Patent Grant
9478209
This patent application claims priority from EP Application No. 11 190 092.4 filed Nov. 22, 2011, which is hereby incorporated by reference.
The present invention relates to the field of active audio noise control, and in particular to tunable multiple-channel noise control systems and methods.
Acoustic noise problems are becoming more and more evident as an increasing amount of industrial equipment such as engines, blowers, fans, transformers, and compressors are being used. The traditional approach to acoustic noise control uses passive techniques such as enclosures, barriers, and silencers to attenuate the undesired noise. These passive silencers are valued for their high attenuation over a broad frequency range; however, they are relatively large, costly, and ineffective at low frequencies. Mechanical vibration is another related type of noise that commonly creates problems in all areas of transportation and manufacturing, as well as in many household appliances. Active noise control (ANC) involves an electroacoustic or electromechanical system that cancels the primary (unwanted) noise based on the principle of superposition; specifically, an antinoise of equal amplitude and opposite phase is generated and combined with the primary noise, thus resulting in the cancellation of both noises. The ANC system efficiently attenuates low-frequency noise where passive methods are either ineffective or tend to be relatively expensive or bulky. ANC permits improvements in noise control, often with potential benefits in size, weight, volume, and cost.
A basic design of acoustic ANC utilizes a microphone, a filter and a secondary source such as a loudspeaker to generate a canceling sound. Since the characteristics of the acoustic noise source and the environment are time varying, the frequency content, amplitude, phase, and sound velocity of the undesired noise are nonstationary. An ANC system must therefore be adaptive in order to cope with these variations.
Multi-channel active noise control is achieved by introducing a canceling “antinoise” wave through an appropriate array of secondary sources. These secondary sources are interconnected through an electronic system using digital signal processing for the particular cancellation scheme. The basic adaptive algorithm for ANC has been developed and analyzed based on single-channel broad-band feedback or feedforward control as set forth by, e.g., S. M. Kuo, D. R. Morgan, “Active Noise Control: A Tutorial Review”, PROCEEDINGS OF THE IEEE, VOL. 87, NO. 6, June 1999. These single-channel ANC solutions are expanded to multiple-channel cases using various online secondary-path modeling techniques and special adaptive algorithms, such as lattice, frequency-domain, subband, and recursive-least-squares. In numerous situations, however, it is not desired to cancel all noise but to modify the noise in order to be perceived as more pleasant by a listener.
There is a need for tunable noise control systems and methods that are suitable also for multi-channel applications.
An active noise control system for tuning an acoustic noise signal at a listening position comprises a microphone that converts acoustic signals into electric signals and that is arranged at the listening position; a loud-speaker that converts electrical signals into acoustic signals and that radiates a noise cancelling signal via a second path to the microphone; a secondary noise source that generates an electrical noise signal modeling the acoustic noise signal; a first filter that has a controllable first transfer characteristic and that is connected between the secondary noise source and the loudspeaker; a second filter that has a second transfer characteristic and that is connected downstream of the secondary noise source; a third filter that has a controllable third transfer characteristic and that is connected downstream of the second filter; a noise signal estimator that is connected downstream of the microphone and that provides an estimate of the acoustic noise signal; and an adaptive filter controller that is downstream of the second filter and downstream of the noise signal estimator and that controls the transfer characteristic of the third filter. The second transfer characteristic is an estimation of the transfer characteristic of the secondary path. The first transfer characteristic is controlled by the third transfer characteristic via a filter coefficient copy path. A first weighting element is connected into the filter coefficient copy path and/or a second weighting element is connected downstream of the noise signal estimator.
In a second embodiment, an active noise control method for tuning an acoustic noise signal at a listening position comprises converting acoustic signals at the listening position into electric signals; generating an electrical noise signal modeling the acoustic noise signal; filtering the electrical noise signal that models the acoustic noise signal with a controllable first transfer characteristic to provide a first filtered noise signal; converting the first filtered noise signal into an acoustic signal which is radiated via a second path to the listening position; filtering the electrical noise signal that models the acoustic noise signal with a second transfer characteristic to provide a second filtered noise signal; adaptively filtering with a third transfer characteristic the second filtered noise signal; providing an estimate of the acoustic noise signal from the converted acoustic signal at the listening position. The second transfer characteristic is an estimate of the transfer characteristic of the secondary path. The first transfer characteristic is controlled by the third transfer characteristic via a filter coefficient copy path. A first weighting process is performed in the filter coefficient copy path and/or a second weighting process is applied to the estimate of the acoustic noise signal.
These and other objects, features and advantages of the present invention will become apparent in light of the detailed description of the embodiments thereof, as illustrated in the accompanying drawings. In the figures, like reference numerals designate corresponding parts.
Various specific embodiments are described in more detail below based on the exemplary embodiments shown in the figures of the drawing. Unless stated otherwise, similar or identical components are labeled in all of the figures with the same reference numbers.
In the following description, noise is defined as any kind of undesirable disturbance, whether it is created by electrical or acoustic sources, vibration sources, or any other kind of media. Therefore, ANC algorithms disclosed herein can be applied to different types of noise using appropriate sensors and secondary sources.
In the ANC system of
The compensation signal y_a is also supplied to a filter 7 to generate a compensation signal y_a_hat therefrom, which is subtracted from the error signal e_a by a subtractor 8 to provide an electrical disturbance signal d_hat. The filter 7 and the subtractor 8 form an estimator that provides an estimate of the acoustic disturbance signal d, i.e., electrical disturbance signal d_hat. However, any other type of estimator may be used.
The reference noise signal x is supplied to a filter 9 that provides a modified noise signal x′, which is provided to an adaptive filter having a controlled filter 10 and a filter controller 11. Adaptive filters adjust (e.g., with their filter controller 11) their coefficients (in their controlled filter 11) to minimize an error signal, and adaptive filters can be realized for example as (transversal) finite impulse response (FIR), (recursive) infinite impulse response (IIR), lattice, or transform-domain filters. The most common form of adaptive filter is the transversal filter using the least-mean-square (LMS) algorithm. In the present example, the modified noise signal x′ is supplied to both the controlled filter 10 and the filter controller 11, whereby the filter controller 11 controls the controlled filter 10, i.e., adapts the filter coefficients of the controlled filter 10. The controlled filter 10 together with a subsequent real part processor 12 provides a signal y′_p to an adder 13, which also receives the electrical disturbance signal d_hat. In addition to the signal x′, the filter controller 11 also receives a modified error signal e_p from the adder 13 (at its error signal input).
The controlled filter 10 has a transfer characteristic W_p and the filter 2 has a transfer characteristic W_a, which is a copy of the transfer characteristic W_p of the controlled filter 10, i.e., both characteristics are identical or the transfer characteristic W_a is updated on a regular basis by the transfer characteristic W_a. Matching of the filters is performed via a filter coefficient copy path between the filters 2 and 10. The filters 7 and 9 both have an identical transfer characteristic S_hat that is an approximation of a transfer characteristic S of the secondary path 5. Accordingly, the ANC system of
The adaptive filter 10 in connection with the real part processor 12 generates from the complex reference noise signal x′ the real signal y′_p, which ideally is identical with or at least rather similar to disturbing noise signal d. In an ideal adapted system the following relations apply:
y′_p=−d_hat
Re{x′·W_p}=−d_hat
Re{x·S_hat·W_p}=−d_hat
in which the active branch may be identical with the passive branch:
W_a=W_p.
Adaption is performed in the present case according to a least-mean-square (LMS) algorithm in a time-discrete manner, according to which:
W_p[n]=W_p[n−1]+μ·x′·e_p,
in which μ stands for the step size of the LMS algorithm that controls the amount of gradient information used to update each coefficient.
The single-channel ANC system described above with reference to
fm=m·rpm/60 with m=1, 2, 3 . . . ,
in which fm is the frequency of the m-th harmonic with the first harmonic (m=1) being the fundamental and rpm are the revolutions per minute.
In the present example, an orthogonal signal generated by the oscillator in connection with complex filters are used so that the adaptive filter and its shadow filter each have a double set of filter coefficients, one for the real part and one for the imaginary part of the complex oscillator signal, i.e., reference noise signal x. However, the complex filter may produce a complex output signal even when its input signal is real. The reference noise signal x can be described as follows:
x=ejwn=cos(w·n)+j sin(w·n) with
w=2πfm/fs,
in which fm is the frequency of the orthogonal noise signal, n is the discrete time index and fs stands for the sample rate of the system.
Accordingly, the complex adaptive transfer characteristics W_a and W_p are:
W_a=w_a_re+j·w_a_im,
W_p=w_p_re+j·w_p_im.
Finally, an operator Re of the real part processors 3 and 12 can be described by
Re(A·ejx)=A cos(x).
The real part processors 3 and 12 convert complex signals into real signals that are to be radiated by the loudspeaker 4. Processing of complex signals with subsequent conversion into real signals is an efficient way of implementing such a signal processing system.
The secondary path 5 has a transfer characteristic S and represents the path between the input circuit of the loudspeaker 4 (including digital-analog converters, amplifiers etc.) and the output circuit of the microphone 6 (including amplifiers, analog-digital converters, etc.), or in terms of signals, between the, e.g., digital signals y_a and e_a. The filters 7 and 9 each have a transfer characteristic S_hat and model the secondary path 5. Accordingly, electrical signal d_hat models/estimates the acoustic disturbance signal d. If S_hat=S, then d_hat=d. d_hat is the target for adaption of the adaptive filter (10, 11), also referred to as the desired signal for adaption of the transfer characteristic W_a and, thus, W_p. Reference signal x′ for the adaptive filter is derived from the reference noise signal x by filtering signal x with the transfer characteristic S_hat. The filtering may be performed in the time or spectral domain using discrete convolution (conv) or complex multiplication. If filtering is performed in the spectral domain, a coefficient corresponding to the transfer characteristic S_hat at frequency fm of signal x is to be used instead and, accordingly, is to be input. The reference noise signal x is input into the (adaptive) filter 2 which compensates for deviations from the actual secondary path 5 having transfer characteristic S, i.e., reference noise signal x is adapted to be the negative of signal d. Signal y′_a is the “real” analog cancelling signal (also referred to as ANC output signal) at the position of the microphone 6.
Referring now to
The system of
y′_p=−Mic_w·d_hat.
Alternatively or additionally to weighting of the passive branch, the active branch, in particular the adaptive filter 2, may be weighted by, e.g., multiplying the copied filter coefficients of the filter 10 with the weighting coefficient(s) Lsp_w, so that
y′_a˜Lsp_w·y′_p.
Provided the transfer characteristic S_hat is an exact model (estimation) of the secondary path transfer characteristic S and the system is in a steady state and has reached a certain degree of adaptation, the weighting coefficients Lsp_w and Mic_w may be selected according to the following considerations:
For the above considerations (1a and 1b), the following conditions ideally are assumed:
Lsp_w=1
a=e_a/d=(d+y′_a)/d≈(d+y′_p)/d
d_hat≈d
d′_hat=Mic_w·d_hat
y′_d≈−d′_hat
a≈(d−Mic_w·d)/d=1−Mic_w.
For the above considerations (2a and 2b), the following conditions ideally are assumed:
Mic_w=1
a=e_a/d=(d+y′_a)/d≈(d+Lsp_w·y′_p)/d
d_hat≈d
d′_hat=Mic_w·d−hat
y′13 p≈−d′_hat
a≈(d−Lsp_w·d)/d=1−Lsp_w.
A major advantage of the system described above with reference to
Referring now to
y′_p=−(d′_hat+d′_ext).
Assuming that Lsp_w=1, the signal y′_p as defined above will be part of the signals y′_a and e_a. Thus, any (e.g., harmonic) signal desired by the listener can be added to the noise. The filter 17 is used to alter the signal d′_ext respective of amplitude and phase, if desired. As can be seen, the additional, external signal d′_ext does not have any effect on disturbance signal d per se. Altering of the disturbance signal d is only performed by the ANC system independent of its system structure.
As shown in
The reference noise signal x is also supplied to the filters 41-46 having transfer characteristics S11, S12, S21, S22, S31, S32 and subsequent controllable filters 47-52 having transfer characteristics W_p_1, W——1, W_p_2, W_p_2, W_p_3, W_p_3. The controllable filters 47-52 are controlled by a filter controller 53 that receives six signals x′ from the filters 41-46 and two signals e_p_1, e_p_2 from adders 54, 55, respectively, to generate control signals for controlling the controllable filters 47-52. The adder 54 receives signal y′_p_1, signal d′_ext_1 and an output signal of the weighting element 39. The adder 55 receives signal y′_p_2, signal d′_ext_2 and an output signal of the weighting element 40. The signals y′_p_1, y′_p_2 are provided by adders 56, 57; the adder 56 receives via real part processors 58, 59, 60 the output signals of the filters 47, 49, 51 and the adder 57 receives via real part processors 61, 62, 63 the output signals of the filters 48, 50, 52. The signals d′_ext_1, d′_ext_2 are derived by filtering the signal x_ext from the external secondary noise source 16 with transfer characteristics −1·H_ext_1, −1·H_ext_2 of filters 64, 65 and taking the real parts thereof with real part processors 66, 67.
In
A modified multi-channel feedforward ANC system based on the system of
Deactivation of noise reduction to “0 dB” in the way described above using weighting coefficients does not mean that ANC is deactivated at the microphone or listening positions. There is still some control present because the system is forced to “0 dB”. When, for instance, an attenuation of “0 db” is desired at a particular microphone position, the ANC system in connection with all its loudspeakers seeks to maintain the instant noise signal d as it is, to the effect that the signals output by the loudspeakers are considered as noise by the ANC system at this point and a compromise has to be made in the ANC system's adaption procedure. Attenuation is desired for each of the remaining microphone signals, however, these signals exhibit a negative effect on the signal of the “0 dB” microphone. For the ANC system, this is a contradiction in itself and the state reached by the ANC system relies heavily on the loudspeaker microphone paths. In particular situations, it may be desirable to deactivate in terms of ANC one of the microphones 22, 23 in
A method of achieving this is to weight (multiply) the error signals e_p_1 and e_p_2 with the weighting coefficients Err_w_1 and Err_w_2 as can be seen in
W_p_1[n+1]=W_p_1[n]+μ·(x′11·e′_p_1+x′12·e′_p_2)
W_p_2[n+1]=W_p_2[n]+μ·(x′21·e′_p_1+x′22·e′_p_2)
W_p_3[n+1]=W_p_3[n]+μ·(x′31·e′_p_1+x′32·e′_p_2)
e′_p_1=Err_w_1·e_p_1
e′_p_2=Err_w_2·e_p_2.
With adequate determination of the weighting coefficients activation or deactivation of a particular microphone can be established to the effect that only a certain share of the respective microphone signal contributes to adaption. According to the above equations, all loudspeakers are affected by equal microphone weighting coefficients during adaption. For even more control options and flexibility, the system may be enhanced by additional weighting of the loudspeaker signals as shown in
W_p_1[n+1]=W_p_1[n]+μ·(x′11·e′_p_1+x′12·e′_p_2)
W_p_2[n+1]=W_p_2[n]+μ·(x′21·e′_p_1+x′22·e′_p_2)
W_p_3[n+1]=W_p_3[n]+μ·(x′31·e′_p_1+X′32·e′_p_2)
e′_p_1=Err_w_1·(e_p_11+e′_p_p_21+e′_p_31)
e′_p_2=Err_w_2·(e_p_21+e′_p_p_22+e′_p_32)
e′_p_11=Err_w_11·e_p_11 and so on.
The systems disclosed herein, in particular their signal processing units such as filters, adders, subtractors, weighting elements etc., may be realized in dedicated hardware and/or in programmable (digital) hardware such as microprocessors, signal processors, microcontrollers or the like, under adequate software-based control. Such a program, i.e., its instructions, may be stored in an adequate memory (or any other computer-readable medium) and are read out for controlling the microprocessor hardware or at least parts thereof to perform the function (method) of certain processing units (e.g., filter, adder, subtractor, weighting element) per se and in combination with other units.
Although various examples of realizing the invention have been disclosed, it will be apparent to those skilled in the art that various changes and modifications can be made which will achieve some of the advantages of the invention without departing from the spirit and scope of the invention. It will be obvious to those reasonably skilled in the art that other components performing the same functions may be suitably substituted. Such modifications to the inventive concept are intended to be covered by the appended claims.
Although the present invention has been illustrated and described with respect to several preferred embodiments thereof, various changes, omissions and additions to the form and detail thereof, may be made therein, without departing from the spirit and scope of the invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 11190092 | Nov 2011 | EP | regional |
| Number | Name | Date | Kind |
|---|---|---|---|
| 5848169 | Clark et al. | Dec 1998 | A |
| 6275592 | Vartiainen | Aug 2001 | B1 |
| 7062049 | Inoue et al. | Jun 2006 | B1 |
| 20010005814 | Dussac | Jun 2001 | A1 |
| 20050207585 | Christoph | Sep 2005 | A1 |
| 20080181422 | Christoph | Jul 2008 | A1 |
| 20100124337 | Wertz et al. | May 2010 | A1 |
| 20110235693 | Lee et al. | Sep 2011 | A1 |
| Number | Date | Country |
|---|---|---|
| H07104770 | Apr 1995 | JP |
| H11149292 | Jun 1999 | JP |
| Entry |
|---|
| Kuo et al., “Active Noise Control: A Tutorial Review”, Proceedings of the IEEE, vol. 87, No. 6, Jun. 1999. |
| Extended European Search Report for corresponding Application No. 111900924, mailed Aug. 29, 2012, 8 pages. |
| Japanese Office Action for corresponding Application No. 2012-255297, mailed Dec. 24, 2013, 7 pages. |
| Number | Date | Country | |
|---|---|---|---|
| 20130129108 A1 | May 2013 | US |
