistant) terminals. Computer terminals include laptop and book computer terminals, as well as desktop computer terminals. The information processing unit 50 may be provided within the housing 10 of the headset 100 or may be provided separately from the headset 100.

[0038] The information processing unit 50 is configured by a computer including a CPU, a RAM, and a ROM in which a program for executing a heart rate estimation processing routine described below is stored, and is functionally configured as follows.

[0039] As in Fig. 6, the information processing unit 50 includes an input unit 60, a calculation unit 70, and an output unit 80. The calculation unit 70 includes a feature point extraction unit 72, a pulse waveform generation unit 74, a frequency analysis unit 76, and a heart rate estimation unit 78.

[0040] The input unit 60 receives a signal output from the microphone 16, which is received by the headset 100.

[0041] As in Fig. 1B, the feature point extraction unit 72 extracts waveform feature points from a signal output from the microphone 16.

[0042] More specifically, the feature point extraction unit 72 extracts, as a waveform feature point, a zero-crossing point having a fluctuation width equal to or greater than a threshold value from a signal output from the microphone 16, the waveform representing a pressure fluctuation in the enclosed space (see the x marks in Fig. 7).

[0043] The lower diagram of Fig. 7 is an enlarged view of the rectangular frame section in the upper diagram of Fig. 7. In the diagram below from Fig. 7. Of the zero-crossing points that exhibit a fluctuation range greater than or equal to a threshold before and after a zero-crossing, the zero-crossing points indicated by the arrow marks are the zero-crossing points presumably due to a heartbeat. The other zero-crossing points are zero-crossing points presumably due to body movements.

[0044] It should be noted that the threshold value of the fluctuation width in zero-crossing point extraction can be varied depending on the situation. It should be noted that since the periodicity of all zero-crossing points extracted as zero crossings with a fluctuation width greater than or equal to the threshold value appears in the frequency characteristic, cyclic feature points will appear in the cycle of a heartbeat, even in cases where a heartbeat and a zero-crossing point due to body movement are intermingled. However, the more waveform feature points appear due to body movement, the more cyclic feature points due to body movement will also appear, and therefore, it is more favorable if there are as few waveform feature points due to body movement as possible; however, it is not necessary to exclude waveform feature points due to body movement.

[0045] As long as a technique is capable of extracting waveform feature points, including points associated with a heartbeat, waveform feature points other than a zero crossing can also be extracted. For example, peak and trough points can be extracted as waveform feature points from a waveform representing a pressure fluctuation in a confined space. In such cases, peak and trough points are captured from a waveform representing a pressure fluctuation in a confined space ( Fig. 8), and from the recorded peaks and troughs, peaks and troughs are removed that are assumed to have been recorded erroneously ( Fig. 9). More precisely, as in Fig. As shown in Figure 8, points that exceed the running average of the respective upper limits are extracted as preliminary peak points (see the white circles) from the waveform representing the pressure fluctuation in the confined space. Points that are below the running average of the respective lower limits are extracted as preliminary trough points (see the black circles). Taking into account the time difference and amplitude magnitude from the preliminary trough to the preliminary peak point, the distance between valid preliminary peak points, and the distance between valid preliminary trough points, preliminary peak points and preliminary trough points that are suspected of being erroneously detected are removed.

[0046] Furthermore, a peak and a trough can be extracted from a waveform representing a pressure fluctuation in a confined space using machine learning and used as waveform feature points. For example, peaks and troughs can be extracted using the method described in Reference 1.

[0047] [Reference document 1]: “Robust ECG R-peak Detection Using LSTM,” Juho Laitala et al., SAC '20: Proceedings of the 35th Annual ACM Symposium on Applied Computing. March 2020, pages 1104-1111, https: / / doi.org / 10.1145 / 3341105.3373945

[0048] The pulse waveform generation unit 74 generates a pulse waveform in which a pulse of a predetermined amplitude is formed at the same time as a waveform feature point. As shown in Fig. For example, as shown in Figure 1D, pulse waveforms are generated in which pulse beats of a constant amplitude are formed at the same time as the waveform feature points. Note that the pulse beat of the pulse waveform does not need to have a constant amplitude and can be changed using any predefined value.

[0049] The frequency analysis unit 76 performs frequency analysis for the pulse waveform. More specifically, the frequency analysis unit 76 performs frequency analysis using discrete Fourier transform and obtains frequency characteristics as shown in Fig. 1F. Note that a fast Fourier transform can be used instead of a discrete Fourier transform.

[0050] The heart rate estimation unit 78 estimates the heart rate by analyzing cyclic feature points obtained from the results of the frequency analysis.

[0051] More precisely, as described above Fig. As shown in Figure 1G, the heart rate estimation unit 78 extracts cycles corresponding to frequencies with a frequency component greater than a threshold from the frequency analysis results as cyclic feature points. The heart rate estimation unit 78 specifies cyclic feature points corresponding to a heart rate interval among the extracted cyclic feature points and estimates a heart rate per unit time based on the specified cyclic feature points. <Betrieb des Headset-Systems der ersten beispielhaften Ausführungsform der Technik der vorliegenden Offenbarung>

[0052] When the casing 10 of the headset 100 is worn on the user's ear, an instruction to estimate the heart rate is received from the user's information processing unit 50 via wireless communication. At this time, the communication unit 23 in the headset 100 transmits a signal output from the microphone 16 to the information processing unit 50 when no sound signal is output from the driver 14.

[0053] When the information processing unit 50 receives the signal output from the microphone 16, the information processing unit 50 performs heart rate estimation processing as shown in Fig. 10 shown.

[0054] First, in step S100, the input unit 60 detects the output signal of the microphone 16 received by the headset 100.

[0055] In step S102, the feature point extraction unit 72 extracts waveform feature points from the signal output from the microphone 16.

[0056] In step S104, the pulse waveform generation unit 74 generates a pulse waveform in which a pulse beat of a predetermined amplitude is formed at the same timing as the waveform feature points.

[0057] In step S106, the frequency analysis unit 76 performs frequency analysis for the pulse waveform.

[0058] In step S108, the heart rate estimation unit 78 estimates the heart rate by analyzing cyclic feature points obtained from the results of the frequency analysis, displays the heart rate on the output unit 80, and ends the heart rate estimation processing.

[0059] As explained above, the heart rate estimation device according to the headset system according to the first exemplary embodiment of the present disclosure extracts waveform feature points from a waveform signal representing pressure fluctuations in a closed space output from a microphone of the headset. The heart rate estimation device estimates the heart rate by generating a pulse waveform in which pulse beats are formed at the same timing as the waveform feature points and by analyzing cyclic feature points obtained from the frequency analysis results of the pulse waveform. This enables accurate estimation of the heart rate from the pressure fluctuation in the closed space. In particular, it is possible to estimate the heart rate even in a state where body movements occur.

[0060] By extracting zero crossings having a fluctuation width equal to or greater than a threshold as waveform feature points from a signal of a waveform representing a pressure fluctuation in a closed space, the estimation accuracy of a heart rate is improved.

[0061] By using a discrete Fourier transform in the frequency analysis of pulse waveforms, a significant reduction in computational complexity is realized compared to cases where a fast Fourier transform is used.

[0062] In the technique of the present disclosure, waveform feature points are extracted from a waveform representing pressure fluctuations in a confined space by a feature point extraction unit. A pulse waveform generation unit generates a pulse waveform in which pulse beats are formed at the same timing as the waveform feature points. A frequency analysis unit performs frequency analysis of the pulse waveform. A heart rate is estimated by a heart rate estimation unit by analyzing cyclic feature points obtained from the frequency analysis results.

[0063] In this way, waveform feature points are extracted from waveforms representing pressure fluctuations in a confined space, a pulse waveform is generated in which pulse beats are formed at the same time as the waveform feature points, and a heart rate is estimated by analyzing cyclic feature points obtained from the frequency analysis results of the pulse waveform. This makes it possible to accurately estimate the heart rate based on the pressure fluctuations in the confined space.

[0064] The feature point extraction unit according to the technique of the present disclosure can extract, from a waveform representing pressure fluctuations in a closed space, a zero crossing point, a peak point, or a trough point having a fluctuation width equal to or larger than a threshold value as a waveform feature point.

[0065] The pulse waveform generation unit according to the technique of the present disclosure can generate a pulse waveform in which a pulse beat having a predetermined amplitude is formed.

[0066] The frequency analysis unit according to the technique of the present disclosure may perform the frequency analysis using a discrete Fourier transform.

[0067] The headset system according to the technique of the present disclosure is configured to include the above-described heart rate estimation device and a headset including a hollow housing worn on a user's ear, a driver for sound signal output provided inside the housing, a microphone provided inside the housing on the back of the driver, and an output unit that outputs an output signal of the microphone to the heart rate estimation device as a waveform representing pressure fluctuations in a closed space.

[0068] According to the technique of the present disclosure, in a headset, a signal output from the microphone via an output unit is output to the heart rate estimator as a waveform representing pressure fluctuations in a closed space.

[0069] Furthermore, the heart rate estimator estimates the heart rate based on the signal output from the microphone.

[0070] In this way, the heart rate can be accurately estimated from the pressure fluctuations in the enclosed space by using the signal emitted by the microphone as a waveform representing the pressure fluctuations in the enclosed space.

[0071] According to the technique of the present disclosure, in a headset, a signal output from the optical detection unit via an output unit is output to the heart rate estimator as a waveform representing pressure fluctuations in a closed space.

[0072] Furthermore, the heart rate estimator estimates the heart rate based on the signal output from the optical detection unit.

[0073] In this way, the heart rate can be accurately estimated from the pressure fluctuations in the enclosed space, using the signal output by the optical detection unit as a waveform representing pressure fluctuations in the enclosed space. [Second exemplary embodiment]

[0074] The following is an explanation regarding a headset system according to a second exemplary embodiment. Portions having the same configuration as in the first exemplary embodiment are denoted by the same reference numerals, and explanations thereof are omitted.

[0075] The second exemplary embodiment differs from the first exemplary embodiment in that biological information is acquired using a photoelectric pulse wave. <Konfiguration des Headset-Systems der zweiten beispielhaften Ausführungsform der Technik der vorliegenden Offenbarung>

[0076] A headset system according to a second exemplary embodiment of the technique of the present disclosure includes the Fig. 11. The headset 200 has the same configuration as the headset 100 according to the first embodiment described above and further includes a photoelectric pulse wave measuring device 216 that measures a photoelectric pulse wave. The photoelectric pulse wave measuring device 216 is an example of an optical detection unit.

[0077] The photoelectric pulse wave measuring device 216 includes a light emitting unit such as a near-infrared LED and an optical detecting unit such as a phototransistor. It measures a photoelectric pulse wave in the external auditory canal and outputs a signal detecting the measured photoelectric pulse wave. The signal output from the photoelectric pulse wave measuring device 216 is transmitted to the information processing unit 50 via the communication unit 23. Here, the signal output from the photoelectric pulse wave measuring device 216 is a waveform representing pressure fluctuations in the enclosed space formed between the hollow portion of the housing 10 and the user's ear, and is, for example, a waveform similar to the one described above. Fig. similar to the waveform described in Figure 1A.

[0078] The input unit 60 of the information processing unit 50 receives a signal output from the photoelectric pulse wave measuring device 216, which is received by the headset 200.

[0079] The feature point extraction unit 72 extracts waveform feature points from the signal output from the photoelectric pulse wave measuring device 216.

[0080] Note that the other configuration and operation of the headset system according to the second exemplary embodiment are the same as those of the first exemplary embodiment, and explanation thereof will be omitted.

[0081] As explained above, in the headset system according to the second exemplary embodiment of the technique of the present disclosure, the heart rate estimator extracts waveform feature points from the waveform signal representing pressure fluctuations in a closed space output from the photoelectric pulse wave measuring device of the headset. The heart rate estimator estimates the heart rate by generating a pulse waveform in which pulse beats are formed at the same timing as the waveform feature points and by analyzing cyclic feature points obtained from the results of frequency analysis performed on the pulse waveform. This makes it possible to accurately estimate the heart rate based on pressure fluctuations in the closed space. In particular, it is possible to estimate the heart rate even in a state where body movements occur. EXAMPLES

[0082] The following is an explanation regarding an example of a headset according to the first exemplary embodiment described above. As shown in Fig. 12, a housing 10 of the headset of the present exemplary embodiment is formed by joining a main housing 1a and a front housing 1b.

[0083] The main body 1a is a hollow member having a cylindrical shape overall, with a rear opening portion closed by a cover 2. Inside the main body 1a, a main board 3 is located facing the opening portion of the body. The main board 3 is a board on which electronic components functioning as the playback unit 20 and the communication unit 23 are mounted.

[0084] In front of the main board 3 there is a battery 6 with a battery pad 7 and a battery cover 8 in between.

[0085] On the outer circumference of the main case 1a there is a case rubber 9. The case rubber 9 is a cylindrical elastic member fitted into the outer circumference of the main case 1a, which weakens the contact with the ear and prevents the ingress of water into the case 10.

[0086] The front housing 1b is arranged to close a front cylindrical opening portion of the main housing 1a. The front housing 1b has an overall shape of an oblique truncated cone, with a peripheral portion thereof slightly raised toward the eardrum.

[0087] At the front of the front housing 1b, an external auditory canal insertion portion 12 is provided, which protrudes from an upper portion of the oblique truncated cone toward the eardrum. The external auditory canal insertion portion 12 has a cylindrical shape provided at a part of the front housing 1b, is open at the front and rear sides, and connects the interior of the front housing 1b to the outside world. A driver 14 having a cylindrical housing is installed inside the external auditory canal insertion portion 12. Thus, a positioning portion 11 of the driver 14 is provided adjacent to the front opening portion of the external auditory canal insertion portion 12, and the driver 14 is fixed to the inner surface of the external auditory canal insertion portion 12 because the front end portion of the driver 14 is engaged with the positioning portion 11.A rear end portion of the driver 14 is arranged to approach the vicinity of a front end portion of the front housing 1b. The driver 14 includes a magnetic circuit, a diaphragm, and the like for generating an output signal within a cylindrical housing, using any suitable known structure.

[0088] The headset of the present exemplary embodiment includes a microphone 16. The microphone 16 is arranged near the external auditory canal insertion portion 12 inside the front housing 1b, namely at the rear of the driver 14.

[0089] The microphone 16 is mounted on a microphone substrate 15. The microphone substrate 15 is attached to a block 416. The block 416 is a block-shaped member that supports the microphone 16 and the microphone substrate 15. Opening portions 15a, 16a are provided in the microphone substrate 15 and the block 416 so that acoustic signals in the external auditory canal can reach the microphone 16.

[0090] On the inner side of the external auditory canal insertion portion 12, there is a hollow portion 16A serving as a groove formed along the axial direction of the external auditory canal insertion portion 12. The hollow portion 16A is a square groove-like space formed between the inner surface and a side surface of the driver 14, and connects a front end portion of the external auditory canal insertion portion 12 to the opening portion 16a of the block 416 fixed to the front housing 1b.

[0091] An earpiece 13 is attached to the outer periphery of the external auditory canal insertion portion 12. The earpiece 13 is also referred to as an ear chip, ear cushion, or ear cap and is made of an elastic member such as silicone rubber. The earpiece 13 includes an adhesion portion 13a with respect to an external auditory canal wall surface and is formed into a hemisphere at a front end of a cylindrical portion 13b fitted into the outer periphery of the external auditory canal insertion portion 12. An earpiece fixing groove 412 is provided on the outer periphery of the external auditory canal insertion portion 12, and a fitting portion 13c is provided on the inner periphery of the cylindrical portion 13b of the earpiece 13, and the fitting portion 13c engages with the earpiece fixing groove 412 so that the earpiece 13 is fixed to the external auditory canal insertion portion 12.

[0092] As explained in the exemplary embodiments described above, the configuration is such that the microphone is located at the rear of the driver. This allows the microphone to be made smaller in relation to the volume of the enclosed space defined by the headset housing, which houses the necessary components of the configuration, and by the external auditory canal than in cases where it is located in front of the driver.

[0093] It should be noted that with respect to the enclosed space formed by the headset body and the external auditory canal, the smaller the volume of the enclosed space, the greater the pressure fluctuation, which includes the expansion and contraction of a blood vessel accompanying blood flow (see Reference Document 2). Therefore, by reducing the volume of the enclosed space using the configuration explained in the above-described embodiments, the pressure fluctuations in the enclosed space can be more accurately detected, and the heart rate can be accurately estimated. [Reference Document 2]: Japanese Patent No. 6082131

[0094] It should be noted that the technique of the present disclosure is not limited to the exemplary embodiments described above, and various modifications and applications are possible within a range that does not deviate from the gist of the technique of the present disclosure.

[0095] For example, in each of the exemplary embodiments described above, examples of cases where the technique of the present disclosure is applied to a headset were explained; however, the present disclosure is not limited thereto. The technique of the present disclosure may also have applications other than a headset, and the technique of the present disclosure may be applied to, for example, an electronic stethoscope.

[0096] The disclosure of Japanese Patent Application No. 2022-171796 is hereby incorporated by reference in its entirety.

[0097] All documents, patent applications, and technical standards described herein are hereby incorporated by reference to the same extent as if each document, patent application, and technical standard were specifically and individually described as being incorporated by reference. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] US 8,998,815

[0003] Cited non-patent literature

[0000] Robust ECG R-peak Detection Using LSTM“, Juho Laitala et al., SAC '20: Proceedings of the 35th Annual ACM Symposium on Applied Computing. März 2020, Seiten 1104-1111, https: / / doi.org / 10.1145 / 3341105.3373945

[0047]

Claims

[1] Heart rate estimation device comprising: a feature point extraction unit that extracts a waveform feature point from a waveform representing pressure fluctuations in a closed space; a pulse waveform generating unit that generates a pulse waveform constituting a pulse beat at the same timing as the waveform feature point; a frequency analysis unit that performs frequency analysis of the pulse waveform; and a heart rate estimation unit that estimates a heart rate by analyzing a frequency feature point obtained from a result of the frequency analysis. [2] The heart rate estimation device according to claim 1, wherein the feature point extraction unit extracts, as the waveform feature point, a zero crossing having a fluctuation range equal to or larger than a threshold value, a peak point, or a trough point from a waveform representing a pressure fluctuation in a closed space. [3] The heart rate estimating apparatus according to claim 1, wherein the pulse waveform generating unit generates the pulse waveform constituting a pulse beat of a predetermined amplitude. [4] The heart rate estimating apparatus according to claim 1, wherein the frequency analysis unit performs the frequency analysis by means of discrete Fourier transform. [5] Headset system comprising: the heart rate estimation device according to any one of claims 1 to 4; and a headset, the headset comprising: a hollow housing worn on a user's ear; a driver for an acoustic signal output provided inside the housing; a microphone located inside the housing on the back of the driver; and an output unit that outputs to the heart rate estimator a signal output from the microphone as a waveform representing a pressure fluctuation in a closed space. [6] Headset system comprising: the heart rate estimation device according to any one of claims 1 to 4; and a headset, the headset comprising: a hollow housing worn on a user's ear; a driver for an acoustic signal output provided inside the housing; an optical detection unit provided inside the housing on the back of the driver; and an output unit that outputs to the heart rate estimating device a signal output from the optical detection unit as a waveform representing a pressure fluctuation in a closed space.

Images