Method, device, earphone and computer program for active interference noise suppression

Abstract

In the method for active interference noise suppression according to the present application, a transfer function of a secondary path between a loudspeaker and an error microphone is measured (20). Based on the measured transfer function of the secondary path, a transfer function of a primary path between a reference microphone and the error microphone is estimated (21). Then, based on the estimated transfer function of the primary path, filter coefficients for a filtering for generating a cancellation signal are determined (22).

Description

Technical Field

[0001] This invention relates to a method for suppressing active interference noise. It also relates to an apparatus for performing the method. Furthermore, the invention relates to an earphone configured to perform the method according to the invention or having an apparatus according to the invention, and a computer program having instructions that cause a computer to perform the steps of the method. Background Technology

[0002] High levels of noise pollution can cause stress and even serious mental and physical illnesses. High levels of noise pollution are caused by things like airplanes, trains, or cars, and are perceived as ambient noise by people both inside and outside these vehicles. For this reason, methods for active noise cancellation (ANC) are known, which reduce this disruptive ambient noise, a key feature of headphones or so-called audible devices.

[0003] Here, an additional sound signal is artificially generated that corresponds as precisely as possible to the sound signal of the interfering sound, but with opposite polarity, so that the interfering noise can be eliminated as much as possible by superimposing the two sound signals using destructive interference. For this purpose, in the case of headphones with active noise suppression, ambient noise is measured using one or more microphones integrated into the headphones, and then the portion remaining at the ear is calculated using the acoustic transfer function of the headphones. For this portion, an inverse polarity signal is then generated in the headphones to compensate and output via a speaker, through which the useful sound is also reproduced. Modern ANC headphones typically use fixed forward and feedback filters for this purpose, achieving attenuation of up to 30 dB at low frequencies, but the filter performance is sensitive to the corresponding wearing position of the headphones and the user's corresponding ear shape. In principle, adaptive algorithms could also be considered to improve the degree of noise suppression. However, such adaptive algorithms require high computational power and are therefore currently not suitable for headphones, hearing devices, or hearing aids.

[0004] Most commercially available ANC headphones come with a built-in speaker and two microphones. One of these microphones is oriented towards the headphone's environment to measure a reference signal in the form of ambient noise, and is often referred to as the reference microphone. The other microphone is oriented towards the user's ear canal or eardrum to determine the internal error signal and is also called the error microphone. The acoustic transmission from the external reference microphone to the internal error microphone is called the primary path; the transmission from the speaker to the error microphone is called the secondary path.

[0005] Measurements of these primary and secondary paths can be customized, significantly improving the performance and robustness of the ANC system. Secondary paths can be measured using a speaker and an internal microphone, where the signal-to-noise ratio is quite high due to the passive isolation of the headphones. Conversely, measuring the primary path requires additional external speaker setup and a suitable measurement environment, and is therefore complex and not easily performed by the end user. Summary of the Invention

[0006] In this context, the technical problem to be solved by the present invention is to provide an improved method and an improved device for suppressing active interference noise, particularly for suppressing interfering environmental noise in headphones, as well as a corresponding headphone and a computer program for performing the method.

[0007] The aforementioned technical problem is solved by a method having the features of claim 1, a corresponding device according to claim 8, a corresponding headset according to claim 10, and a computer program according to claim 11. Preferred embodiments of the invention are the subject of the dependent claims.

[0008] The present invention utilizes the understanding that, particularly in the case of in-ear headphones, but also in the case of headphones with other structural forms, there can be a significant correlation between the spectra of the primary path and the secondary path, and this correlation can be used to optimize interference noise suppression without measuring the primary path.

[0009] According to the inventors' knowledge, in the method for active interference noise suppression according to the present invention, the transfer function of the secondary path between the loudspeaker and the error microphone is measured. Based on the measured transfer function of the secondary path, the transfer function of the primary path between the reference microphone and the error microphone is estimated. Then, based on the estimated transfer function of the primary path, the filter coefficients for generating the canceled signal are determined.

[0010] Specifically, here, at least one reference microphone acquires the interfering sound signal, the speaker outputs a cancellation signal, and an error microphone acquires the remaining signal after the cancellation signal and the interfering sound signal are superimposed.

[0011] According to an embodiment of the invention, active interference noise suppression is performed when reproducing a useful audio signal using headphones, wherein one or more reference microphones are located on the outside of the headphones and an error microphone is located on the inside of the headphones.

[0012] Preferably, the transfer function of the secondary path is measured for the user in a personalized manner, and the transfer function of the primary path is estimated based on the transfer function of the secondary path measured for the user in a personalized manner.

[0013] Advantageously, filtering can be performed here using a forward FIR filter or an IIR filter.

[0014] According to another embodiment of the present invention, the estimated function of the primary path is determined by measuring and analyzing the transfer function of the secondary path and the transfer function of the primary path respectively during training for different people and / or headphones.

[0015] The advantage here is:

[0016] - For measurements within the frequency range of the transfer function, where there are deterministic changes for the primary and secondary paths, principal component analysis is performed, followed by dimensionality reduction of the measurements obtained during training;

[0017] - Based on the principal components and means already determined through principal component analysis, determine the complex gain vectors for the primary and secondary paths; and

[0018] - Determine a linear mapping that minimizes the error between the determined and estimated gain vectors of the primary path.

[0019] Accordingly, the device for active interference noise suppression according to the present invention includes:

[0020] - At least one reference microphone;

[0021] -speaker;

[0022] - Error microphone;

[0023] - Digital filters are used to generate canceled signals;

[0024] - A digital signal processor, which is configured to be used for:

[0025] - Generate a measurement signal that can be output through a speaker, and analyze the signal acquired using an error microphone to measure the transfer function of the secondary path between the speaker and the error microphone.

[0026] - Estimate the transfer function of the primary path between the reference microphone and the error microphone based on the measured transfer function of the secondary path; and

[0027] - Adjust the filter coefficients of the digital filter based on the estimated transfer function of the primary path.

[0028] According to one embodiment of the invention, the digital filter is designed herein as an FIR filter or an IIR filter.

[0029] The present invention also relates to an earphone designed for performing the method according to the invention or having a device according to the invention, and a computer program having instructions that cause a computer to perform the steps of the method according to the invention. Attached Figure Description

[0030] Other features of the invention will become apparent from the following description and claims, taken in conjunction with the accompanying drawings.

[0031] Figure 1 The illustration schematically shows an in-ear headphone with a primary acoustic path and a secondary acoustic path;

[0032] Figure 2 A flowchart of a method for active interference noise suppression according to the present invention is shown;

[0033] Figure 3 A block diagram of the headphones according to the present invention is shown;

[0034] Figure 4 The spectra of the measured primary path (a) and secondary path (b) are shown;

[0035] Figure 5 The following are shown: a) the spectrum based on personalized secondary path and averaged primary path measurements, and b) the spectrum of the active transfer function from the reference microphone to the error microphone based on personalized secondary path and separately estimated primary path measurements.

[0036] Figure 6 The median of the primary path |P(z)| and spectrum |H(z)| for different primary path estimates is shown;

[0037] Figure 7 Box plots of energy ratios for different primary path estimates are shown; and

[0038] Figure 8 The illustration shows the headphones being used in conjunction with an external computing device. Detailed Implementation

[0039] To better understand the principles of the present invention, embodiments of the invention are explained in more detail below with reference to the accompanying drawings. It should be understood that the present invention is not limited to these embodiments, and the described features can be combined or modified without departing from the scope of protection of the invention as defined in the claims.

[0040] The method according to the invention can be used, in particular, for active interference noise suppression in the case of in-ear headphones, such as... Figure 1As schematically shown. Here, the in-ear headphones 10 are positioned at the user's ear, with the ear insert 14 of the in-ear headphones inserted into the external auditory canal 15 to hold them in place. Depending on the individual wearing position in the ear canal, external interference noise can be partially blocked by the ear insert, so that this interference noise subsequently reaches the user's eardrum 16 only at a reduced level.

[0041] Interference sound signals x(t) reaching the headphones from the surrounding environment are acquired using a reference microphone 11 positioned away from the ear canal. Furthermore, the in-ear headphones 10 have an error microphone 12 pointing towards the ear canal 15 and a speaker 13 located near the error microphone 12. A cancellation signal can be output using the speaker 13. Error microphone 12 in signal cancellation The residual signal e(t) is acquired after superimposing the interfering sound signal x(t). Primary acoustic path P a (s) describes the transfer function from the reference microphone 11 to the error microphone 12, while the acoustic secondary path S a (s) describes the transfer function from speaker 13 to error microphone 12. The in-ear headphones shown have only one reference microphone, but multiple reference microphones can also be used, each with its own primary path.

[0042] Figure 2 The schematic illustration shows a basic scheme for active interference noise suppression, which can be implemented, for example, in such in-ear headphones. Here, in a first step 20, the transfer function of the secondary path between the speaker and the error microphone is measured. In a subsequent step 21, the transfer function of the primary path between the reference microphone and the error microphone is then estimated based on the measured transfer function of the secondary path. For this purpose, the relationship between the primary and secondary paths, determined during the training phase (described below), is used in the current headphones. The estimated transfer function then allows the filter coefficients for the filter used to generate the cancellation signal to be determined in a further step 22. In this way, the filter can be adjusted so that the output cancellation signal can achieve the best possible compensation for the interference signal. After determining the filter coefficients by means of the measurement of the secondary path and the subsequent estimation of the primary path, the filter can then be used unchanged until further notice to prevent or at least reduce the impairment of the user's perception due to interference noise when reproducing the useful audio signal with the in-ear headphones. Similarly, for example, when a user is traveling by train or plane and thus lowers the volume level, interference noise suppression can be perceived as more comfortable by the user, even without reproducing the useful audio signal.

[0043] Figure 3A block diagram of a device according to the present invention is shown. It includes components according to... Figure 1 The analog unit 30 of the hardware components is extended via an electronic back-end, which is connected to microphones 11 and 12 via analog-to-digital converters 31 and 32 and to speaker 13 via digital-to-analog converter 33. The electronic back-end includes a digital filtering unit 34 and a processor unit 35.

[0044] The device according to the invention can be fully integrated into an ANC headset, or it can be partly a component of an external device, such as a smartphone. Therefore, for example, the processor unit 35 can be part of such an external device.

[0045] Here, processor unit 35 has one or more digital signal processors, but may also contain different types of processors or combinations thereof. Digital filter 34 is designed as a time-invariant FIR feedforward filter. It receives the digitally converted interference signal x(n) and generates a cancellation signal. However, similarly, digital filter 34 can also be designed as an IIR filter, typically a dual second-order filter. Digital signal processor 35 generates the measurement signal m(n) and evaluates the digitized error signal e(n) to measure the secondary path. Furthermore, the digital filter is adjusted by the digital signal processor. The filter coefficients. For this purpose, instructions are stored in a memory (preferably integrated into a processor unit), which, when executed by the processor unit, cause the device to perform the steps according to the method according to the invention.

[0046] The overall transfer function H(s) describes the transfer function from the reference microphone 11 to the error microphone 12, and includes the effects of the ANC system relative to the primary path. The primary path P(z) and the secondary path S(z) include the effects of the analog-to-digital converter and digital-to-analog converter, the speaker, and the microphone.

[0047] Then the total transmission path is defined as

[0048]

[0049] Here, s and z represent the complex frequency parameters of the Laplace transform or z-transform, and n represents the discrete-time marker.

[0050] The following section first derives how to select the FIR feedforward filter based on the personalized measurements of the secondary path. The filter quotient is then introduced. Subsequently, an estimator for the primary path is introduced, which is trained based on a series of previously measured primary and secondary paths. Individual measurements of the secondary paths can then be fed into the estimator after the training phase to estimate individual primary paths.

[0051] This is the set of measured impulse responses of length L. The optimal FIR feedforward filter. To minimize the average energy of the total transmission path, as defined by the following cost function:

[0052]

[0053] It has a primary path vector expanded by 0. The convolution matrix s of the secondary path j .

[0054] Optimal FIR forward filter with respect to average value As given below

[0055]

[0056] However, in order to optimize the FIR feedforward filter in a personalized way An accurate understanding of the corresponding primary and secondary paths is required.

[0057] As mentioned earlier, personalized secondary paths can be measured using the headphones' speakers and internally placed error microphones. Then, if for all the s in the above equation... j Replace the personalized secondary path, and The average value of intermediate and primary paths, i.e.

[0058]

[0059] Used as an estimate of p, the optimal filter for a given personalized secondary path is obtained as follows:

[0060]

[0061] Because both the primary and secondary paths depend on the headphone wearing position and the physiological characteristics of the user's ear, this correlation can be used to apply the estimator to a personalized primary path based on the characteristics of the measured personalized secondary path. For this purpose, a window function Q is utilized in the z-domain. p (z) and Q s (z) Extract the frequency range of the transfer function affected by deterministic changes.

[0062] Principal component analysis (PCA) is used to analyze the set of... The first Kth complex frequency domain vector set from which the primary and secondary paths are extracted. p K sPrincipal component U p,k , And mean.

[0063] The gain vector g of complex numbers p,j and g s,j The frequency domain vectors and principal components based on the primary and secondary paths are used to minimize the Euclidean distance between the reconstructed frequency domain vectors. Then, a linear mapping is used. This linear mapping will transform the gain vector g of the primary path. p,j Gain vector g projected onto the secondary path s,j superior.

[0064] After measuring the personalized secondary paths, the window function Q in the z-domain is... s (z) is applied to the measured secondary path, and then the gain vector g of the secondary path is calculated using the principal components and mean of the secondary path. s,j Next, we first use linear mapping... Estimate the gain vector g of the primary path p,j Subsequently, based on the principal components and mean of the primary path and the estimated gain vector g of the primary path, p,j This is used to estimate the primary path. Finally, the estimation of the single primary path is used to replace... To obtain a personalized feedforward filter.

[0065] The effectiveness of the proposed estimator was verified through simulation, and the results are shown below. For this purpose, measurements were performed on 25 subjects and different engagement patterns at the in-ear headphone location, using a sampling rate of 48 kHz. The sets of measured primary and secondary paths are shown below. This includes a total of J = 173 pairs of impulse responses.

[0066] Figure 4 The measured spectra of the primary path (a) and secondary path (b) are shown. The shaded frequency range 40 illustrates the region of the selected frequency range window. The lengths of the primary and secondary paths are chosen to be L = 1024, and the length of the feedforward filter is L. w =64. The sets of primary and secondary paths measured were randomly divided into two subsets, one for training (80%) and one for validation (the remaining 20%). The training set was used to train the estimator as described above. Furthermore, K was chosen for the number of principal components. p =1 and K s =3. Then check the total transmission path. To verify the performance of the estimator, the measurement was repeated 100 times for a randomly divided subset.

[0067] Figure 5The amplitude spectrum |H(z)| measured here is shown, where the filter design in a) is based on the personalized secondary path and the averaged primary path, and in b) it is based on the personalized secondary path and the separately estimated primary path. Here, in addition to the median 50, 50th percentile 52, and 90th percentile 53 of |H(z)|, the median 51 of the primary path |P(z)| is also given to show the passive attenuation of the headphones.

[0068] Figure 6 The median of the primary path |P(z)| and spectrum |H(z)| for different primary path estimates is shown. Here, H... avg (z) Based on the mean of the primary paths in the training set, H est (z) Based on primary path estimation, H ppg The same applies to (z), where, however, a perfect PCA gain vector (PPG)g is used. p To replace its estimate, finally H opt (z) Based on the actual primary path. Shaded area 60 (where: |Q) p The plot (z)|>0 marks the frequency range in which H(z) is affected by the primary path estimator. As can be seen from the figure, the median of |H(z)| in the spectrum between 250 Hz and 2.5 kHz is reduced by up to 7 dB, approaching the median based on the personalized primary path.

[0069] Figure 7 The boxplot in the image correspondingly shows the data from... Figure 6 The energy ratios estimated in dB for different primary paths are: (a) mean, (b) estimate, (c) estimate using PPG, and (d) optimal value given the actual primary path. Here, Q is used... p In the case of (z), the energy ratio ε of the total transmission path and the primary path of the window is defined as

[0070]

[0071] For different primary path estimates, the median, as well as the minimum (so-called lower whisker) and maximum (so-called upper whisker), are shown as horizontal lines, and the lower and upper quartiles are shown as rectangles enclosing the median.

[0072] As can be seen from the figure, when using the median estimator (b), the energy ratio ε is reduced by 3.1 dB compared to using the mean (a), while the difference between the maximum values, the so-called upper whisker, is 5.0 dB.

[0073] Figure 8The illustration schematically shows the use of headphones 10 in conjunction with an external computer device 80, such as a so-called audio device. The external computer device 80 can, in particular, be a mobile terminal device suitable for audio reproduction. For example, a smartphone, a so-called wearable device such as a smartwatch, fitness tracker, or digital glasses, or a computer tablet can be connected to the headphones.

[0074] The device communicates wirelessly via a radio connection, such as Bluetooth. After a connection is established, audio signals can be transmitted from an external computing device 80 to headphones 10, and then reproduced in a conventional manner using one or more speakers integrated into the headphones.

[0075] Additionally, active interference noise suppression according to the invention can also be performed using an external computer device 80. For this purpose, the external computer device 80 can transmit a measurement signal to the headphones, particularly when the headphones 10 are first used by the user, and this measurement signal is then output through a speaker integrated in the headphones. An error microphone integrated in the headphones 10 then acquires an error signal, which is transmitted to the external computer device 80. Based on this, the external computer device 80 calculates the secondary path, estimates the primary path, and then determines the filter coefficients for the filter used to generate the canceled signal. The filter coefficients are then transmitted from the external computer device 80 to the headphones 10 via a wireless connection, wherein the filter is adjusted accordingly so that interference noise is suppressed to the greatest extent possible during the reproduction of the audio signal.

[0076] This invention can be used for active interference noise suppression in any field of audio reproduction technology.

Claims

1. A method for suppressing active interference noise, wherein, - Measure the transfer function of the secondary path between the loudspeaker and the error microphone; - Based on the measured transfer function of the secondary path, the transfer function of the primary path between the reference microphone and the error microphone is estimated using an estimator of the primary path, wherein the estimator of the primary path is determined by measuring and analyzing the transfer functions of the secondary path and the primary path in advance during training. and - Determine the filter coefficients for generating the canceled signal based on the estimated transfer function of the primary path.

2. The method according to claim 1, wherein, At least one reference microphone (11) acquires the interfering sound signal, the speaker (13) outputs the cancellation signal, and the error microphone (12) acquires the remaining signal after the cancellation signal and the interfering sound signal are superimposed.

3. The method according to claim 2, wherein, Active interference noise suppression is performed by means of headphones (10) when reproducing useful audio signals, and one or more reference microphones (11) are located outside the headphones and the error microphone (12) is located inside the headphones.

4. The method according to claim 1, wherein - A transfer function for measuring secondary paths in a personalized way for each user; - Estimate the transfer function of the personalized primary path based on the transfer function of the secondary path measured individually for the user.

5. The method according to claim 1, wherein, The filtering is performed using a forward FIR filter or an IIR filter.

6. The method according to claim 1, wherein, The estimator for the primary path is determined by measuring and analyzing the transfer functions of the secondary path and the primary path respectively during pre-training for different people and / or headphones.

7. The method according to claim 6, wherein, - For measurements within the frequency range of the transfer function, where there are deterministic changes for the primary and secondary paths, principal component analysis is performed, followed by dimensionality reduction of the measurements obtained during the training process; - Based on the principal components and means already determined through principal component analysis, determine the complex gain vectors for the primary and secondary paths; and - Determine a linear mapping that minimizes the error between the determined and estimated gain vectors of the primary path.

8. An apparatus for active noise suppression, comprising: - At least one reference microphone (11); - Speaker (13); - Error microphone (12); - Digital filter (34) is used to generate the cancellation signal; - Digital signal processor (35), the digital signal processor being configured to: - Generate a measurement signal that can be output through the speaker, and analyze the signal acquired using the error microphone in order to measure the transfer function of the secondary path between the speaker and the error microphone; - Based on the measured transfer function of the secondary path, the transfer function of the primary path between the reference microphone and the error microphone is estimated using an estimator of the primary path, wherein the estimator of the primary path is determined by measuring and analyzing the transfer functions of the secondary path and the primary path respectively in advance during training. and - Adjust the filter coefficients of the digital filter based on the estimated transfer function of the primary path.

9. The device according to claim 8, wherein, The digital filter (34) is designed as a forward FIR filter or an IIR filter.

10. An earphone (10) configured to perform the method according to claim 1.

11. A computer program having instructions that cause a computer to perform the steps of the method according to claim 1.

Images