WO2009034536A2 - Détection d'activité audio - Google Patents

Détection d'activité audio Download PDF

Info

Publication number
WO2009034536A2
WO2009034536A2 PCT/IB2008/053666 IB2008053666W WO2009034536A2 WO 2009034536 A2 WO2009034536 A2 WO 2009034536A2 IB 2008053666 W IB2008053666 W IB 2008053666W WO 2009034536 A2 WO2009034536 A2 WO 2009034536A2
Authority
WO
WIPO (PCT)
Prior art keywords
signal
audio
hearing aid
audio activity
adaptive filter
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/IB2008/053666
Other languages
English (en)
Other versions
WO2009034536A3 (fr
WO2009034536A9 (fr
Inventor
Sriram Srinivasan
David A. C. M. Roovers
Cornelis P. Janse
Valery S. Kot
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Koninklijke Philips NV
Original Assignee
Koninklijke Philips Electronics NV
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Koninklijke Philips Electronics NV filed Critical Koninklijke Philips Electronics NV
Publication of WO2009034536A2 publication Critical patent/WO2009034536A2/fr
Publication of WO2009034536A3 publication Critical patent/WO2009034536A3/fr
Publication of WO2009034536A9 publication Critical patent/WO2009034536A9/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; ELECTRIC HEARING AIDS; PUBLIC ADDRESS SYSTEMS
    • H04R25/00—Electric hearing aids
    • H04R25/40—Arrangements for obtaining a desired directivity characteristic

Definitions

  • the invention relates to audio activity detection and in particular, but not exclusively, to audio activity detection for a hearing aid apparatus or hearing aid device arrangement.
  • Advanced processing of audio signals has become increasingly important in many areas including e.g. telecommunication, content distribution etc.
  • detection of audio activity is an important operation which for example may be used to control and adapt the operation of audio processing equipment to the current requirements and conditions.
  • audio activity detection can be used to e.g. switch of transmissions during periods of the no audio activity
  • conferencing applications the audio activity detection may be used to identify and optimize operation for the current speaker etc.
  • a particular important area for audio signal processing is in the field of hearing aids.
  • hearing aids have increasingly applied complex audio processing algorithms to provide an improved user experience and assistance to the user.
  • audio processing algorithms have been used to provide an improved signal to noise ratio between a desired sound source and an interfering sound source resulting in a clearer and more perceptible signal being provided to the user.
  • hearing aids have been developed which include more than one microphone with the audio signals of the microphones being dynamically combined to provide a directivity for the microphone arrangement.
  • Such directivity may be achieved by beamforming algorithms which in some cases may be adaptive such that they are dynamically directed towards a desired sound source.
  • noise canceling algorithms may be applied to reduce the interference caused by undesired sound sources and background noise.
  • audio detection is aimed at differentiating between the presence and absence of speech.
  • the interfering signal is in such applications considered to be stationary background noise.
  • audio detection is significantly more complex when the interfering signal is also speech, such as may often be the case in the context of hearing aids.
  • conventional energy based activity detectors typically only detects the strongest source that is active, which is not necessarily the desired sound source.
  • energy threshold based detections schemes tend to often fail in reverberant environments with significant echoes. Accordingly, such detection algorithms tend to be relatively inaccurate.
  • an improved audio detection would be advantageous and in particular an audio detection allowing increased flexibility, facilitated implementation, improved suitability for hearing aids, increased accuracy, improved differentiation between interfering and desired audio sources and/or improved performance would be advantageous.
  • the Invention seeks to preferably mitigate, alleviate or eliminate one or more of the above mentioned disadvantages singly or in any combination.
  • audio activity detection apparatus comprising: a first sound sensor having a substantially omni-directional sensitivity and providing a first signal; a second sound sensor having a directional sensitivity and providing a second signal; a first adaptive filter for filtering the first signal to generate a first filtered signal; a first adaptation unit for adapting the first adaptive filter to reduce a difference between the first filtered signal and the second signal; and a detection unit for detecting audio activity in response to at least one filter coefficient of the first adaptive filter .
  • the invention may provide improved audio detection in many scenarios and applications.
  • the invention may be suitable for implementation in a hearing aid.
  • the invention may be suitable for implementation with a small distance between the microphones of the first and second sound sensors.
  • the invention may in many scenarios provide an accurate audio detection in reverberant audio environments and/or with long distances between the apparatus and a desired sound source.
  • the invention may in many scenarios provide improved differentiation between a desired and/or interfering sound source.
  • the detected audio activity may specifically be an activity of a desired sound source, activity of one or more interfering sound sources and/or of activity of both a desired sound source and of one or more interfering sound sources.
  • the invention may provide improved differentiation between detection of audio activity from a desired or interfering sound source and detection of audio activity from both a desired and interfering sound source.
  • the invention may in many embodiments allow a decorrelated detection of the presence of a desired sound source and an interfering sound source.
  • the audio detection apparatus may in many embodiments be implemented with reduced complexity compared to many other algorithms providing equivalent performance.
  • the audio detection apparatus may in many examples reuse functionality already used for other purposes.
  • a beamforming algorithm of a hearing aid may be reused to provide the directional sensitivity of the second sound sensor.
  • the invention may efficiently exploit directional characteristics of the audio environment to provide improved audio detection.
  • the invention may in a hearing aid context exploit the realization of the inventors that a desired signal is often in front of the user whereas interferers are typically located in the rear-half plane behind the user and that an audio detection scheme may be based on this fact.
  • the first adaptation unit may adapt the adaptive filter in response to an error signal generated as the difference between the first filtered signal and the first signal.
  • the first adaptation unit may be arranged to adapt the adaptive filter such that the error signal is reduced towards zero and may seek to minimize the error signal.
  • the substantially omni-directional sensitivity may e.g. vary less than 5 dB as a function of the direction.
  • the directional sensitivity may be such that sensitivity in at least one angle interval is at least 10 dB lower than the directional sensitivity of at least one other angle interval.
  • the first adaptive filter is a multiple tap filter and the detection unit is arranged to detect the audio activity in response to a peak filter coefficient for the first adaptive filter.
  • the feature may in many cases allow an improved audio detection for signals which do not have a flat frequency spectrum such as e.g. for speech signals.
  • the peak filter coefficient may be the filter coefficient of the tap having the highest amplitude.
  • the audio activity detection may not be in response to a peak filter coefficient but may use another detection criterion such as e.g. a criterion based on a (weighted) average filter coefficient measure.
  • the first adaptive filter is a single tap filter.
  • sufficiently accurate audio detection may be achieved using only a single tap first adaptive filter thereby reducing complexity, resource requirements and/or cost.
  • the audio activity detection apparatus of claim 1 further comprises: a third sound sensor having a directional sensitivity in a different direction than the second sound sensor and providing a third signal; a second adaptive filter for filtering the first signal to generate a second filtered signal; a second adaptation unit for adapting the second adaptive filter to reduce a difference between the second filtered signal and the third signal; and wherein the detection unit is further arranged to detect the audio activity in response to filter coefficients of the second adaptive filter .
  • the feature may allow an improved differentiation between audio activity of a desired source and audio activity of an interfering source.
  • the feature may allow improved audio detection in a reverberant audio environment.
  • the feature may provide improved differentiation between detection of audio activity from a desired or interfering sound source and detection of audio activity from both a desired and interfering sound source.
  • the feature may efficiently exploit directional characteristics of the audio environment to provide improved audio detection.
  • the feature may in a hearing aid context exploit the realization of the inventors that a desired signal is often in front of the user whereas interferers are typically located in the rear-half plane behind the user.
  • the second adaptation unit may adapt the second adaptive filter in response to an error signal generated as the difference between the second filtered signal and the first signal.
  • the second adaptation unit may be arranged to adapt the second adaptive filter such that the error signal is reduced towards zero and may e.g. seek to minimize the error signal.
  • the detection unit is arranged to determine an activity of a desired sound source if the filter coefficients meet a first criterion and an activity of an interfering sound source if the filter coefficients meet a second criterion. This may allow an improved audio detection.
  • an improved operation may be achieved based on an improved differentiation between detection of different types of audio activity. For example, in a hearing aid context, an improved application of either noise canceling and/or beam steering may be achieved based on an improved detection of the origin of the detected audio activity thereby resulting in improved user experience.
  • the detection unit is arranged to determine a combined activity of a desired sound source and at least one interfering sound source if the filter coefficients meet a third criterion and/or an absence of an activity of either the desired sound source and the at least one interfering sound source if the filter coefficients meet a fourth criterion (which may be the same as the third criterion).
  • the first sound sensor comprises a first substantially omni-directional microphone for generating the first signal and wherein the second sound sensor comprises the first substantially omni-directional microphone and a second substantially omni-directional microphone and an audio beamforming unit arranged to generate a directional audio beam by combining signals from the first substantially omni-directional microphone and the second substantially omnidirectional microphone into the second signal.
  • the feature may allow a reduced number of microphones and/or the use of simpler and cheaper omni-directional microphones.
  • a beamforming unit otherwise used for providing a desired audio processing may be reused by the audio activity detection apparatus.
  • the beamforming unit may specifically be arranged to generate a cardiod beam pattern.
  • the beam forming unit may in embodiments wherein a second sound sensor is employed be used to generate a different directional beampattern for the third sound sensor than for the first sound sensor.
  • the audio activity detection apparatus may use two substantially omni-directional microphones wherein the first sound sensor provides the signal from one of the microphones, the second sound sensor comprises the two microphones and the beamforming unit and provides the second signal as a beamformed output signal from the beamforming unit and the third sound sensor comprises the two microphones and the beamforming unit and provides the third signal as a different beamformed output signal from the beamforming unit
  • the audio activity detection apparatus is comprised in a hearing aid apparatus.
  • the invention may provide an improved hearing aid.
  • an improved audio detection may provide improved audio processing in the hearing aid resulting in improved performance and clarity of the signal provided to the user.
  • the invention may allow improved beamforming and/or noise canceling resulting from improved audio activity detection.
  • a hearing aid arrangement comprising two hearing aid devices, each of the two hearing aid devices comprising an audio activity detection apparatus in accordance with the abover; and wherein: a first hearing aid device of the two hearing aid devices comprises a transmitter for transmitting an audio activity detection indication to a second hearing aid device of the two hearing aid devices; and the second hearing aid device comprises: a receiver for receiving the audio activity detection indication from the first hearing aid device, and the second hearing aid device is arranged to detect the audio activity in response to the audio activity detection indication.
  • the feature may allow improved performance.
  • the feature may allow two hearing aids located at the different ears of the user to cooperate to provide improved audio detection. Specifically, the head shadowing effect on the signals received by hearing aids of each ear may be reduced, mitigated and/or compensated for.
  • the second hearing aid device may be arranged to detect audio activity for an interfering sound source if the at least one filter coefficient of the first adaptable filter of the second hearing aid device meets an activity criterion for an interfering sound source or if the audio activity detection from the first hearing aid device is indicative of an activity detection for an interfering sound source.
  • the second hearing aid device is arranged to detect audio activity for a desired sound source only if the at least one filter coefficient of the first adaptable filter of the second hearing aid device meets an activity criterion for a desired sound source and the audio activity detection from the first hearing aid device is indicative of an activity detection for a desired sound source.
  • the inventors have realized that it is typically best to adapt these processes only when the desired signal is in front of the user such that a clear signal is received.
  • the feature allows separation between desired signal received directly and a desired signal received from a different directions due to the head shadowing effect.
  • a method of audio activity detection comprising: receiving a first signal from a first sound sensor having a substantially omni-directional sensitivity; receiving a second signal from a second sound sensor having a directional sensitivity; filtering the second signal in a first adaptive filter to generate a first filtered signal; adapting the first adaptive filter to reduce a difference between the first filtered signal and the first signal; and detecting audio activity in response to at least one filter coefficient of the first adaptive filter .
  • FIG. 1 illustrates an example of an audio activity detection apparatus in accordance with some embodiments of the invention
  • FIG. 2 illustrates an example of a cardioid audio beam form
  • FIG. 3 illustrates an example of an audio activity detection apparatus in accordance with some embodiments of the invention.
  • FIG. 4-6 illustrates examples of impulse responses of an adaptive filter in accordance with some embodiments of the invention.
  • FIG. 1 illustrates an example of an audio activity detector for a hearing aid.
  • the illustrated functionality is implemented in a hearing aid to be worn by a user.
  • the audio activity detector generates an estimated activity indication of whether a desired sound source, such as a person speaking, is currently active.
  • the activity indication is in the specific example used to control other functionality of the hearing aid and in particular to control a beamforming and noise cancellation algorithm depending on whether the signal is a desired signal or an interfering signal.
  • the audio activity detector comprises a first and second microphone 101, 103 which in the example are omni-directional microphones mounted in an end- fire configuration (mounted along a line towards the front when worn by a user).
  • an omni-directional microphone is a microphone for which the sensitivity variation as a function of the angle between a sound source and a reference direction is less than a given value.
  • a microphone may for example be considered an omni-directional microphone if the difference between the highest and lowest signal level recorded for the same sound source at different angles is less than a threshold, which e.g. may be 1, 5 or 10 dB.
  • the first and second microphone 101, 103 are coupled to a beamforming unit (or processor) 105.
  • the beamforming unit 105 is arranged combine the signals recorded by the microphones to generate an output signal corresponding to a beamformed sensitivity pattern as will be known to the person skilled in the art.
  • the beamforming unit 105 is arranged to combine the audio signals from the first and second microphone 101, 103 to generate a directional sensitivity corresponding to a forward facing cardiod beam pattern. An example of such a beam pattern is illustrated in FIG. 2.
  • the first microphone 101 is furthermore coupled to a first adaptive filter 107 which filters the signal from the first microphone 101 to generate a first filtered signal.
  • the first adaptive filter 107 is a Finite Impulse Response (FIR) filter which filters the signal from the first microphone 101 in the time domain.
  • the first adaptive filter 107 may e.g. in some embodiments be a multi-tap FIR filter or may in other embodiments be a single tap FIR filter having only a single filter coefficient.
  • FIG. 1 will initially be described with reference to a multi-tap FIR filter comprising a large number of taps/filter- coefficients, such as e.g. 128 filter coefficients.
  • the first adaptive filter 107 and beamforming unit 105 are coupled to a difference element 109 which receives the first filtered signal and the beamform signal and generates a difference signal reflecting the difference between these.
  • the difference signal is simply generated by subtracting the samples of the first filtered signal from the samples of the beamform signal.
  • the difference element 109 is coupled to a first adaptation unit 111 which is further coupled to the first adaptive filter 107.
  • the first adaptation unit 111 receives the difference signal and uses this as an error signal to adapt the first adaptive filter 107.
  • the error signal is used to drive the adaptation of the filter coefficients of the first adaptive filter 107 by the first adaptation unit 111 such that the error signal is minimized.
  • the first adaptive filter 107 adapts the first adaptive filter to reduce the difference between the first filtered signal and the beamform signal.
  • the first adaptation unit 111 may be adapted using the normalized least mean square algorithm described in S. Haykin, "Adaptive Filter Theory", Ch. 6, 3 rd ed. Prentice Hall, 1995.
  • the first adaptive filter 107 is coupled to a detection unit 113 which is operable to generate an audio detection estimate in response to one or more filter coefficients of the first adaptive filter 107.
  • the detection unit 113 may at regular intervals receive a copy of the adapted filter coefficients of the first adaptive filter 107 and may set an audio activity detection indication accordingly. Thus, if the filter coefficients currently meet a given criterion, the detection unit 113 determines that (desired) audio activity has been detected and otherwise the detection unit 113 determines that (desired) audio activity has not been detected.
  • the detection unit 113 is arranged to find a peak in the impulse response of the first adaptive filter 107. The amplitude of the largest filter coefficient is determined and compared to a threshold. If the amplitude exceeds the threshold, the detection unit 113 generates an indication of a detection of an audio activity and otherwise it generates an indication of not having detected any (desired) audio activity.
  • the audio activity detector adapts an adaptive filter such that the filtered signal from an omni-directional microphone most closely resembles the signal received from a directional beam pattern which is aimed in the presumed direction of a wanted sound source (e.g. a speaker in front of the user of the hearing aid).
  • a wanted sound source e.g. a speaker in front of the user of the hearing aid.
  • the first adaptive filter 107 will be adapted to have a single filter coefficient with a unity amplitude (ignoring noise and assuming ideal scaling of each signal path etc).
  • the first adaptive filter 107 will be adapted to have filter coefficients with very low amplitude values as the directional beam associated with the beamform signal attenuates this signal to substantially zero.
  • the audio activity detector of FIG. 3 comprises the same elements as described with reference to the audio activity detector of FIG. 1.
  • the beamforming unit 105 is arranged to not only generate a first beamform signal for a forward facing beam pattern but is also arranged to generate a second beamform signal for a rear facing beam pattern.
  • the beamforming unit 105 is also arranged to combine the signals from the first and second microphones 101, 103 to generate a second beamform signal that corresponds to a directional sensitivity in another direction than for the first beamform signal.
  • the beamforming unit 105 generates a second cardiod beam pattern in the opposite direction of the beam pattern for the first beamform signal.
  • the audio activity detector also comprises a second adaptation filter 301 which like the first adaptive filter 107 receives the signal from the first microphone 101 and filters this to generate a second filtered signal.
  • the second adaptive filter 301 and beamforming unit 105 are coupled to a second difference element 303 which receives the second filtered signal and the second beamform signal and generates a second difference signal reflecting the difference between these.
  • the difference signal is simply generated by subtracting the samples of the second filtered signal from the samples of the second beamform signal.
  • the second difference element 303 is coupled to a second adaptation unit 305 which is further coupled to the second adaptive filter 301.
  • the second adaptation unit 305 receives the second difference signal and uses this as an error signal to adapt the second adaptive filter 301.
  • the second error signal is used to drive the adaptation of the filter coefficients of the second adaptive filter 301 by the second adaptation unit 305 such that the error signal is minimized.
  • the second adaptation unit 305 adapts the second adaptive filter 301 to reduce the difference between the second filtered signal and the second beamform signal.
  • the second adaptation unit 305 may apply the same algorithm as the first adaptation unit 111.
  • the second adaptive filter 301 is coupled to the detection unit 113 which is furthermore arranged to generate the estimated activity detection in response to filter coefficients of the second adaptive filter 301.
  • the detection unit 113 may perform an audio detection which not only considers the presence of wanted signals relative to a general (omni-directional) audio level but also considers the presence of dominant interfering signals in different directions relative to the general (omni-directional) audio level. This allows not only a more accurate detection of desired audio activity (and dominance of a desired signal) but also provides additional information in the form of an indication of dominance of interferers in other directions.
  • the active signal is assumed to be located around 0 degrees resulting in an the impulse response of the first adaptive filter 107 being dominated by a peak whose magnitude is close to unity as the cardioid beamforming of the beamforming unit 105 has a gain of unity in this direction. More generally, for sources in the front half plane, the magnitude of the peak is between 0.5 and 1 as can be seen from considering the cardioid polar pattern of Fig. 2.1. Specifically, the forward cardioid beam pattern provides a unit gain for a signal arriving from zero degrees, and a gain of around 0.5 for a signal arriving from 90 degrees.
  • the first adaptive filter 107 is adapted to filter the signal from the omni microphone such that the filtered signal corresponds to this signal resulting in an impulse peak between 0.5 and 1 depending on the direction to the sound source.
  • the first adaptive filter 107 is adapted to filter the signal from the omni microphone such that the filtered signal corresponds to this signal resulting in an impulse peak between 0.5 and 1 depending on the direction to the sound source.
  • the magnitude of the peak of the impulse response for the second adaptive filter 301 is between 0 and 0.5.
  • the magnitude of the peak of the impulse response of the second adaptive filter 301 is between 0.5 and 1
  • the magnitude of the peak of the impulse response of the first adaptive filter 107 is between 0 and 0.5.
  • FIG. 4-6 Examples of adapted impulse responses of the first adaptive filter 107 and second adaptive filter 301 for different scenarios are shown in FIG. 4-6. Specifically, FIG. 4 illustrates an example where there is a single dominant desired signal, FIG. 5 illustrates an example where there is a single dominant interferer and FIG. 6 illustrates an example where there is both a desired signal and interfering signal. In the example, each of the first adaptive filter 107 and the second adaptive filter 301 have 128 taps.
  • the detection unit 113 may for example determine an activity of a desired sound source if the filter coefficients of the first adaptive filter 107 and the second adaptive filter 301 meet a first criterion and an activity of an interfering sound source if the filter coefficients of the first adaptive filter 107 and the second adaptive filter 301 meet a second criterion.
  • the detections may specifically indicate that the respective signal is currently a dominant signal. If none of the criterions are met, the detection unit 113 may generate an indication reflecting that no signal is currently dominant. As a specific example, the detection unit 113 may first identify a peak filter coefficient for both the first adaptive filter 107 and the second adaptive filter 301.
  • the values of these coefficients may then be compared to suitable thresholds to determine whether audio activity has been detected for a desired source, from an interfering source, from both a desired and interfering source or from neither an interfering source nor a desired source.
  • the detection unit 113 can set the audio activity indication to reflect that a dominant signal has been detected from a desired source.
  • the hearing aid may proceed to e.g. control and adapt a beamforming unit to direct a beam towards the current audio source in order to improve the received signal quality.
  • the detection unit 113 can set the audio activity indication to reflect that a dominant interfering signal has been detected (from one or more interfering sources).
  • the hearing aid may proceed to e.g. control and adapt a noise canceling unit to provide improved suppression of the currently detected audio signal.
  • the detection unit 113 may set the audio detection indication to reflect that no dominant audio source has been detected. Accordingly, no strong and clear signal is present which can be used to efficiently adapt the beamforming or noise canceling and accordingly the hearing aid may continue to operate with the current settings.
  • the described approach may allow a hearing aid to provide improved adaptation of various algorithms depending on the current audio conditions.
  • algorithms may be updated when this is likely to result in the most accurate update and specifically when the audio environment is dominated by signals which the algorithm are directed at.
  • the first adaptive filter 107 is referred to as wf and the second adaptive filter 301 is referred to as wb.
  • this is changed to reflect the presence of a dominant signal only if a dominant signal has been detected for C consecutive segments, where C is a suitable design parameter that depends on the requirements and preferences of the individual embodiment.
  • the described audio activity detectors may be affected by head shadow in the rear quarter plane corresponding to the ear without the hearing aid.
  • the impulse response does not behave as expected due to head shadow effects (even if an interferer is a point source).
  • the shadow effect may result in the strongest signal for the interferer being received at a different angle which may e.g. fall within beam pattern of the forward cardioid (e.g. due to reflections and attenuation of the direct signal). This may in some scenarios result in a risk that the audio activity detector may detect an interfering signal as activity from a desired signal.
  • a hearing aid arrangement may be used which comprises two hearing aid devices for respectively a left and right ear of a user where each of the hearing aids comprise an audio activity detector as previously described.
  • At least one of the hearing aids furthermore comprises functionality for transmitting the audio activity detection indication generated by the detection unit 113 of the hearing aid to the other hearing aid.
  • This second hearing aid furthermore comprises functionality for receiving the audio activity detection indication from the first hearing aid. The second hearing aid may then proceed to determine a combined activity indication which reflects both the audio activity detection indication from the first hearing aid as well as the activity detection generated by the detection unit 113 of the second hearing aid.
  • the detection unit 113 of the second hearing aid can proceed to determine that an activity of a dominant desired source is only present if both hearing aids detect a dominant wanted signal.
  • the second hearing aid only indicates an audio activity for a desired sound source if at least one filter coefficient of the first adaptable filter of the second hearing aid device meets an activity criterion for a desired sound source (i.e. the audio activity detector of the second hearing aid detects the presence of a dominant desired signal) and the audio activity detection indication from the first hearing aid is also indicative of the audio activity of a desired sound source being detected.
  • the hearing aids may comprise BluetoothTM communication functionality for exchanging data.
  • the first and second hearing aids may both comprise the described functionality and that audio activity detection indications may be exchanged and processed in both directions.
  • both the first and second hearing aids may be arranged to only update a beamforming algorithm if the other hearing aid also indicates that a desired source has been detected.
  • Such a binaural scheme may specifically address, reduce or mitigate the problems resulting from the head shadow effect. For a source at the front, the head shadow effect will typically be negligible or non-existent and the individual audio activity detections of the two hearing aids operate reliably. Thus, both hearing aids will detect the desired audio activity.
  • At least one of the hearing aids will not be shadowed by the head and will therefore correctly detect an interferer even if the other hearing aid does not.
  • any interferer will typically only be in the head shadow for one of the hearing aids, at least one of the audio activity detections will be correct thereby avoiding an interferer incorrectly being detected as a desired signal by both hearing aids and thus by the arrangement as such.
  • the first adaptive filter 107 has only a single tap thereby reducing complexity.
  • a hearing aid with two omni-directional microphones spaced d meters apart in an end- fire configuration (oriented toward 0 degrees) is considered.
  • the origin O is defined to be the center of the end- fire array. It is assumed that the point sources under consideration are at ear-level and thus at an elevation of around zero degrees.
  • Many modern hearing aids include a directional processing unit that combines the two omni signals to produce a forward facing cardioid response and a backward facing cardioid response, corresponding to the processing by the beamforming unit 105.
  • the output of the forward cardioid can be given by: 1_ c Jn) ⁇ -(1 + cos ⁇ )s(n,Q ) ⁇ c ⁇ s(n,Q ),
  • the array at ⁇ deg., normalized to have unit response to a signal arriving from 0 degrees.
  • the first adaptive filter 107 of FIG. 1 corresponds to a single adaptive weight w between one of the omni-directional microphones 101, and the forward cardio id generated by the beamforming unit 105.
  • the weight w is adapted continuously, regardless of whether a desired or interfering source is active, to minimize the expected energy of the error signal defined by
  • the equation holds even in the presence of multiple interferers located within the region [90 + ⁇ ; ,270 - ⁇ ; ] .
  • a simple detection scheme can be aplied by the detection unit 113 to differentiate between a source in the front half plane and an interferer in the rear half plane.
  • a desired source is detected whenever w ⁇ c s and an interferer is detected whenever w ⁇ C 1 . It should be noted that the detections are decoupled such that an interferer is not automatically detected when the desired source is not active.
  • the method can also be applied in the frequency domain e.g. to obtain detections per frequency bin. This is particularly relevant for small microphone spacings where the cardioid response (1 + cos ⁇ )/2 can be assumed to be frequency- independent. In the following the effect of reverberation and ambient diffuse noise will be considered.
  • the microphone signal of the first microphone 101 can be written as
  • V 1 O) J;J;: O VO, ⁇ , ⁇ )J ⁇ J ⁇ ,
  • V 1 OO includes both the ambient noise and the reverberation.
  • the optimal adaptive weight w is given by:
  • the detection threshold for detecting a dominant desired source is considered. This is followed by a discussion on detecting a dominant interferer.
  • the two detections are decoupled, i.e., a dominant interferer is not automatically detected when there is no dominant desired source (e.g., both could be inactive or both equally dominant).
  • a dominant desired detecton threshold T s is desired such that w(j,f,c ⁇ ,c ⁇ ) > T s when the desired source is dominant.
  • Two approaches for determining T s are considered corresponding to either minimizing the false alarm rate or maximizing the detection rate.
  • the design parameters ⁇ s , ⁇ , , J , and y d ow can be set during the individual fitting session between the user and an audiologist, or automatically be set to different values based on the output of the environment classification algorithm that is present in modern hearing aids.
  • a dominant interferer detection threshold T 1 must be determined such that w(y,f,c ⁇ ,c ⁇ ) ⁇ T ⁇ when the interferer is dominant.
  • the procedure we adopt here is analogous to that adopted for the desired source detection, and we briefly state the relevant results.
  • the threshold T 1 determines the SIR y l ° w such that: Condition: 4 For all G 1 e [360 - ⁇ j S 0] u [0, ⁇ J , for all ⁇ 2 e [90 + ⁇ ! ,270 - ⁇ J , and for a given y , w(y,f,c ⁇ ,c ⁇ ) ⁇ T 1 whenever ⁇ ⁇ y] ow .
  • an embodiment of the invention may be physically, functionally and logically implemented in any suitable way. Indeed the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units. As such, the invention may be implemented in a single unit or may be physically and functionally distributed between different units and processors.

Landscapes

  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Neurosurgery (AREA)
  • Otolaryngology (AREA)
  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Acoustics & Sound (AREA)
  • Signal Processing (AREA)
  • Circuit For Audible Band Transducer (AREA)

Abstract

La présente invention concerne un appareil de détection d'activité audio qui comprend un premier capteur sonore (101) doté essentiellement d'une sensibilité omnidirectionnelle et fournissant un premier signal et un second capteur sonore (101, 103, 105) ayant une sensibilité directionnelle et offrant un second signal. Un premier filtre adaptatif (107) filtre le second signal pour générer un premier signal filtré et une première unité d'adaptation (111) adapte le premier filtre adaptif (107) afin de réduire une différence entre le premier signal filtré et le premier signal. Une unité de détection (113) détecte une activité audio en réponse à au moins un coefficient de filtre du premier filtre adaptatif (107). L'invention peut permettre la détection audio améliorée dans plusieurs cas et peut en particulier fournir une détection audio améliorée pour des aides auditives utilisant une pluralité de microphones. Des détections décorrélées d'émetteurs brouilleurs dominants et des signaux désirés peuvent être réalisés.
PCT/IB2008/053666 2007-09-14 2008-09-11 Détection d'activité audio Ceased WO2009034536A2 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP07116406.5 2007-09-14
EP07116406 2007-09-14

Publications (3)

Publication Number Publication Date
WO2009034536A2 true WO2009034536A2 (fr) 2009-03-19
WO2009034536A3 WO2009034536A3 (fr) 2009-08-20
WO2009034536A9 WO2009034536A9 (fr) 2009-11-05

Family

ID=40452645

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2008/053666 Ceased WO2009034536A2 (fr) 2007-09-14 2008-09-11 Détection d'activité audio

Country Status (1)

Country Link
WO (1) WO2009034536A2 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2242289A1 (fr) * 2009-04-01 2010-10-20 Starkey Laboratories, Inc. Système d'assistance auditive avec détection de sa propre voix
US7929713B2 (en) 2003-09-11 2011-04-19 Starkey Laboratories, Inc. External ear canal voice detection
WO2012098427A1 (fr) * 2011-01-18 2012-07-26 Nokia Corporation Appareil de sélection de scène audio
US9219964B2 (en) 2009-04-01 2015-12-22 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9552845B2 (en) 2009-10-09 2017-01-24 Dolby Laboratories Licensing Corporation Automatic generation of metadata for audio dominance effects

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1018854A1 (fr) * 1999-01-05 2000-07-12 Oticon A/S Procédé et dispositif pour l' amelioration de l' intelligibilité de la parole
CA2538021C (fr) * 2003-09-19 2011-11-22 Widex A/S Procede de commande de la directionalite de la caracteristique de reception de son d'une aide auditive, et aide auditive dans laquelle est applique ledit procede
DE102005032274B4 (de) * 2005-07-11 2007-05-10 Siemens Audiologische Technik Gmbh Hörvorrichtung und entsprechendes Verfahren zur Eigenstimmendetektion

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9369814B2 (en) 2003-09-11 2016-06-14 Starkey Laboratories, Inc. External ear canal voice detection
US9036833B2 (en) 2003-09-11 2015-05-19 Starkey Laboratories, Inc. External ear canal voice detection
US7929713B2 (en) 2003-09-11 2011-04-19 Starkey Laboratories, Inc. External ear canal voice detection
EP3169085A1 (fr) * 2009-04-01 2017-05-17 Starkey Laboratories, Inc. Système d'assistance auditive avec détection de sa propre voix
US10171922B2 (en) 2009-04-01 2019-01-01 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9094766B2 (en) 2009-04-01 2015-07-28 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US11388529B2 (en) 2009-04-01 2022-07-12 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9219964B2 (en) 2009-04-01 2015-12-22 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US10715931B2 (en) * 2009-04-01 2020-07-14 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
EP2242289A1 (fr) * 2009-04-01 2010-10-20 Starkey Laboratories, Inc. Système d'assistance auditive avec détection de sa propre voix
US10652672B2 (en) 2009-04-01 2020-05-12 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9699573B2 (en) 2009-04-01 2017-07-04 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9712926B2 (en) 2009-04-01 2017-07-18 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US8477973B2 (en) 2009-04-01 2013-07-02 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US10225668B2 (en) 2009-04-01 2019-03-05 Starkey Laboratories, Inc. Hearing assistance system with own voice detection
US9552845B2 (en) 2009-10-09 2017-01-24 Dolby Laboratories Licensing Corporation Automatic generation of metadata for audio dominance effects
WO2012098427A1 (fr) * 2011-01-18 2012-07-26 Nokia Corporation Appareil de sélection de scène audio
US9195740B2 (en) 2011-01-18 2015-11-24 Nokia Technologies Oy Audio scene selection apparatus

Also Published As

Publication number Publication date
WO2009034536A3 (fr) 2009-08-20
WO2009034536A9 (fr) 2009-11-05

Similar Documents

Publication Publication Date Title
KR102352928B1 (ko) 가변 마이크로폰 어레이 방향을 갖는 헤드셋들을 위한 듀얼 마이크로폰 음성 프로세싱
JP5249207B2 (ja) 適応性のある方向性信号処理を備えた補聴装置
WO2009034524A1 (fr) Appareil et procede de formation de faisceau audio
EP3407627B1 (fr) Système d'assistance auditive incorporant une personnalisation de microphone directionnel
US10079026B1 (en) Spatially-controlled noise reduction for headsets with variable microphone array orientation
US7983907B2 (en) Headset for separation of speech signals in a noisy environment
CN104254029B (zh) 一种具有麦克风的耳机、及改善耳机的音频灵敏度的方法
CA2475183C (fr) Methode permettant de reduire le couplage acoustique dans des systemes d'audioconference
CN110140359B (zh) 使用波束形成的音频捕获
Jeub et al. Blind estimation of the coherent-to-diffuse energy ratio from noisy speech signals
JP2009542057A5 (fr)
WO2011010292A1 (fr) Formation de faisceau audio
KR20080059147A (ko) 노이즈 환경에서 스피치 신호의 강건한 분리
WO2008041878A2 (fr) Système et procédé de communication libre au moyen d'une batterie de microphones
WO2013049737A1 (fr) Traitement de signaux
EP3545691A1 (fr) Capture sonore en champ lointain
WO2009034536A2 (fr) Détection d'activité audio
US20260006392A1 (en) Hearing system comprising a hearing aid and an external processing device
Marquardt et al. Performance comparison of bilateral and binaural MVDR-based noise reduction algorithms in the presence of DOA estimation errors
Braun et al. Directional interference suppression using a spatial relative transfer function feature
Srinivasan et al. Spatial audio activity detection for hearing aids
Alkaher et al. Adaptive Time-Domain ROI Beamforming with Acoustic-Feedback Suppression for Assistive Hearing Glasses

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 08807601

Country of ref document: EP

Kind code of ref document: A2

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 08807601

Country of ref document: EP

Kind code of ref document: A2