Headset on-ear status detection

Abstract

A method and apparatus for detecting whether a headset is on an ear. i ) of the acoustic playback detection signal (V i ). Receives microphone signal (X) from microphone (Ei) Ei ), the microphone signal comprising at least a portion of the detection signal received at the microphone. The microphone signal is transmitted to a state estimator (350) to generate an estimate of at least one parameter #imgabs0# of the portion of the detection signal contained in the microphone signal. The estimate of the at least one parameter is processed to determine whether the headset is on the ear.

Description

[0001] This application is a divisional application of the invention patent application with application date of June 29, 2018, application number 201880063689.9, and name “On-ear status detection of head-mounted receiver”. Technical Field

[0002] The present invention relates to headsets, and more particularly to headsets configured to determine whether the headset is in place on or in an ear of a user, and to methods for making such a determination. Background Art

[0003] A headset is a popular device for delivering sound to one or both ears of a user, such as for playback of music or audio files or telephone signals. A headset typically also captures sounds from the surrounding environment, such as the user's voice for voice recording or voice calls, or background noise signals for enhancing signals processed by the device. A headset may provide a wide range of signal processing capabilities.

[0004] For example, one such function is active noise cancellation (ANC, also called active noise control), which combines a noise cancellation signal with a playback signal and outputs the combined signal via a speaker so that the noise cancellation signal component acoustically cancels ambient noise, while the user hears only or primarily the playback signal of interest. The ANC process typically takes as input an ambient noise signal provided by a reference (feedforward) microphone and a playback signal provided by an error (feedback) microphone. The ANC process continues to consume a significant amount of power even if the headset is removed.

[0005] Therefore, in ANC, and similarly in many other signal processing functions of a headset, it is desirable to know whether a headset is being worn at any particular time. For example, it is desirable to know whether an on-ear headset is placed on or over the pinna of a user, and whether an earbud headset has been placed in the ear canal or concha of a user. Both of these use cases are referred to herein as the respective headset being "on ear". An unused state, such as when the headset is worn around the user's neck or removed entirely, is referred to herein as being "off ear".

[0006] Previous methods for on ear detection include the use of dedicated sensors, such as capacitive sensors, optical sensors, or infrared sensors, which can detect when a headset is worn on or near the ear. However, providing such non-acoustic sensors increases hardware costs and increases power consumption. Another previous method for on ear detection is to provide a sensing microphone that is positioned to detect acoustic sounds inside the headset when worn, based on the fact that acoustic reverberation inside the ear canal and / or auricle will cause a detectable rise in the power of the sensing microphone signal compared to when the headset is off the ear. However, the sensing microphone signal power is affected by noise sources such as wind noise, so this method may output a false positive that the headset is on the ear when the headset is off the ear and affected by noise in practice. These and other methods for on ear detection may also output false positives when the headset is held in the user's hand, placed in a box, etc.

[0007] Any discussion of documents, acts, materials, devices, articles and the like that has been included in this specification is solely for the purpose of providing a context for the present invention. It is not to be taken as an admission that any or all of these matters formed part of the prior art base or were common general knowledge in the field relevant to the present invention as it existed before the priority date of each claim of this application.

[0008] Throughout this specification, the word "comprise" or variations such as "comprises" or "comprising" will be understood to imply the inclusion of stated elements, integers or steps, or groups of elements, integers or steps, but not the exclusion of any other elements, integers or steps, or groups of elements, integers or steps.

[0009] In the present specification, a statement that an element may be “at least one of” a list of options should be understood that the element may be any one of the listed options, or may be any combination of two or more of the listed options. Summary of the invention

[0010] According to a first aspect, the present invention provides a signal processing device for on-ear detection of a headphone, the device comprising:

[0011] a probe signal generator configured to generate a probe signal for acoustic playback from a speaker;

[0012] an input for receiving a microphone signal from a microphone, the microphone signal comprising at least a portion of the detection signal received at the microphone; and

[0013] A processor is configured to apply a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the detection signal contained in the microphone signal, and the processor is further configured to process the estimate of the at least one parameter to determine whether the headphone is on the ear.

[0014] According to a second aspect, the present invention provides a method for on-ear detection of a headphone, the method comprising:

[0015] generating a probe signal for acoustic playback from a speaker;

[0016] receiving a microphone signal from a microphone, the microphone signal comprising at least a portion of the detection signal received at the microphone;

[0017] applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, and

[0018] Whether the headphone is on the ear is determined from the estimation of the at least one parameter.

[0019] According to a third aspect, the present invention provides a non-transitory computer readable medium for on-ear detection of a headset, the non-transitory computer readable medium comprising instructions which, when executed by one or more processors, result in the following operations:

[0020] generating a probe signal for acoustic playback from a speaker;

[0021] receiving a microphone signal from a microphone, the microphone signal comprising at least a portion of the detection signal received at the microphone;

[0022] applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, and

[0023] Whether the headphone is on the ear is determined from the estimation of the at least one parameter.

[0024] According to a fourth aspect, the present invention provides a system for on-ear detection of a headphone, the system comprising a processor and a memory, the memory containing instructions executable by the processor, and wherein the system is operable to:

[0025] generating a probe signal for acoustic playback from a speaker;

[0026] receiving a microphone signal from a microphone, the microphone signal comprising at least a portion of the detection signal received at the microphone;

[0027] applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, and

[0028] Whether the headphone is on the ear is determined from the estimation of the at least one parameter.

[0029] In some embodiments of the invention, the processor is configured to process the estimate of the at least one parameter to determine whether the headset is on the ear by comparing the estimated parameter to a threshold.

[0030] In some embodiments of the invention, the at least one parameter is an amplitude of the detection signal. In some embodiments, when the amplitude is above a threshold, the processor is configured to indicate that the headset is on the ear.

[0031] In some embodiments of the present invention, the detection signal comprises a single tone. In other embodiments of the present invention, the detection signal comprises a weighted multi-tone signal. In some embodiments of the present invention, the detection signal is limited to an inaudible frequency range. In some embodiments of the present invention, the detection signal is limited to a frequency range less than a threshold frequency below a typical human hearing range. In some embodiments of the present invention, the detection signal varies over time. For example, the detection signal may vary in response to a change in the level of ambient noise within the frequency range of the detection signal.

[0032] Some embodiments of the present invention may further include a down converter configured to down-convert the microphone signal before the state estimation to reduce the computational burden required for the state estimation.

[0033] In some embodiments of the invention, a Kalman filter implements the state estimation.In such embodiments, a copy of the detection signal generated by the detection signal generator may be transmitted to a prediction module of the Kalman filter.

[0034] In some embodiments of the present invention, the determination device module is configured to generate a first probability that the headphone is on the ear and a second probability that the headphone is off the ear from the at least one parameter, and the processor is configured to use the first probability and / or the second probability to determine whether the headphone is on the ear. In such an embodiment, the determination device module may compare the at least one parameter with an upper threshold level to determine the first probability. In some embodiments, the state estimation generates a sample-by-sample estimate of the at least one parameter, and the estimate is considered on a frame basis to determine whether the headphone is on the ear, each frame including N estimates, and for each frame, the first probability is calculated as N ON / N, where N ON is the number of samples in the frame where the at least one parameter exceeds the upper threshold.

[0035] In some embodiments of the invention, the decision device module may compare the at least one parameter to a lower threshold level to determine the second probability. In some embodiments, the state estimation produces a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, the second probability is calculated as N OFF / N, where N OFF is the number of samples in the frame in which the at least one parameter is less than the lower threshold.

[0036] In some embodiments of the present invention, the decision device module is configured to generate an uncertainty probability from the at least one parameter, the uncertainty probability reflecting the uncertainty of whether the headphone is on the ear or off the ear, and the processor is configured to use the uncertainty probability to determine whether the headphone is on the ear. In some embodiments, the state estimation can produce a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headphone is on the ear, each frame including N estimates, and wherein for each frame, the uncertainty probability is calculated as N UNC / N, where N UNC is the number of samples in the frame in which the at least one parameter is greater than the lower threshold and less than the upper threshold. In some such embodiments, the processor may be configured to not change a previous determination of whether the headset is on the ear when the uncertainty probability exceeds an uncertainty threshold.

[0037] In some embodiments of the present invention, a change is made in the determination of whether the headphone is on the ear having a first decision delay from off-ear to on-ear, and having a second decision delay from on-ear to off-ear, the first decision delay being less than the second decision delay, thereby biasing the determination toward an on-ear determination.

[0038] In some embodiments of the present invention, the level of the detection signal can be dynamically changed to compensate for changing headphone occlusion. Such embodiments can further include an input for receiving a microphone signal from a reference microphone of the headphone for capturing external ambient sound, and wherein the processor is further configured to apply a state estimation to the reference microphone signal to produce a second estimate of the at least one parameter of the detection signal, and wherein the processor is further configured to compare the second estimate with the estimate to distinguish ambient noise from supra-auricular occlusion.

[0039] In some embodiments of the present invention, the system is a headset, such as an earbud. In some embodiments, an error microphone is mounted on the headset so that when the headset is worn, the error microphone senses the sound that occurs in the space between the headset and the eardrum of the user. In some embodiments, a reference microphone is mounted on the headset so that when the headset is worn, the reference microphone senses the sound that occurs outside the headset. In some embodiments of the present invention, the system is a smart phone or other such host device that can interoperate with the headset. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Embodiments of the present invention will now be described with reference to the accompanying drawings, in which:

[0041] Figure 1a and Figure 1b A signal processing system including a wireless earbud-type headset is illustrated, wherein on-ear detection is implemented;

[0042] Figure 2 is a generalized schematic diagram of an ANC headset with the proposed on-ear detector;

[0043] Figure 3 yes Figure 2 A more detailed block diagram of an ANC headset illustrating the state tracking of the on-ear detector of the present invention in more detail;

[0044] Figure 4 Is Figure 2 and Figure 3 A block diagram of a Kalman amplitude tracker implemented by a supra-auricular detector;

[0045] Figure 5a to Figure 5e The application of multiple decision thresholds and decision probabilities to improve the stability of the supra-ear detector output is illustrated;

[0046] Figure 6 is a block diagram of a supra-ear detector according to another embodiment of the present invention, the supra-ear detector implementing dynamic control of the detection signal; and

[0047] Figure 7 is an example Figure 6 Flow chart of an implementation scheme for dynamic control of the detection signal.

[0048] Corresponding reference numerals indicate corresponding parts throughout the drawings. DETAILED DESCRIPTION

[0049] Figure 1a and Figure 1b An ANC headset 100 is illustrated in which on-ear detection is implemented. The headset 100 comprises two wireless earbuds 120 and 150, each of which comprises two microphones 121, 122 and 151, 152, respectively. Figure 1b 1 is a system schematic diagram of the earplug 120. The earplug 150 is configured in substantially the same manner as the earplug 120 and is therefore not shown or described separately. The digital signal processor 124 of the earplug 120 is configured to receive microphone signals from earplug microphones 121 and 122. Microphone 121 is a reference microphone and is positioned to sense ambient noise from outside the ear canal and outside the earplug. In contrast, microphone 122 is an error microphone and is positioned inside the ear canal in use to sense acoustic sounds including the output of speaker 128 in the ear canal. When the earplug 120 is positioned in the ear canal, microphone 122 is blocked to a certain extent from the external ambient acoustic environment, but maintains good coupling with the output of speaker 128, while microphone 121 is blocked to a certain extent from the output of speaker 128, but maintains good coupling with the external ambient acoustic environment. The headphone 100 is configured for a user to listen to music or audio, make phone calls, and pass voice commands to a voice recognition system, as well as other such audio processing functions.

[0050] Processor 124 is also configured to adapt the manipulation of such audio processing functions in response to one or both earplugs being placed on or removed from the ear. Earplug 120 further includes memory 125, which can actually be set as a single component or multiple components. Memory 125 is configured to store data and program instructions. Earplug 120 further includes transceiver 126, which is configured to allow earplug 120 to communicate wirelessly with external devices (including earplug 150). In alternative embodiments, such communication between earplugs can include wired communication, wherein a suitable line is set directly between the left and right sides of the headset (such as, in the overhead band) or via an intermediate device (such as, a smart phone). Earplug 120 also includes a speaker 128 for transmitting sound to the user's ear canal. Earplug 120 is powered by a battery and may include other sensors (not shown).

[0051] Figure 2 1 is a generalized schematic diagram of an ANC headset 100, illustrating in more detail the process of on-ear detection according to an embodiment of the present invention. L , and the right reference microphone 151 is also denoted as R R The left and right reference microphones generate signals X RL and X RR The left error microphone 122 is also denoted as E L , and the right error microphone 152 is also denoted as E R , and the two error microphones generate signals X EL and X ER The left earbud speaker 128 is also denoted as S L , the right earbud speaker 158 is also denoted as S R The left earbud playback audio signal is represented as U PBL , the audio signal played back by the right earbud is represented as U PBR .

[0052] According to the current embodiment of the present invention, the processor 124 of the earplug 120 executes the on-ear detector 130 or OED L , so as to acoustically detect whether the earbud 120 is on or in the user's ear. The earbud 150 performs the equivalent of OED R 160. In this embodiment, the output of the corresponding on-ear detector 130, 160 is transmitted to the corresponding acoustic detection generator GEN as an enable signal or a disable signal. L , GEN R When the acoustic detection generator is enabled, the acoustic detection generator will generate an inaudible acoustic detection signal U IL, U IR The inaudible acoustic detection signal will be added to the corresponding playback audio signal. The output of the corresponding on-ear detector 130, 160 is also used as the signal D L , D R is sent to the decision combiner 180, which generates the overall ear decision D ∑ .

[0053] In the following, i is used to represent L [left] or R [right], and it should be understood that according to various embodiments of the present invention, the described processing can be operated in only one headphone, independently in two headphone, or in cooperation with each other in two headphone. Figure 2 As shown, each earbud is equipped with a speaker S i , reference microphone R i and error microphone E i To play back the signal from the host playback device U PBi , the inaudible detection signal U may be added depending on the value of the "enable" flag from the control module Ii :1-add detection; 0-do not add detection. Inaudible detection U Ii By the corresponding detection generator GEN i The specific value of the "enabled" flag, 0 or 1, depends on a variety of factors, such as the operating environment conditions of the device, ambient noise levels, the presence of playback, headphone design, and other such factors. The resulting signal is passed through the ANC i , the ANC i A conventional ANC function is provided which adds a signal constituting a certain amount of estimated unwanted anti-phase noise. To this end, the ANC i From the reference microphone R i and error microphone E i Get input. Then, ANC i The output is sent to the speaker S i , to be played to the user's ears. Therefore, ANC requires the presence of microphones 121 and 122 and speaker 128, and the on-ear detection solution of the present invention does not require additional microphones, speakers or sensors. The output from the speaker generates a signal X Ri , the signal X Ri contains a certain amount of uncompensated noise in the i reference microphone; similarly, it generates the signal X in the i error microphone Ei .

[0054] Figure 31 is a block diagram of an earbud of an ANC headset 100 including an on-ear detector according to an embodiment of the present invention. Each earbud 120, 150 is equipped with a speaker S i , reference microphone R i and error microphone E i The playback signal U from the host playback device i With inaudible detection signal V i Added, the inaudible detection signal V i By the corresponding detection generator GEN i 320 is generated. You can use a high pass filter HPF i 310 Filter the playback signal to prevent the playback content U i and detect V i The resulting signal is sent to the ANC i 330, ANC i 330 provides a conventional ANC function that adds a certain amount of estimated unwanted anti-phase noise. i The generated signal X Si is transmitted to the speaker S i , the speaker S i The signal is played back acoustically. From the loudspeaker S i The output of the generated signal X Ri , the signal X Ri Includes reference microphone R i A certain amount of uncompensated noise in the error microphone E i The signal X in Ei .

[0055] Error microphone signal X Ei In the down converter ↓N i 340 is down-converted to the necessary sampling rate and then fed into the state tracker 350. The state tracker 350 performs state estimation to continuously estimate or track the down-converted error microphone signal For example, the state tracker 350 may track the down-converted error microphone signal The amplitude of the detection signal present in . The estimated detection signal parameters is transmitted to the decision device DD 360, where a decision D is made as to whether the corresponding earplug is on the ear. i In this way, the individual decisions D generated in both the left earplug and the right earplug are iThey may be used independently, or may be combined (eg, ANDed) to produce an overall determination of whether a corresponding headphone is on the ear, or whether both heads are on the ear.

[0056] In this embodiment, by limiting the detection signal to have spectral content B below the nominal human audible threshold IPS To make the detection signal inaudible, in this embodiment B IPS ≤ 20 Hz. In other embodiments, the detection signal may occupy slightly higher frequency components, rather than being strictly inaudible.

[0057] Importantly, according to the present invention, the form that the probe signal must take is capable of being tracked using a state estimation or state space representation to track the acoustic coupling of the probe signal from the playback speaker to the microphone. This is important because considerable noise may be generated at the same frequency as the probe signal, such as wind noise. However, the present invention recognizes that such noise typically has an incoherent variable phase and therefore will not tend to corrupt or deceive a state space estimator that is tuned to seek a known coherent signal. This is in contrast to simply monitoring the power in the frequency band occupied by the probe signal, as such power monitoring will be corrupted by the noise.

[0058] An example of an inaudible detection signal according to one embodiment of the present invention may be represented as follows:

[0059]

[0060]

[0061] Where N is the number of harmonic components; w n ∈[0, 1] is the weight of the corresponding component; A n 、f 0n and f s are the amplitude, fundamental frequency and sampling frequency respectively. For example, if N = 1 and w 1 =1, then the detection signal has amplitude A and frequency f 0 In other embodiments within the scope of the present invention, it is contemplated that many other suitable detection signals may be used.

[0062] The estimated amplitude output by the state tracker 350 (or estimated range The sum of ) can be used as an on-ear detection feature. This can be achieved by limiting the higher The value corresponds to the on-ear state because during this state, more of the probe signal energy is captured by the error microphone due to the occlusion of the ear canal and the limitation of the speaker output within the ear canal. The value φ is defined to correspond to the off-ear state because during this state, more sound pressure of the probe signal output by the loudspeaker escapes into free space without being restricted by the ear canal, and therefore less of the probe signal is captured by the error microphone.

[0063] In the following, for clarity, single component detection is discussed, but it should be understood that other embodiments of the present invention may equivalently utilize weighted multi-tone detection according to EQ1, or any other detection representable by a state-space model, within the scope of the present invention.

[0064] For clarity, we now omit the index i and introduce k to denote the sample. It is important to note that for a given nth fundamental frequency f 0 , detect V k can be generated recursively as follows:

[0065]

[0066] Where V 1,k is the in-phase (cosine) component at time k, V 2,k is the orthogonal (sinusoidal) component at time k, V 1,k-1 is the in-phase (cosine) component at time k-1, V 2,k-1 is the orthogonal (sinusoidal) component at time k-1, and φ is defined by EQ2.

[0067] The amplitude of the generated detection is determined by the initial state vector is limited and can be calculated as follows:

[0068]

[0069] In matrix form, EQ3 can be written as

[0070]

[0071]

[0072] Each nth component in EQ1 has a dedicated recursive generator matrix Φ n .

[0073] Other types of recursive orthogonal generators are possible. The orthogonal generator described by EQ3 is given as an example only.

[0074] In this embodiment, HPF 310 filters the input audio to prevent spectral overlap between the playback content and the detection. For example, if the detection is a signal with a frequency f 0 =20Hz cosine wave (EQ1, N=1), the cutoff frequency of the HPF should be selected so that f 0 Again, alternative embodiments within the scope of the invention may utilize higher cutoff frequencies as the intended use permits, with the caveat that such filtering will remove low frequency components of the playback signal of interest, which may become undesirable.

[0075] The detection generator GEN 320 generates an inaudible detection signal whose spectral content is below the nominal human audible threshold. One embodiment considered here is that the detection signal has an amplitude A and a fundamental frequency f 0 The cosine wave, as given by EQ1 (N = 1, w 1 =1).

[0076] The inaudible detection signal may be a continuous stationary signal, or its parameters may vary over time while remaining within the scope of the present invention. The characteristics of the detection signal (e.g., number of components N, frequency f 0n , Amplitude A n , spectrum shape w n ) can vary depending on a preconfigured sequence or in response to signals on other sensors. For example, if a large amount of ambient noise is present at the same frequency as the detection, the detection signal can be conditioned by GEN 320 to change the detection frequency or any detection signal parameters (amplitude, frequency, spectral shape, etc.) to keep the detection signal clearly observable even in the presence of such ambient noise.

[0077] The detection generator GEN 320 may be implemented as a hardware tone / multi-tone generator, a recursive software generator, a look-up table, and any other suitable signal generation means.

[0078] Turning again to the down converter ↓N 340, it is noted that the highest f in the error microphone signal 0n The above spectral content is not necessary for on-ear detection, which must only consider the low frequency band occupied by the detection signal. Therefore, in this embodiment, the error microphone signal sampling rate f is first converted by a down-converter ↓N 340. s Down-conversion is performed to reduce the computational burden added by the on-ear detection and further reduce the power consumption of the on-ear detector. The down-converter ↓N 340 can be implemented as a low-pass filter (LPF) followed by a down-sampler. For example, the sampling frequency of the on-ear detector can be reduced to a value f by correspondingly selecting the LPF cut-off frequency and the down-sampling rate.s ≥2*f 0n Naturally, the sampling rate of the detection generator 320 and the output of the down-converter ↓N 340 should be the same. 0n =20Hz, it is recommended to use f s ∈[60, 120] Hz.

[0079] Figure 4 The state tracker 350 is illustrated in more detail. In this embodiment, the on-ear state tracker 350 is based on a Kalman filter used as an amplitude estimator / tracker. Again, the playback audio signal is high pass filtered at 310 and then compared with the detection signal V generated by the detection generator 320. 1,K The resulting audio signal is played through the speaker S 128. It should be emphasized that the inaudible detection does not have to be generated by the recursive generator Φ(EQ5). This is shown only to highlight the state space nature of the method adopted by the present invention. In practice, the detection V 1,K It can be generated by a hardware tone / multi-tone generator, a recursive software generator, a lookup table, or other suitable means.

[0080] The audio signal acoustically output by the speaker S 128 is captured by the error microphone 122 and after being reduced in rate by the down converter ↓ N 340, the signal is input into the state tracker 350. The Kalman filter-based state tracker 350 includes a "prediction" module 410 and an "update" module 420. During the "prediction" step, the corresponding submodule 410 locally regenerates the detection signal V 1,K Here again, the inaudible detection does not have to be generated by the recursive generator Φ(EQ5), but it is shown as such to highlight the state-space nature of the approach taken by the present invention. In other embodiments within the scope of the present invention, the detection can be generated in module 410 by a hardware tone / multi-tone generator, a recursive software generator, a lookup table, etc.

[0081] The “update” module 420 obtains the down-converted error microphone signal and the inaudible detection signal V provided by module 410 1,K , and perform a convex combination of the two:

[0082]

[0083] Where G is the Kalman gain. The Kalman gain G can be calculated "on the fly" using Kalman filter theory and is therefore not discussed further. Alternatively, in the case where the Kalman gain calculation does not rely on real-time data, the gain G can be pre-calculated to reduce the real-time computational load.

[0084] After the prediction / update step is completed, the amplitude of the detection signal is estimated by the amplitude estimator (AE 430) according to EQ4.

[0085] return Figure 3 , the estimated amplitude of the detection signal is fed to decision device DD 360 where it can be integrated from the current sampling rate to the required detection time resolution (in one embodiment, a suitable time resolution value is 200ms) and compared to a predefined threshold value T D The comparison is performed to produce a binary decision D. In more detail, this step is implemented as follows:

[0086]

[0087] The decision device 360 ​​is input with the instantaneous (sample-by-sample) detection amplitude estimate from the Kalman amplitude tracker 350 and is calculated by t D The defined temporal resolution is used to generate binary supra-aural decisions.

[0088] While in some applications a simple threshold decision made by DD 360 in this implementation may be sufficient, in some cases this may return a high rate of false positives or false negatives as to whether the headphone is on the ear, or may be overly variable in alternating between on-ear and off-ear decisions.

[0089] Therefore, the following embodiments of the present invention are also proposed to provide a more sophisticated method for the decision device 360 ​​to improve the robustness and stability of the on-ear detection output. Figure 5a to Figure 5e The derivation of this solution is illustrated in the signal diagram of FIG.

[0090] produce Figure 5a to Figure 5e The test scenario for the data included a LiSheng headset with a mold, in a public bar environment with the user's own voice, and no playback audio. The probe signal used consisted of a 20Hz tone producing 66dB SPL. ANC was turned off and there was no wind noise. Figure 5a shows a down-converted error microphone signal (the estimate is based on the down-converted error microphone signal), and Figure 5b The output of the Kalman tracker 350 is shown, which is the estimated pitch amplitude. Figure 5a and Figure 5bVisual inspection of the may indicate that the earplug is removed at approximately sample 4,000 and then returned to the ear at approximately sample 7,500, but it can also be seen that the user's manipulation of the earplug makes these transitions unclear and non-instantaneous, particularly around the time period of sample 7,000 to sample 8,500.

[0091] Figure 5c is a graph of the raw pitch amplitude estimate generated by the tracker 350. It is worth noting that it is difficult to use any one threshold as the decision point for whether the headset is on or off the ear, because if only one decision threshold is used to evaluate Figure 5c If the data is large, there will inevitably be many false positives and / or false negatives. Figure 5c As shown in FIG, instead, the Kalman tracker and decision module in this embodiment does not apply one detection threshold, but two thresholds, an upper threshold T Upper and lower threshold T Lower Then, the original pitch amplitude estimate A in this implementation is EST Divided into N D - sample frame, and T upper and T Lower It should be noted that the threshold T depends on the speaker and microphone hardware, the headset form factor and the degree of occlusion of the headset when worn, and the power of the playback detection signal. upper and T Lower The value to which it is set will vary, so choosing an appropriate such threshold below the "on the ear" amplitude and above the "off the ear" amplitude will be an implementation step.

[0092] Figure 5d The application of this two-threshold determination device is illustrated. OFF ), the probability that the headset is on the ear (P ON ) and uncertainty probability (P UNC ) is calculated. If P UNC Less than the uncertainty threshold T unc , then by OFF With confidence threshold T Confidence Compare to update the on-ear detection decision. UNC Exceeding the uncertainty threshold T unc , the previous state will remain because there is too much uncertainty to make any new decisions. Although there is uncertainty throughout the period of about 7,500 samples to 8,500 samples (this is Figure 5a to Figure 5d ), but if Figure 5eAs shown in , the method described in this embodiment still outputs a clear on-ear judgment or off-ear judgment. A further improvement of this embodiment is to bias the final judgment towards on-ear judgment rather than off-ear judgment, because most DSP functions should be enabled quickly when the device is on the ear, and can be disabled more slowly when the device is off the ear. To this end, Figure 5d The confidence threshold in is greater than 0.5. In addition, a rule is applied that the state decision is changed from on-ear to off-ear only when the off-ear state is indicated at least a minimum number of times in a row.

[0093] Therefore, in the embodiment of FIG. 5, increasing t D A window is thus created that spans multiple data points to reduce the variability associated with instantaneous (sample-by-sample) decisions, noting that it is unlikely that the user will be able to change the position of the headset at a rate even close to the sampling rate. Additionally, it is worth noting that two thresholds are considered to increase the confidence in either the on-ear or off-ear decisions and to create an intermediate "uncertain" state that is useful for disabling changes to the on-ear state decision when the confidence is low. In other words, a confidence level is introduced so that the output state indication is only changed when the confidence level is sufficient, and this is changed repeatedly over time, which introduces a certain lag into the output indication, thereby reducing variability in the output, such as Figure 5e Clearly shown in.

[0094] The algorithm is applied to implement the process illustrated in FIG5 as follows. First, the incoming estimated pitch amplitude A EST Each is conditionally subdivided into N D multiple frames of samples, so N D =t D *F S , where F S is the sampling frequency after down-conversion (e.g., 125 Hz). Then, N D Each of the amplitude estimates is compared with two predefined thresholds T upper and T Lower The comparison is done to generate three probabilities: ON , p OFF and p UNC (respectively, the probability of the headset being on the ear, the probability of the headset being off the ear, and the probability of being in an uncertain state) are as follows:

[0095] a. If A EST <T Lower , then the ear-off counter N is incremented OFF

[0096] b. If A EST >T upper , then increment the ear counter N ON

[0097] c. If A EST >=T Lower And A EST <=T upper , then increment the uncertainty counter N UNC

[0098] d. After processing all N D After samples, estimate the probability:

[0099] P OFF =N OFF / N D ;P ON =N ON / N D ;P UNC =N UNC / N D ,

[0100] Therefore, every N D samples (or equivalently, every t D seconds) to update the probability.

[0101] If the uncertainty probability is low (below a predefined threshold T UNC ), so that P UNC <T UNC , then the above-ear judgment is updated as follows, where the low P UNC Represents a reliable estimate:

[0102] a. If P OFF >=T Conf , then DECISION = OFF-EAR ("1"), where T Conf is the predefined confidence level

[0103] b. If P OFF <T Conf , then DECISION=ON-EAR(“0”)

[0104] If the uncertainty probability is high (higher than a predefined threshold T UNC ), so that P UNC >=T UNC , then it will remain in the previous judgment interval t D The ear judgment made at the place. High P UNC Indicates an unreliable estimate (such as a low SNR that may be caused by a loose fit or high levels of low frequency noise).

[0105] If uncertain, the resulting on-ear decision is further biased towards on-ear. For this reason, only one "positive" decision (DECISION == ON-EAR) is sufficient to switch from the off-ear state to the in-ear state. This means that the decision delay in this case is exactly t D However, for the transition from the on-ear state to the off-ear state, M consecutive “positive” decisions (e.g., 4) are necessary. This means that the delay for this situation is at least M*t D seconds. Therefore, if DECISION == ON-EAR, it is passed to the detector output as is. If DECISION == OFF-EAR, the corresponding counter C OFF If DECISION is not equal to OFF-EAR during M decision intervals, C OFF Reset. If C OFF ==M, then only DECISION ==OFF-EAR is passed to the output.

[0106] Supra-aural detection according to any embodiment of the present invention may be performed independently for each ear. The resulting decisions may then be combined in one overall decision (eg, by ANDing the decisions made for the left and right channels).

[0107] It has been shown that the embodiments described above perform well in the task of on-ear detection, especially if there is a considerable blockage from the inside of the ear canal to the external environment, since in this case there is a high detection-to-noise ratio in the error microphone signal.

[0108] On the other hand, the following embodiments of the invention may be particularly suitable for headset form factors that have poor occlusion, which may occur, for example, due to poor headset design, different user anatomy, improper positioning, and the use of inappropriate tips on earbuds. The following embodiments may additionally or alternatively be suitable when high levels of low-frequency noise are present. These scenarios effectively reflect a reduced SNR (in this context, it refers to the detection-to-noise ratio). The SNR can be reduced "from above" in the sense that the detector receives fewer detection signals, and / or the SNR can be reduced "from below" when a large amount of low-frequency noise deteriorates the SNR. The following embodiments address this scenario by implementing a Kalman state tracker within a closed-loop control system.

[0109] Figure 6 is a block diagram of another embodiment of an on-ear detector that specifically allows for dynamic control of the amplitude of the detection signal in response to adverse blockages and / or high noise. Specifically, Figure 6The on-ear detector includes a closed-loop control system in which the level of the detection signal is dynamically changed to compensate for the effects of adverse occlusion.

[0110] exist Figure 6 6, the speaker S 628 transmits a probe signal at a nominal (loud) level in order to maintain a nominal sound level at the error microphone 622. The probe signal is generated by the generator 620 and mixed with the playback audio, high pass filtered by the HPF 610 to remove (inaudible) frequency content occupying the same frequency band as the probe signal. It should be noted that the mixing is done at the sampling rate of the playback audio. The probe signal mixed with the audio playback content is played by the speaker 628 and captured by the error microphone 622, downsampled to a lower sampling rate in the down converter ↓ module 640. This has the effect of largely removing the playback content from the error microphone signal. The level of the probe signal generated at the error microphone is estimated and tracked by the "Kalman E" amplitude tracker 650.

[0111] Once a blockage is detected, i.e., an increase in the error microphone 622 signal level, the level of the detection signal from the generator 620 is dynamically reduced by applying a gain G. The gain G is calculated and interpolated in the gain interpolation module 680 and is used to control the level of the detection signal at the speaker S 628 to maintain the desired level at the error microphone 622. G is also used as a metric by the decision device DD 690 to assist in making a decision as to whether the earbud is on-ear or off-ear. If the gain G becomes low (a large negative number), the on-ear state is indicated and / or output.

[0112] This embodiment also recognizes that if only the error microphone 622 signal is used for detection, false positives (decision device 690 indicates a headphone is on the ear condition when in fact the headphone is off the ear) may occur too often. This is because when the error microphone 622 signal level increases due to in-band ambient noise (which does not indicate an on-ear condition), it may have the same effect on the detector as a blockage (which indicates an on-ear condition), resulting in a false positive. Therefore, in Figure 6 In an embodiment of the present invention, this problem is solved by utilizing reference microphone 624 for the purpose of determining whether an increase in the error microphone 622 signal level is due to a blockage.

[0113] When in-band ambient noise is present, the reference microphone 624 will experience the same (or within a certain range Δ) increase in noise level as the error microphone 622. Therefore, an additional Kalman state tracker Kalman R 652 is provided to track the signal level of the reference microphone 624. The gain G can then be increased to amplify the detection signal (up to a maximum level) to compensate for the in-band noise, thereby maintaining the SNR within the range necessary for reliable detection. This is implemented by simultaneously tracking the detection signal levels at the error microphone 622 and the reference microphone 624. In turn, when the gain G applied to the detection at the speaker provides P ERR >P REF +Δ, the decision device 690 reports that the headset is on the ear, where P ERR is the tracked detection level at the error microphone 622, P REF is the tracked detection level at the reference microphone 624, and Δ is a predefined constant. If this condition is not met and the speaker 628 reaches its maximum value, the decision device 690 reports that the headphone is off the ear.

[0114] Figure 7 It is a further example Figure 6 Flowchart of implementation plan. Figure 7 The OED starts at 700 in an off-ear state (corresponding to a nominal level of the radiation detection signal) and sets the gain G to G at 710. MAX And the decision state is set to off-ear at 720. The process then continues to 730 where a "control" signal comprising the difference between the reference microphone signal (plus a constant offset Δ) and the error microphone signal is used to adjust the gain G as described above. At step 740, G is compared to G MAX If the adjusted gain output by step 730 is less than the maximum gain G MAX , then at 750, the determination is updated to indicate that the headphone is on the ear. Otherwise, at 720, the determination is updated to indicate that the headphone is off the ear.

[0115] In a similar Figure 6 In another embodiment of the invention, the level of the probe signal at the loudspeaker can be used as a detection metric. This exploits the observation that the lower the level of the probe signal at the loudspeaker, the more likely the headset is on the ear. Thus, such other embodiments of the invention may provide another Kalman filter "Kalman S" to track the level of the probe signal at the loudspeaker S for this purpose.

[0116] Other embodiments of the present invention may provide an averaged or smoothed hysteresis in the determination of whether the headphone is on the ear or off the ear. This may be applied to a single threshold embodiment, such as the embodiment of DD 360, or may be applied to a multi-threshold embodiment, such as the embodiment shown in FIG. 5. In particular, in such other embodiments, the hysteresis may be implemented, for example, by providing that the indication of the state of changing from off the ear to on the ear is only after the determination device indicates that the headphone is on the ear for more than 1 second. Similarly, the indication of the state of changing from on the ear to off the ear is only after the determination device indicates that the headphone is off the ear for more than 3 seconds. The time periods of 1 second and 3 seconds are suggested here only for illustrative purposes, and any other suitable value within the scope of the present invention may be adopted instead.

[0117] The preferred embodiment also provides for automatically shutting down the OED 130 once the headset has been away from the ear for more than 5 minutes (or any suitable equivalent period of time). This allows the OED to provide a useful function when the headset is frequently used and frequently moved on the ear, but also allows the headset to conserve power when away from the ear for extended periods of time, after which the OED 130 can be reactivated the next time the device is powered on or activated for playback.

[0118] Embodiments of the present invention may include a USB headset with a USB cable connection that enables a data connection to a host device and enables power from the host device. In providing on-ear detection that requires only one or more acoustic microphones and one or more acoustic speakers, the present invention may be particularly advantageous in such embodiments because USB earbuds typically require very small components and have a very low price point, inspiring the elimination of non-acoustic sensors, such as capacitive sensors, infrared sensors, or optical sensors. Another benefit of eliminating non-acoustic sensors is that it avoids the requirement to provide additional data lines and / or power lines in the cable connection, otherwise the additional data lines and / or power lines must be dedicated to such non-acoustic sensors. Therefore, in this case, it is particularly advantageous to provide an on-ear detection method that does not require non-acoustic components.

[0119] Other embodiments of the present invention may include a wireless headset (such as a Bluetooth headset) with a wireless data connection to a host device, and having an onboard power source (such as a battery). In such an embodiment, the present invention may also provide particular advantages in avoiding the need for limited battery power consumed by the non-acoustic on-ear sensor components.

[0120] Therefore, the present invention attempts to solve on-ear detection by acoustic means only (ie, by using the existing speakers / drivers, error microphones and reference microphones of the headset).

[0121] Knowing whether a headset is on the ear can be used in simple cases to disable or enable one or more signal processing functions of the headset. This can save power. This can also avoid undesirable scenarios where signal processing functions adversely affect device performance when the headset is not in the expected position (whether on the ear or off the ear). In other embodiments, knowing whether a headset is on the ear can be used to modify the operation of one or more signal processing functions or playback functions of the headset so that such functions adaptively respond to whether the headset is on the ear.

[0122] It will be appreciated by persons skilled in the art that numerous variations and / or modifications may be made to the invention as shown in the specific embodiments without departing from the spirit or scope of the invention as broadly described.

[0123] For example, although in the described embodiments the state tracker is based on a Kalman filter used as an amplitude estimator / tracker, other embodiments within the scope of the invention may alternatively or additionally use other techniques for state estimation to estimate the acoustic coupling of the detection signal from the speaker and the microphone, such as an H∞ (H infinite) filter, a nonlinear Kalman filter, an unscented Kalman filter, or a particle filter.

[0124] The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive.

[0125] Thus, those skilled in the art will recognize that some aspects of the apparatus and methods described above (e.g., calculations performed by a processor) may be embodied as, for example, processor control code located on a non-volatile carrier medium such as a disk, CD-ROM or DVD-ROM, a programmed memory such as a read-only memory (firmware) or on a data carrier such as an optical signal carrier or an electrical signal carrier. For many applications, embodiments of the present invention will be implemented on a DSP (digital signal processor), an ASIC (application specific integrated circuit), or an FPGA (field programmable gate array). Thus, the code may include conventional program code or microcode or, for example, code for setting up or controlling an ASIC or FPGA. The code may also include code for dynamically configuring a reconfigurable apparatus such as a reprogrammable logic gate array. Similarly, the code may include code for a hardware description language such as Verilog TM or VHDL (very high speed integrated circuit hardware description language). As will be appreciated by those skilled in the art, the code may be distributed between multiple coupled components that communicate with each other. Where appropriate, the embodiments may also be implemented using code that runs on a field (re)programmable analog array or similar device to configure analog hardware.

[0126] Embodiments of the present invention may be arranged as part of an audio processing circuit, such as an audio circuit that may be provided in a host device.Circuitry according to embodiments of the present invention may be implemented as an integrated circuit.

[0127] Embodiments may be implemented in a host device, particularly a portable and / or battery-powered host device (e.g., a mobile phone, an audio player, a video player, a PDA, a mobile computing platform such as a laptop or tablet computer, and / or, for example, a gaming device). Embodiments of the invention may also be implemented in whole or in part as an accessory attachable to a host device, such as an active speaker or headset, etc. Embodiments may be implemented in other forms of devices, such as remote controller devices, toys, machines (such as robots), home automation controllers, etc.

[0128] It should be noted that the above-mentioned embodiments illustrate rather than limit the present invention, and those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims. The use of "a" or "an" herein does not exclude a plurality, and a single feature or other unit may perform the functions of several units recited in the claims. Any reference signs in the claims shall not be construed as limiting the scope of the claims.

Claims

1. A signal processing device for on-ear detection of a headphone, the device comprising: a probe signal generator configured to generate a probe signal for acoustic playback from a speaker; an input for receiving a microphone signal from a microphone located within the headset, the microphone signal comprising at least a portion of the detection signal received at the microphone; as well as a processor configured to apply a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, the processor further configured to process the estimate of the at least one parameter to determine whether the headphone is on the ear; Wherein the processor is configured to dynamically reduce the level of the detection signal upon detecting an increase in the level of the microphone signal.

2. The device of claim 1, wherein the processor is configured to process the estimate of the at least one parameter to determine whether the headphone is on the ear by comparing the estimated parameter to a threshold.

3. An apparatus according to claim 1 or claim 2, wherein the at least one parameter is the amplitude of the detection signal.

4. The device of claim 3, wherein when the amplitude is above a threshold, the processor is configured to indicate that the headphone is on the ear.

5. The apparatus of claim 1, wherein the detection signal comprises a single tone.

6. The apparatus of claim 1, wherein the probe signal comprises a weighted multi-tone signal.

7. The apparatus of claim 1, wherein the detection signal is limited to an inaudible frequency range.

8. The apparatus of claim 1, wherein the detection signal is limited to a frequency range less than a threshold frequency below a typical human hearing range.

9. The apparatus of claim 1, wherein the detection signal varies with time.

10. The apparatus of claim 9, wherein the detection signal varies in response to a change in the level of ambient noise within a frequency range of the detection signal. 11 . The apparatus according to claim 1 , further comprising a down-converter configured to down-convert the microphone signal before the state estimation to reduce a computational burden required for the state estimation.

12. The apparatus of claim 1, wherein the processor is configured to implement a Kalman filter to achieve the state estimation.

13. The apparatus of claim 11, wherein a copy of the detection signal generated by the detection signal generator is transmitted to a prediction module of a Kalman filter.

14. The device according to claim 1 comprises a determination device module, wherein the determination device module is configured to generate a first probability that the headphone is on the ear and a second probability that the headphone is off the ear from the at least one parameter, and wherein the processor is configured to use the first probability and / or the second probability to determine whether the headphone is on the ear.

15. The apparatus of claim 14, wherein the decision device module compares the at least one parameter to an upper threshold level to determine the first probability.

16. The apparatus of claim 15 , wherein the state estimation produces a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, the first probability is calculated as N ON / N, where N ON is the number of samples in the frame where the at least one parameter exceeds the upper threshold.

17. The apparatus of claim 14, wherein the decision device module compares the at least one parameter to a lower threshold level to determine the second probability.

18. The apparatus of claim 17, wherein the state estimation produces a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headphone is on the ear, each frame comprising N estimates, and wherein for each frame, the second probability is calculated as N OFF / N, where N OFF is the number of samples in the frame in which the at least one parameter is less than the lower threshold.

19. The device of claim 14, wherein the determination device module is configured to generate an uncertainty probability from the at least one parameter, the uncertainty probability reflecting uncertainty about whether the headphone is on the ear or off the ear, and wherein the processor is configured to use the uncertainty probability to determine whether the headphone is on the ear.

20. The apparatus of claim 19, wherein the state estimation produces a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, the uncertainty probability is calculated as N UNC / N, where N UNC is the number of samples in the frame in which the at least one parameter is greater than the lower threshold and less than the upper threshold.

21. A device according to claim 19 or claim 20, wherein the processor is configured to not change a previous determination of whether the headset is on the ear when the uncertainty probability exceeds an uncertainty threshold.

22. The apparatus of claim 1, wherein a change in the determination of whether the headphone is on the ear has a first decision delay from off-ear to on-ear, and has a second decision delay from on-ear to off-ear, the first decision delay being less than the second decision delay, thereby biasing the determination toward an on-ear determination.

23. The apparatus of claim 22, further comprising an input for receiving a microphone signal from a reference microphone of the headphone that captures external ambient sounds, and wherein the processor is further configured to apply a state estimation to the reference microphone signal to produce a second estimate of the at least one parameter of the detection signal, and wherein the processor is further configured to compare the second estimate with the estimate to distinguish ambient noise from supra-auricular occlusion.

24. A method for on-ear detection of a headphone, the method comprising: generating a probe signal for acoustic playback from a speaker; receiving a microphone signal from a microphone located within the headset, the microphone signal comprising at least a portion of the detection signal received at the microphone; applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the detection signal contained in the microphone signal, and determining whether the headset is on the ear based on an estimate of the at least one parameter; Wherein the level of the detection signal is dynamically reduced once an increase in the level of the microphone signal is detected.

25. The method of claim 24, wherein determining whether the headset is on the ear comprises comparing the estimated parameter to a threshold.

26. A method according to claim 24 or claim 25, wherein the at least one parameter is the amplitude of the detection signal.

27. The method of claim 26, comprising indicating that the headset is on the ear when the amplitude is above a threshold.

28. The method of claim 24, wherein the probing signal comprises a single tone.

29. The method of claim 24, wherein the probe signal comprises a weighted multi-tone signal.

30. The method of claim 24, wherein the detection signal is limited to an inaudible frequency range.

31. The method of claim 24, wherein the detection signal is limited to a frequency range less than a threshold frequency below a typical human hearing range.

32. The method of claim 24, wherein the detection signal varies with time.

33. The method of claim 32, wherein the detection signal varies in response to changes in the level of ambient noise within a frequency range of the detection signal.

34. The method of claim 24, further comprising down-converting the microphone signal prior to the state estimation to reduce the computational burden required for the state estimation.

35. The method of claim 24, wherein application state estimation is implemented via a Kalman filter.

36. The method of claim 34, wherein a copy of the detection signal is transmitted to a prediction module of a Kalman filter.

37. The method of claim 24, comprising generating, from the at least one parameter, a first probability that the headphone is on the ear and a second probability that the headphone is off the ear, and using the first probability and / or the second probability to determine whether the headphone is on the ear.

38. The method of claim 37, comprising comparing the at least one parameter to an upper threshold level to determine the first probability.

39. The method of claim 38, wherein a sample-by-sample estimate of the at least one parameter is generated, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, a first probability is calculated as N ON / N, where N ON is the number of samples in the frame where the at least one parameter exceeds the upper threshold.

40. The method of claim 37, further comprising comparing the at least one parameter to a lower threshold level to determine the second probability.

41. The method of claim 40, wherein a sample-by-sample estimate of the at least one parameter is generated, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, the second probability is calculated as N OFF / N, where N OFF is the number of samples in the frame where the at least one parameter is less than the lower threshold.

42. The method of claim 37, further comprising generating an uncertainty probability from the at least one parameter, the uncertainty probability reflecting uncertainty as to whether the headphone is on the ear or off the ear, and using the uncertainty probability to determine whether the headphone is on the ear.

43. The method of claim 42, wherein the state estimation produces a sample-by-sample estimate of the at least one parameter, and wherein the estimate is considered on a frame basis to determine whether the headset is on the ear, each frame comprising N estimates, and wherein for each frame, the uncertainty probability is calculated as N UNC / N, where N UNC is the number of samples in the frame in which the at least one parameter is greater than the lower threshold and less than the upper threshold.

44. A method according to claim 42 or claim 43, wherein when the uncertainty probability exceeds an uncertainty threshold, a previous determination of whether the headphone is on the ear is not changed.

45. The method of claim 24, wherein a change is made in determining whether the headphone is on the ear having a first decision delay from off-ear to on-ear, and having a second decision delay from on-ear to off-ear, the first decision delay being less than the second decision delay, thereby biasing the determination toward an on-ear determination.

46. ​​The method of claim 45, further comprising receiving a microphone signal from a reference microphone that captures external ambient sounds, and further comprising applying state estimation to the reference microphone signal to produce a second estimate of the at least one parameter of the detection signal, and comparing the second estimate with the estimate to distinguish between ambient noise and supra-auricular occlusion.

47. A non-transitory computer readable medium for on-ear detection of a headset, the non-transitory computer readable medium comprising instructions that, when executed by one or more processors, result in the following operations: generating a probe signal for acoustic playback from a speaker; receiving a microphone signal from a microphone located within the headset, the microphone signal comprising at least a portion of the detection signal received at the microphone; applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, and determining whether the headset is on the ear based on an estimate of the at least one parameter; The level of the detection signal is dynamically reduced upon detecting an increase in the level of the microphone signal.

48. The non-transitory computer readable medium of claim 47, further configured to perform the method of any one of claims 25 to 46.

49. A system for on-ear detection of a headphone, the system comprising a processor and a memory, the memory containing instructions executable by the processor, and wherein the system is operable to: generating a probe signal for acoustic playback from a speaker; receiving a microphone signal from a microphone located within the headset, the microphone signal comprising at least a portion of the detection signal received at the microphone; applying a state estimation to the microphone signal to produce an estimate of at least one parameter of the portion of the probe signal contained in the microphone signal, and determining whether the headset is on the ear based on an estimate of the at least one parameter; Wherein the level of the detection signal is dynamically reduced once an increase in the level of the microphone signal is detected.

50. The system of claim 49, wherein the system is a headset.

51. The system of claim 50, wherein the headset is an earbud.

52. The system of claim 49, wherein the system is a smartphone.

53. A system according to any one of claims 50-51, wherein an error microphone is mounted on the headphone so that when the headphone is worn, the error microphone senses sound occurring in the space between the headphone and the user's eardrum.

54. The system of claim 50, wherein a reference microphone is mounted on the headphone such that when the headphone is worn, the reference microphone senses sounds occurring outside of the headphone.

Images