NOISE CANCELATION SYSTEMS WITH SIMULTANEOUS HARMONIC FILTERING AND PROCESSES

Abstract

A noise cancellation system (100) with harmonic filtering for a vehicle audio system, comprising: at least one first input sensor configured to receive a reference signal (x r1 [n], x r2 [n]); at least one second input sensor configured to transmit at least two narrowband input signals, wherein each of the input signals (x r1 [n], x r2 [n]) contains overtone noises; a processor (105) programmed to: Receiving the reference signal, wherein the reference signal comprises at least two narrowband reference signals (x r1 [n], x r2 [n]) includes, Receiving the narrowband input signals, Applying a gain reference control (346) to the reference signals to determine whether the frequencies of each of the reference signals (x r1 [n], x r2[n]) are within a predefined range of each other, Applying a bandpass filter (156) to one of the reference signals (x r1 [n], x r2 [n]), Applying a secondary path (376) based on the filtered reference signals (x r1 [n], x r2 [n]) to generate anti-noise signals, and Summing (378) the anti-noise signals and the input signals to produce an error signal (e m [n]) at the output sensor (145), Applying an adaptive filter (392) to the error signal (e m [n]) to remove overtone noise from the error signal (e m [n]) and Removing one of the reference signals (x r1 [n], x r2 [n]) in response to the frequencies of each of the reference signals (x r1 [n], x r2[n]) are within the predefined range of each other to prevent common overtone content from being detected on both reference signals (x r1 [n], x r2 [n]) during the adaptation of the algorithm.

Description

TECHNICAL FIELD

[0001] This discloses noise suppression systems with harmonic filtering. GENERAL STATE OF THE ART

[0002] Vehicles often generate airborne and structural noise while driving. In an effort to suppress this noise, active noise cancellation is often used to eliminate such noise by emitting a sound wave with an amplitude similar to that of the noise, but with an inverted phase. Such active noise cancellation can rely on both narrowband and broadband cancellation algorithms.

[0003] US 7,885,417 B2 describes an active noise control system and method for controlling an acoustic noise generated by a noise source at a listening position. In this system, sound in the vicinity of the listening position is recorded by a sound sensor; an electrical noise signal corresponding to the acoustic noise of the noise source is generated and adaptively filtered in accordance with control signals. The adaptively filtered noise signal is radiated by a sound reproduction device into the vicinity of the listening position, with a secondary path transfer function extending between the sound reproduction device and the sound sensor. The noise signal is filtered with a transfer function that models the secondary path transfer function. The signals supplied by the sound sensor after the initial filtering serve as control signals for the adaptive filtering.

[0004] US 5,425,105 A describes an active adaptive noise suppressor that does not require a training mode and operates over an extended noise bandwidth. The noise suppressor divides the noise bandwidth into frequency subbands, and multiple adaptive filter channels are used, one for each subband, to suppress the noise energy in the respective subbands. Each channel contains bandpass filters to limit the channel's operation to the respective subband, and delays are introduced into the operation of the filter weight update. Since each channel is stable in its subband, the noise suppressor operates over the extended noise bandwidth of all subbands.

[0005] US 5,841,876 A describes a vibration control system comprising a processor-based circuit that monitors and controls an analog vibration control circuit. The system includes a sensor for detecting unwanted vibrations and a synchronous pulse generator for determining the fundamental frequency of the vibrations emanating from the source. An actuator generates counter-noise to counteract unwanted vibrations. The processor circuit tests the system, monitors system functions, and adjusts various parameters to achieve optimal performance. SUMMARY

[0006] The invention is defined by a noise suppression system and a method according to the independent claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The embodiments of the present disclosure are particularly pointed out in the appended claims. However, other features of the various embodiments will become more apparent and best understood by reference to the following detailed description taken in conjunction with the accompanying drawings, in which: Fig. 1 illustrates an exemplary active noise cancellation system according to one embodiment; Fig. 2 illustrates an exemplary narrowband and broadband filter system of the system from Fig. 1; Fig. 3 illustrates another exemplary narrowband and broadband filter system of the system from Fig. 1; Fig. 4 illustrates another exemplary narrowband and broadband filter system of the system from Fig. 1; Fig. 5 illustrates an exemplary adaptive notch filter; Fig. Figure 6A illustrates a graphical representation of the size of the spectra of the broadband reference signal (x r-bb ); Fig. Figure 6B illustrates a graphical representation of the magnitude of the narrowband reference signal x r-nb ; Fig. Figure 6C illustrates a graphical representation of the size of the spectra of the reference signal with applied adaptive notch filter x̂ r-bb-flt [k,n]; Fig. Figure 7 illustrates an exemplary harmonic rejection filter as applied to error and reference signals; Fig. Figure 8A illustrates a graphical representation of the magnitude of the error signal E m ; Fig. Figure 8B illustrates a graphical representation of the spectra of the narrowband reference signal magnitude x r-nb ; Fig. Figure 8C illustrates a graphical representation of the spectra of the filtered error reference signal E m-fit ; Fig. 9 illustrates an adaptive reference filter system; Fig. Figure 10 illustrates another example of two simultaneous narrowband adaptive filter systems compared to the system of Fig. 1; Fig. Figure 11 illustrates another example of two simultaneous narrowband adaptive filter systems compared to the system of Fig. 1; Fig. Figure 12 illustrates an example of an adaptive harmonic rejection filter; Fig. Figure 13A illustrates an example of a graphical representation of the magnitude of the spectra of the error signal E 1 ; Fig. Figure 13B illustrates an exemplary graphical representation of the size of the spectra of the reference signal X r1 ; Fig. Figure 13C illustrates a graphical representation of the magnitude of the spectra of the error signal ê m ; Fig. 14 illustrates an exemplary process for adaptive error filtering, which is provided to the system from the Fig. 2 and Fig. 3 above; Fig. 15 illustrates an exemplary process 600 for adaptive reference filtering that may be applied to the system of Fig. 4 above; Fig. Figure 16 illustrates an exemplary process 700 for adaptive error filtering provided to the system of Fig. 11 above. DETAILED DESCRIPTION

[0008] Detailed embodiments of the present invention are disclosed herein as necessary; however, it is to be understood that the disclosed embodiments are merely exemplary of the invention that may be embodied in various and alternative forms. The figures are not necessarily to scale; some features may be exaggerated or reduced to show details of particular components. Therefore, the specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a representative basis for teaching those skilled in the art to widely employ the present invention.

[0009] Disclosed herein is an active noise cancellation system for increasing stability and quality during simultaneous operation of narrowband and broadband cancellation systems. For example, narrowband and broadband cancellation systems or algorithms often use a common error sensor and therefore both receive similar noise. In this case, both the narrowband and broadband algorithms may attempt to cancel the same frequency content, which may have different propagation paths. For example, some of the noise may be airborne in nature and coherent with the speed signal (e.g., engine RPM). Another portion of the noise may be structural in nature and coherent with a separate reference signal, such as a vehicle accelerometer.

[0010] In automotive applications, such coherent noise can be common, particularly under steady-state conditions such as idling, where engine noise emitted from the exhaust tailpipe is inherently airborne. This noise can be suppressed using an adaptive filtering technique such as FxLMS, ​​where the reference signal is provided by engine speed. At the same time, engine sway may be present, transmitting coherent structural noise at the same frequency, which can be suppressed using the adaptive filtering algorithm, with reference signals provided by accelerometers located on the chassis. When the two suppression algorithms run in parallel, instabilities or amplification may occur.

[0011] Current systems attempt to avoid such coherence by driving different outputs from each of the narrowband and wideband frequencies. This means that current systems operate independently in the narrowband and wideband frequencies and, as a result, cannot operate in overlapping frequency ranges. However, this is inefficient and can be inaccurate. The system disclosed herein filters out the common narrowband harmonic content from the error and / or reference signals. This prevents the wideband cancellation algorithm from providing an output that matches the narrowband content. If the narrowband content is filtered out from the error and / or reference signal, then the wideband algorithm should not adapt to the harmonic content to ensure a more stable system. This method is computationally more efficient than other coherence processing.The system can allow for lower latency because the computational overhead for the DSP is lower.

[0012] Fig. 1 illustrates an example active noise control system 100 including a controller 105, at least one input sensor 110, and at least one transducer 140. The controller 105 may be a standalone device including a combination of hardware and software components and may include a processor configured to analyze and process audio signals. In particular, the controller 105 may be configured to perform wideband and narrowband noise cancellation for Engine Order Cancellation (EOC) and / or Active Road Noise Cancellation (ARNC) within a vehicle based on data received from the input sensor 110. The controller 105 may include various systems and components for achieving ANC, such as an adaptive filtering system 132. The controller 105 may also operate a concurrent narrowband system, such as EOC.

[0013] The input sensor 110 may be configured to provide an input or reference signal to the controller 105. The input sensor 110 may include an accelerometer configured to detect motion or acceleration and provide an accelerometer signal to the controller 105. The acceleration signal may be indicative of vehicle acceleration, engine acceleration, wheel acceleration, etc. The input sensor 110 may also include a microphone and / or a sound intensity sensor configured to detect noise. The input sensor 110 may detect both narrowband noise and broadband noise, as described with respect to Fig. 2. Input sensor 110 may also detect multiple sets of noise, including a first set of narrowband noise signals and a second set of narrowband noise signals. For narrowband systems, the input sensor may simply be engine speed, engine torque, or other reference signals.

[0014] The transducer 140 may be configured to acoustically generate an audio signal provided by the controller 105 at an output channel (not labeled). In one example, the transducer 140 may be included in a motor vehicle. The vehicle may include multiple transducers 140 arranged at various locations throughout the vehicle, such as the front right, front left, rear right, and rear left. The audio output at each transducer 140 may be controlled by the controller 105 and may be subject to noise cancellation and other parameters that affect its output. The transducer 140 may provide the noise cancellation signal to assist the RNC in improving the sound quality in the vehicle.

[0015] The active noise control (ANC) system 100 may include a feedback or output sensor 145, such as a microphone, located on a secondary path 176 and capable of receiving audio signals from the transducer 140. The feedback sensor may be a microphone configured to transmit a microphone output signal or error signal to the controller 105. The feedback sensor may also receive unwanted noise from the vehicle, such as road noise and engine noise.

[0016] Fig. 2 illustrates an exemplary narrowband and wideband filter system 132 of the ANC system 100. The narrowband filter system 132 may include a narrowband primary path 152 that provides a time-dependent primary narrowband propagation noise signal P r,mn [n], and a broadband primary path 154 providing a time-dependent primary broadband propagation noise signal P r,mb[n]. In one example, the narrowband propagation path P r,mn [n] can be captured by a microphone, accelerometer, sound intensity sensor, etc., as it correlates with one or more speed sensors configured to detect the rotation of engine fan shafts or other speed-related noise. The broadband noise propagation path P r,mb [n] can be detected by a microphone, accelerometer, sound intensity sensor, etc.

[0017] The system 132 can provide two feedforward reference signals, a narrowband reference signal x rn [n] and a broadband reference signal x rb [n]. Additionally or alternatively, the two feedforward reference signals may be two narrowband reference signals, as described below with respect to Fig. 10. Each of the reference signals can contain overtone noise. The narrowband reference signal x rn [n] can be fed to a bandpass filter 156. The bandpass filter 156 can select certain frequencies from the narrowband reference signal x rn [n] which is then fed to a narrowband adaptive filter 160 and then to the secondary path 176.

[0018] The broadband reference signal x rb [n] can be fed to a broadband adaptive filter 174. The broadband adaptive filter 174 can filter the broadband reference signal x rb [n] filter and a broadband secondary signal y lb [n] generate.

[0019] The broadband reference signal x rb [n] and the time-dependent primary narrowband propagation path P r,mn[n] can be provided to a Fast Fourier Transform block 164. An FFT can be applied to the estimation block 158 of the secondary path of the broadband reference signal x rb [n] can be applied.

[0020] The secondary path estimation block 158 may include a secondary path in the frequency range Ŝ l,rn [k] and an estimated secondary path in the time domain ŝ l,m [k]. The secondary path estimation block 158 may provide a broadband least mean square block 170 with an RxLxM matrix, where: R is the total dimensional number of reference signals, L is the total dimensional number of secondary sources and M is the total dimensional number of error signals.

[0021] The wideband least mean square (LMS) block 170 may be a sum cross-spectrum comparator configured to provide a vector for applying filter coefficients to the least mean square of the error signals. An inverted FFT may then be applied to this signal at the IFFT block 172. An RxL matrix may then be fed to a wideband adaptive filter 174.

[0022] The secondary path estimation block 158 may also provide an RxLxM matrix to a narrowband least mean square (LMS) block 162, which may be a summed cross-spectrum comparator or time-domain comparator configured to provide a vector configured to apply least-mean-square filter coefficients to the error signals. The narrowband least-mean-square block 162 may provide an RxL matrix to the narrowband adaptive filters 160. A bandpass filter (BPF) 171 may be disposed between the summed error signal and the narrowband LMS 162 for time alignment.

[0023] The broadband adaptive filter 174 can filter the broadband secondary source signal y lb [n] and the narrowband adaptive filter 160 can supply the narrowband secondary source signal y ln[n], which are summed together. The summed secondary source signals y lb [n], y ln [n] can then use the secondary path s l,m [n] 176. The secondary path s l,m [n] 176 represents the electroacoustic transfer function of the system (electronics, loudspeakers, microphones and vehicle interior acoustics).

[0024] In summation 178, the anti-noise signals that travel via the secondary path s l,m [n] 176, which are sent to primary paths 152 and 154, are summed, resulting in an error signal e m [n]. The error signal e m [n] can be detected by the output sensors 145, such as a microphone. The summed signal can be input to a Fast Fourier Transform 180, which produces an estimated error signal E m [n] forms.

[0025] A harmonic rejection filter 182 can then be applied to the estimated error signal E m[n] using the narrowband reference signal x nb [n] are applied. The harmonic rejection filter 182 is described herein with respect to Fig. 7. The harmonic rejection filter 182 can filter the harmonic signals in the error signal E m [n]. The broadband signal can then be found in the error signal E without the narrowband signal and without taking the overtone noise into account. m [n] adjust.

[0026] The harmonic rejection filter 182 subtracts the output of a narrowband harmonic content from E m [n]. The adaptive filter 240 converts the narrowband reference signal into the best estimate of the signal in E m [n] existing narrowband interference. The algorithm of the LMS 170 (or similar) updates the adaptive filter coefficients by feedback from E m-flt[k,n]. The adaptive filter 240 may have one of several filter structures: FIR, IIR, or simply sinusoids with adjustable magnitude and phase.

[0027] Fig. 3 illustrates an exemplary narrowband filter system 132 of the ANC system 100, similar to Fig. 2, assume that the harmonic rejection filter 182 is replaced by a Wiener filter 186. The Wiener filter can filter the harmonic noises based on the reference signals x rb [n], x rn [n] and the noise from the broadband error signal E m [n] filter. Narrowband interference suppression can remove the narrowband content from the broadband error signal. Narrowband interference can be achieved through several mechanisms. One example is the use of LMS, as described in the Fig. 5 and Fig. 7. Using LMS, prior matrix inversion methods are unnecessary. Another example is the use of RLS (recursive least squares) or other adaptive filters. Furthermore, linear forward-backward prediction can also be used. In this case, the Wiener filter 186 can be implemented.

[0028] The one in the Fig. 2 and Fig. 7 illustrated overtone rejection filter 182 and the one in Fig. The Wiener filters 186 illustrated in Figure 3 can allow the harmonic content to be separated from the broadband content. This prevents the road noise cancellation system from adapting to the harmonic content, at least because the harmonic content has been removed from the error signals. The use of a notch filter can allow phase distortion and the loss of broadband content. The use of a Wiener filter may work best when the signals are uncorrelated, or there may be a partial correlation.

[0029] Fig. 4 illustrates an exemplary narrowband filter system 132 of the ANC system 100, similar to Fig. 2, except that an adaptive notch filter 192 is arranged between the bandpass filters 156 and the secondary path estimation block 158. In this example, the adaptive notch filter 192 is illustrated as including both the narrowband reference signal x rn [n] as well as the broadband reference signal x rb [n]. However, in another example, the adaptive notch filter 192 may only receive the broadband reference signal x rb [n]. In this latter example, the adaptive notch filter 192 can extract harmonic content from the broadband reference signal x rb[n] filter to avoid duplication of harmonic content in both the narrowband and broadband. If both the narrowband and broadband have similar harmonic noise, and each is filtered separately by the narrowband and broadband, the narrowband filtering may suppress the same frequency content as the broadband filtering. This can create instabilities or gains and result in poorer noise rejection.

[0030] Fig. 5 illustrates an exemplary adaptive notch filter 192. The harmonic content is removed via the notch filter by applying an adaptive filter 240 and LMS 242 to the narrowband reference signal x r-nb[n] from the wideband content. The adaptive filter 240 may be similar to the narrowband adaptive filter 160 and is configured to apply least-mean-square filter coefficients to the error signals. The LMS 242 may be similar to the narrowband LMS block 162 and may provide a dimension R matrix to the adaptive filter 240.

[0031] Fig. Figure 6A illustrates a graphical representation of the magnitude of the broadband reference signal x r-bb in the frequency range.

[0032] Fig. Figure 6B illustrates a graphical representation of the magnitude of the narrowband reference signal x r-nd in the frequency domain. In the illustrated example, the magnitude of the narrowband reference signal x r-nd two peaks. These two peaks can be compared with the magnitudes of the broadband reference signal x r-bbcorrelate. Based on this correlation, the similarity between the broadband and narrowband reference signals is determined. Due to this similarity and current noise cancellation systems, the narrowband and broadband reference signals may attempt to suppress this similar content. Filtering the similar narrowband content from the broadband reference signals can enable more accurate noise cancellation.

[0033] Fig. Figure 6C illustrates a graphical representation of the magnitude of the reference signal in the time domain with applied adaptive notch filter x̂ r-flt[k,n]. As illustrated, the correlated peaks present before filtering are no longer present. Thus, the similar content between the narrowband and broadband reference signals has been removed from the broadband reference signal, providing a more suitable reference signal for noise cancellation.

[0034] Fig. 7 illustrates an exemplary harmonic rejection filter 182 that is similar to the harmonic rejection filter 182 of Fig. 2. The harmonic content is removed via the harmonic rejection filter 182 by applying an adaptive filter 240 and LMS 242 to the narrowband reference signal x r-nb [n] from the broadband content. The filtered narrowband reference signal is derived from the error signal E m [n] is removed to obtain the filtered error reference signal E m-flt [k,n] to generate.

[0035] Fig. Figure 8A illustrates a graphical representation of the magnitude of the error signal Em [n] in the frequency domain.

[0036] Fig. Figure 8B illustrates a graphical representation of the magnitude of the narrowband reference signal x r-nd in the frequency range, similar to Fig. 6B.

[0037] Fig. Figure 8C illustrates a graphical representation of the filtered error reference signal E m-flt [k,n]. As illustrated, the correlated peaks present before filtering in the filtered error reference signal E m-flt [k,n] is no longer present. Thus, the similar content between the error and narrowband reference signals has been removed from the error signal, allowing for a more suitable signal for ANC system adaptation.

[0038] Fig. Figure 9 illustrates an adaptive reference filter system 250. Coherence calculations can only apply to stationary processes that are stochastic and whose joint probability does not change with time shift. Stationary processes are a fundamental assumption in certain statistical calculations that use time series analysis. The cross-spectrum analysis within the coherence calculation may depend on the stochastic process. Furthermore, some narrowband reference signals do not have a mean value of 0. For these reasons, it may be advantageous to use time-frequency coherence to evaluate the reference signals x m [n], x rb [n] to be used.

[0039] The adaptive reference filter system 250 may include a time-frequency coherence analysis block 252. This block 252 may analyze the narrowband reference signal x rn [n] and the broadband reference signal x rb[n] to determine the coherence between the two signals. In one example, wavelet coherence may be used to detect oscillations in the non-stationary signals. Time-frequency coherence analysis block 252 may determine a coherence value based on the comparison.

[0040] At block 256, the controller 105 may determine whether the coherence value meets or exceeds a coherence threshold. The threshold may be a minimum degree of coherence. This value may be defined as any number X, where 0 <X≤1. In einem Beispiel kann die Wavelet-Kohärenz 0,5 oder größer sein. Ein Wert von 0,5 oder größer kann also den Filterprozess auslösen. Wenn der Schwellenwert nicht erreicht wird, kann die Steuerung 105 die Frequenz identifizieren, bei der die Kohärenz nicht erreicht wird, und zwar im Frequenzidentifikationsblock 258, und einen Sperrfilter bei dieser Frequenz bei Block 260 anwenden. Die Steuerung 105 kann Sperrfilter anwenden, bis der Schwellenwert für jede Frequenz erreicht ist.

[0041] With reference to the Fig. 10 and Fig. 11, in addition to scenarios with a broadband and a narrowband reference, there may also be systems with two (or more) independent narrowband references. Since these narrowband signals approach each other in frequency, attenuation of the tracking bandpass filters can cause slapping effects.

[0042] Fig. 10 illustrates an exemplary narrowband filter system 342 of the ANC system 100 for systems with simultaneous multiple inputs, multiple outputs, multiple least-mean-square systems with adaptive reference filtering applied to multiple narrowband signals with different references. The filter system 342 may include a narrowband primary path 352 that defines a time-dependent first narrowband primary propagation path P r,rml [n], and a second primary narrowband propagation path 354 which provides a time-dependent primary narrowband propagation path P r,m2[n]. The primary noise signals 352, 354 can be detected by the output sensors 145 (as in Fig. 1 shown).

[0043] The system 342 may include two feedforward reference signals (e.g., input signals), a first narrowband reference signal x r1 [n] and a second narrowband reference signal x r2 [n] are received. Each of the reference signals may contain overtone noise. The narrowband reference signals Xr1 [n], x r2 [n] can be fed to a reference gain controller 346. The reference gain controller 346 can control the reference signals x r1 [n], x r2[n] to favor the suppression of one noise signal over the other. As the frequencies of the two systems approach each other, one reference can be switched off, disabling suppression for that sequence. This way, some noise will still be present in the system.

[0044] The controller 105 can determine whether the frequencies of the narrowband reference signals are within a predefined threshold of each other. If so, the controller 105 can exclude one of the reference signals from consideration. The threshold can relate to frequency, magnitude, or coherence, to name a few. In one example, if two reference signals produce the same frequency content (but perhaps a different phase), one of the reference signals can be muted if they are within 5 Hz of each other. This can be beneficial if one of the noise sources associated with a reference signal is known to be significantly more dominant than the other in certain frequency ranges. Similarly, a signal can be muted based on the magnitude of the reference signals, e.g., if the magnitudes are within 3 dB of each other.Again, this is a reason for a phase shift, where if the signals are significantly similar in amplitude, amplification can be prevented.

[0045] The reference signals x r1 [n], x r2 [n] can be fed to a first secondary path estimation block 358 and a second secondary path estimation block 359, respectively. The secondary path estimation blocks 358, 359 can estimate a secondary path for the time domain and the frequency domain, respectively, and generate an estimated secondary path in the frequency domain Ŝ l,m [k] and an estimated secondary path in the time domain ŝ l,m[k]. In particular, these secondary paths may be unique or common. For example, a first reference signal may be reproduced only by the rear subwoofer because that specific speaker couples with the exhaust noise, which is the dominant source. In this example, the other speakers are used for the second set of reference signals. The secondary path estimation blocks 358, 359 may provide an RxLxM matrix to a broadband least mean square block 170, where: R is the total dimensional number of reference signals, L is the total dimensional number of secondary sources and M is the total dimensional number of error signals.

[0046] The secondary path estimation blocks 358, 359 may provide the RxLxM matrices for the respective first narrowband least mean square (LMS) block 362 and second narrowband least mean square (LMS) block 363. The LMS blocks 362, 363 may be adaptive filters configured to apply least mean square filter coefficients to the error signals. The narrowband least mean square blocks 362, 363 may provide an R 1 xL matrix for a first narrowband adaptive filter 360 and an R 2 xL matrix for a second narrowband adaptive filter 361.

[0047] The first narrowband adaptive filter 360 may receive a first secondary source signal y 1 [n] and the second narrowband adaptive filter 361 can supply a second secondary source signal y 2[n], which are summed together. The summed secondary source signals y 1 [n], y 2 [n] can then use the secondary path s l,m [n] 376. Again, this may be a common secondary path or a unique secondary path, as in the estimates in Fig. 10. The secondary path s l,m [n] 376 represents the transfer function of the acoustic system (loudspeakers, microphones and vehicle interior acoustics).

[0048] In summation 378, the signals that travel through the secondary path s l,m [n] 376, which are sent to primary paths 352 and 354, are summed, resulting in an error signal e m [n]. The error signal e m[n] can be detected by the output sensors 145, such as a microphone. The summed signal can be input to a first bandpass filter 356 and a second bandpass filter 357, which select certain frequencies from the error signal e m [n]. The filtered error signal is then fed to the respective first block 362 of the narrowband LMS and the second block 363 of the narrowband LMS.

[0049] Fig. 11 illustrates an exemplary narrowband filter system 342 of the ANC system, similar to Fig. 10, except that a first adaptive notch filter 392 is arranged between the first bandpass filter 356 and the first LMS block 362. Furthermore, a second adaptive notch filter 393 is arranged between the second bandpass filter 357 and the second LMS block 363. The adaptive filters 392, 393 can then filter out harmonic content from the filtered error signal to avoid duplication of the harmonic content in both the first and second narrowband signals. If both narrowband signals contain similar harmonic noise and each of the sets of narrowband signals is filtered separately, the filtering can suppress the same frequency content in each set of signals. This can create instabilities or amplifications and lead to poorer noise suppression.

[0050] That is, to automatically isolate the error signals related to the two references, adaptive filters 392, 393 can be used to remove the noise component of the first reference before the adaptive filter responsible for suppressing signals related to the second reference. A parallel system can also be operated that isolates only the noise related to the first reference in the same way. This can be repeated for more than two narrowband reference signals. In this system, the reference gain control block 346 can be adjusted to keep both references active based on the performance of the narrowband adaptive filter, thus maximizing the amount of suppression in the system.

[0051] Fig. 12 illustrates an exemplary harmonic rejection filter 392 or harmonic rejection filter 393. The harmonic content is removed via the rejection filter by applying an adaptive filter 340 and LMS 380 to the narrowband reference signal x rl [n] from the wideband content. Adaptive filter 340 may be similar to narrowband adaptive filter 160 and configured to apply least-mean-square filter coefficients to the error signals. LMS 380 may be similar to narrowband LMS block 162 and may provide an RxL matrix to adaptive filter 340.

[0052] Fig. Figure 13A illustrates a graphical representation of the magnitude of the error signal E 1 in the frequency range.

[0053] Fig. Figure 13B illustrates a graphical representation of the magnitude of the reference signal X r1 in the frequency domain. In the illustrated example, the magnitude of the reference signal X r1a peak. This peak can be measured with the magnitude of the error signal E 1 (f) correlate. Based on this correlation, the similarity between the broadband and narrowband reference signals is determined. Due to this similarity and current noise cancellation systems, the two narrowband reference signals may attempt to suppress this similar content when the frequencies are very close to each other.

[0054] Fig. Figure 13C illustrates a graphical representation of the magnitude of the reference signal in the frequency domain with the error signal ê m As illustrated, the correlated peaks present before filtering are no longer present. Thus, the similar content between the two narrowband reference signals has been removed from one of the reference signals, allowing for a more suitable reference signal for noise cancellation.

[0055] Fig. Figure 14 illustrates an exemplary process 500 for adaptive reference filtering provided to the system from the Fig. 12 and Fig. 13 above. The process 500 may begin at block 505, where the controller 105 receives input signals, including a narrowband input signal from the input sensor 110.

[0056] At block 515, the controller 105 may apply a filter to at least one of the input signals x rn [n], x rb [n]. The filter may include a bandpass filter, such as bandpass filter 156. The filter may be an adaptive filter, such as narrowband adaptive filter 160.

[0057] At block 520, the controller 105 may generate a secondary path representing the electroacoustic transfer function of the system, similar to the secondary path estimation block 158 of the Fig. 2 and Fig. 3.

[0058] At block 525, the controller 105 may sum the antinoise and primary noise to generate an error signal. In this example, the antinoise signals sent via the second path s l,m [n] 176 are summed with the noise coming from the primary paths 152, 154, resulting in an estimated error signal E m [n] leads.

[0059] At block 530, the controller 105 may apply an adaptive filter (e.g., the harmonic rejection filter 182 of the Wiener filter 186) to the estimated error signal. The adaptive filter may reduce the harmonic signals in the error signal. m [n]. The broadband signal can then be found without the narrowband signal and without taking the overtone noise into account in the error signal e m [n] adjust.

[0060] At block 510, the controller 105 may apply the secondary path estimate to the input signals.

[0061] At block 535, the controller 105 may take the least mean square (LMS) of the filtered error signal from block 530 and the secondary estimate from block 510.

[0062] At block 540, the controller 105 may take the IFFT of the signal.

[0063] At block 545, the controller 105 may update the system with the filter based on the process 500.

[0064] Process 500 can then end.

[0065] Fig. 15 illustrates an exemplary process 600 for adaptive reference filtering that may be applied to the system of Fig. 4 above. The process 600 may begin at block 605, where the controller 105 receives input signals, including narrowband input signals and wideband input signals from the input sensor 110 (or multiple narrowband signals).

[0066] At block 615, the controller 105 may apply an adaptive filter (e.g., the harmonic rejection filter 182 of the Wiener filter 186) to one or more of the input signals. The adaptive filter may filter out the harmonic signals in the reference. In one example, the adaptive rejection filter 192 is in Fig. 4 in such a way that it contains both the narrowband reference signal x rn [n] as well as the broadband reference signal x rb [n]. However, in another example, the adaptive notch filter 192 may only receive the broadband reference signal x rb [n]. In this latter example, the adaptive notch filter 192 can extract harmonic content from the broadband reference signal x rb[n] filter to avoid duplication of harmonic content in both the narrowband and broadband. If both the narrowband and broadband have similar harmonic noise, and each is filtered separately by the narrowband and broadband, the narrowband filtering may suppress the same frequency content as the broadband filtering. This can create instabilities or gains and result in poorer noise rejection.

[0067] At block 620, the controller 105 may apply a secondary path representing the electroacoustic transfer function of the system, similar to the secondary path estimation block 158 of Fig. 4.

[0068] At block 608, the controller 105 may apply a filter to the input signals.

[0069] At block 612, the controller 105 may apply an estimate of the secondary path to the filtered input signal.

[0070] At block 625, the controller 105 may sum the anti-noise and primary noise signals to generate an error signal. In this example, the anti-noise signals sent via the second path s l,m [n] 176 are summed with the noise coming from the primary paths 152, 154, resulting in an estimated error signal E m [n] leads.

[0071] At block 630, the controller 105 may take the least mean square (LMS) of the secondary estimate from block 620 and block 612.

[0072] At block 630, the controller 105 may take the least mean square (LMS) of the filtered error signal from block 530 and the secondary estimate from block 510.

[0073] At block 635, the controller 105 may take the IFFT of the signal.

[0074] At block 640, the controller 105 may update the system with the filter based on the process 600.

[0075] Process 600 can then end.

[0076] Fig. Figure 16 illustrates an exemplary process 700 for adaptive reference filtering provided to the system from the Fig. 10 and Fig. 11 above. The process 700 may begin at block 705, where the controller 105 receives input signals, including at least two narrowband input signals from the input sensor 110.

[0077] At block 710, the controller 105 may apply a bandpass filter to the input signals.

[0078] At block 715, the controller 105 may apply a reference gain control 346 to adjust the filtered input signals x r1 [n], x r2 [n] to favor the suppression of one noise signal over the other.

[0079] At block 718, the controller 105 may apply a filter to the reference gain controlled signal from block 715.

[0080] At block 720, the controller 105 may generate a secondary path representing the electroacoustic transfer function of the system, similar to the secondary path estimation block 358 of the Fig. 10 and Fig. 11.

[0081] At block 725, the controller 105 may sum the primary noise signal from the primary path and the anti-noise signal from the secondary path to generate an estimated error signal.

[0082] At block 718, the controller 105 may apply a bandpass filter to the summed signal.

[0083] At block 730, the controller 105 may apply an adaptive filter (e.g., the adaptive notch filters 392, 393) to the filtered estimated error signal. The adaptive filter may reduce the harmonic signals in the error signal m [n] filter out.

[0084] At block 735, the controller 105 may apply the secondary path estimate to the input signals.

[0085] At block 740, the controller 105 may take the least mean square (LMS) of the filtered error signal from block 730 and the secondary estimate from block 735.

[0086] At block 745, the controller 105 may update the system with the filter based on the process 700.

[0087] Process 700 can then end.

[0088] The embodiments of the present disclosure generally provide a plurality of circuits, electrical devices, and at least one controller. All references to the circuits, the at least one controller, and other electrical devices, and the functions provided by each, are not intended to be limited to including only what is illustrated and described herein. While specific designations may be assigned to the various circuit(s), controller(s), and other electrical devices disclosed, such designations are not intended to limit the scope of operation for the various circuit(s), controller(s), and other electrical devices. Such circuit(s), controller(s), and other electrical devices may be combined and / or separated from one another in any manner based on the particular type of electrical implementation desired.

[0089] It will be appreciated that controllers disclosed herein may include any number of microprocessors, integrated circuits, memory devices (e.g., FLASH, random access memory (RAM), read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or other suitable variations thereof), and software that cooperate with one another to perform the operation(s) disclosed herein. In addition, any disclosed controller utilizes any one or more microprocessors to execute a computer program embodied in a non-transitory computer-readable medium that is programmed to perform any number of disclosed functions.Furthermore, any controller as provided herein includes a housing and the varying number of microprocessors, integrated circuits, and memory devices (e.g., FLASH, Random Access Memory (RAM), Read-Only Memory (ROM), Electrically Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM)) positioned within the housing. The disclosed controller(s) also include hardware-based inputs and outputs for receiving and transmitting data, respectively, from and to hardware-based devices, as discussed herein.

[0090] With respect to the processes, systems, methods, heuristics, etc. described herein, it is to be understood that although the steps of such processes, etc., are described as occurring in a particular order, these processes may also be performed such that the described steps are performed in an order different from that described herein. It is further understood that certain steps may be performed concurrently, other steps may be added, or certain steps described herein may be omitted. In other words, the descriptions of processes in this specification are for the purpose of illustrating particular embodiments and should in no way be construed to limit the claims.

[0091] While exemplary embodiments are described above, these embodiments are not intended to describe all possible forms of the invention. Rather, the terms used in the description are words of description rather than limitation, and it is understood that various changes may be made without departing from the spirit and scope of the invention. In addition, the features of different implementing embodiments may be combined to form further embodiments of the invention.

Claims

[1] A noise cancellation system (100) with harmonic filtering for a vehicle audio system, comprising: at least one first input sensor configured to receive a reference signal (x r1 [n], x r2 [n]); at least one second input sensor configured to transmit at least two narrowband input signals, each of the input signals (x r1 [n], x r2 [n]) contains overtone noises; a processor (105) programmed to: Receiving the reference signal, wherein the reference signal comprises at least two narrowband reference signals (x r1 [n], x r2 [n]) includes, Receiving the narrowband input signals, Applying a gain reference control (346) to the reference signals to determine whether the frequencies of each of the reference signals (x r1 [n], x r2[n]) are within a predefined range of each other, Applying a bandpass filter (156) to one of the reference signals (x r1 [n], x r2 [n]), Applying a secondary path (376) based on the filtered reference signals (x r1 [n], x r2 [n]) to generate anti-noise signals, and Summing (378) the anti-noise signals and the input signals to produce an error signal (e m [n]) at the output sensor (145), Applying an adaptive filter (392) to the error signal (e m [n]) to remove overtone noise from the error signal (e m [n]) and Removing one of the reference signals (x r1 [n], x r2 [n]) in response to the frequencies of each of the reference signals (x r1 [n], x r2[n]) must be within the predefined range of each other to prevent common overtone content from being detected on both reference signals (x r1 [n], x r2 [n]) during the adaptation of the algorithm. [2] The system of claim 1, wherein the adaptive filter (392, 393) includes at least one harmonic rejection filter. [3] The system of claim 1, wherein the adaptive filter (392, 393) includes at least one Wiener filter. [4] The system of claim 2, wherein the processor (105) is further configured to apply a narrowband adaptive filter (360) to the incoming reference signals (x r1 [n], x r2 [n]) after applying a gain control (346). [5] A method of a noise cancellation system (100) with harmonic filtering for a vehicle audio system, comprising: Receiving at least two narrowband reference signals (x r1 [n], x r2 [n]), Receiving at least two narrowband input signals, Applying a gain reference control (346) to the narrowband reference signals (x r1 [n], x r2 [n]) to determine whether the frequencies of each of the reference signals (x r1 [n], x r2 [n]) are within a predefined range of each other, Applying a bandpass filter (156) to one of the reference signals (x r1 [n], x r2 [n]), Applying a secondary path (376) based on the filtered reference signals (x r1 [n], x r2 [n]) to generate anti-noise signals, and Summing (378) the anti-noise signals and the input signals to produce an error signal (e m [n]) at the output sensor (145), Applying an adaptive filter (392) to the error signal (e m [n]) to remove overtone noise from the error signal, Removing one of the reference signals (x r1 [n], x r2 [n]) in response to the frequencies of each of the reference signals (x r1 [n], x r2 [n]) are within the predefined range of each other.

Images