System and method for adaptive control of online extraction of loudspeaker parameters

Abstract

In at least one embodiment, an audio system for extracting online parameters is provided. The system includes a speaker and at least one controller. The speaker transmits an audio signal in a listening environment. The at least one controller includes a signal processing block and an adaptive filter. The signal processing block is programmed to provide a drive signal u(n) to drive the speaker to transmit the audio signal. The adaptive filter is programmed to receive the drive signal and, in response to the speaker transmitting the audio signal, receive a first change signal i(n) from the speaker. The adaptive filter is further programmed to generate an admittance curve of the speaker based on at least the drive signal and the first change signal.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Application Serial No. 62 / 955,125, filed December 30, 2019, the disclosure of which is incorporated herein by reference in its entirety. This application may also be generally related to U.S. Provisional Application Serial No. 62 / 955,138, filed December 30, 2019, entitled “SYSTEM AND METHOD FOR PROVIDING ADVANCED LOUDSPEAKER PROTECTION WITH OVER-EXCURSION, FREQUENCY COMPENSATION AND NON-LINEAR CORRECTION” (“the '138 Application”); and U.S. Provisional Application Serial No. 62 / 955,138, filed December 30, 2019, entitled “SYSTEM AND METHOD FOR PROVIDING ADVANCED LOUDSPEAKER PROTECTION WITH OVER-EXCURSION, FREQUENCY COMPENSATION AND NON-LINEAR CORRECTION.” and U.S. Provisional Application Serial No. 62 / 955,149, filed on December 30, 2019, entitled “SYSTEM AND METHOD FOR PROVIDING A LINEARIZER FOR LOUDSPEAKER APPLICATIONS” (“the '141 Application”), the disclosures of which are hereby incorporated by reference in their entireties. Technical Field

[0003] One or more aspects disclosed herein generally relate to systems and methods for adaptive control of online extraction of loudspeaker parameters. These and other aspects will be discussed in more detail below. Background Art

[0004] Current loudspeaker implementations are based on previously measured loudspeaker parameters or functions (such as the loudspeaker's impedance curve), which have several drawbacks because loudspeaker parameters can, in principle, change with respect to reverberation volume (applied sound pressure level), time, temperature, aging, individual differences, etc. Consequently, any control algorithm, such as a (thermal) limiter, based on such loudspeaker-related parameters will not perform optimally. It would be highly desirable to introduce a system that can estimate those desired / required loudspeaker parameters in an automated, adaptive manner, and in real time. Summary of the Invention

[0005] In at least one embodiment, an audio system for extracting online parameters is provided. The system includes a speaker and at least one controller. The speaker transmits an audio signal in a listening environment. At least one controller includes a signal processing block and an adaptive filter. The signal processing block is programmed to provide a drive signal u(n) to drive the speaker to transmit the audio signal. The adaptive filter is programmed to receive the drive signal and, in response to the speaker transmitting the audio signal, receive a first change signal i(n) from the speaker. The adaptive filter is further programmed to generate an admittance curve of the speaker based on at least the drive signal and the first change signal.

[0006] In at least another embodiment, a computer program product embodied in a non-transitory computer-readable medium programmed to extract online parameters associated with a loudspeaker is provided. The computer program product includes instructions for providing a drive signal u(n) to drive the loudspeaker to transmit an audio signal; and receiving a change signal i(n) from the loudspeaker in response to the loudspeaker transmitting the audio signal. The computer program product includes instructions for generating one of an admittance curve or an impedance curve of the loudspeaker based at least on the drive signal and the change signal.

[0007] In at least another embodiment, a method for extracting online parameters associated with a loudspeaker is provided. The method includes providing a drive signal u(n) to drive the loudspeaker to transmit an audio signal and receiving a change signal i(n) from the loudspeaker in response to the loudspeaker transmitting the audio signal. The method also includes generating an admittance curve or an impedance curve of the loudspeaker based on at least the drive signal and the change signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] 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 will be best understood by referring to the following detailed description taken in conjunction with the accompanying drawings.

[0009] Figure 1 a first graph depicting a magnitude-frequency response of the loudspeaker's admittance and a second graph depicting its corresponding impulse response;

[0010] Figure 2 a first graph depicting a magnitude frequency response of the impedance of the loudspeaker and a second graph depicting a group delay frequency response of the impedance;

[0011] Figure 3 Describing a system for performing online extraction of speaker parameters according to one embodiment;

[0012] Figure 4A generally depicts a detailed implementation of an online parameter estimation block according to one embodiment;

[0013] Figure 4B Generally depicting a Figure 4A A more detailed implementation of the online parameter estimation block;

[0014] Figure 5 Describes an embodiment of the Figure 4B The adaptive filter used in the system;

[0015] Figure 6A generally depicts a high-level system for providing spectrum system identifiers and adaptive control according to one embodiment;

[0016] Figure 6B Generally depicting a Figure 6A Detailed implementation of the system;

[0017] Figure 6C Generally depicting a Figure 6B Detailed implementation of the system;

[0018] 7A to 7B depicting respectively a first graph in which windowing of the input block signal has been avoided and a second graph in which windowing of the input block signal has been applied;

[0019] Figure 8 generally depicts an example weighting function based on an impedance curve of a loudspeaker;

[0020] Figure 9 generally depicts another implementation of an online parameter estimation block that provides an estimate of total harmonic distortion (THD) in the spectral domain according to one embodiment;

[0021] Figure 10 generally depicts another implementation of an online parameter estimation block that provides estimation of a nonlinear fingerprint (NLF) in the spectral domain according to one embodiment;

[0022] Figure 11 generally depicts a three-dimensional graph of total harmonic distortion (THD) according to one embodiment;

[0023] Figure 12 generally depicts a system for determining an equalization filter for a spectral compressor according to one embodiment;

[0024] Figure 13 generally depicting a first graph of the magnitude frequency response of the raw and smoothed impedance curves and a second graph of the magnitude frequency response of the raw and smoothed impedance curves, respectively, in addition to corresponding NFP and THD functions;

[0025] Figure 14 Generally depicts the corresponding EQ filter based on the underlying NFP implemented as a Finite Impulse Response (FIR) (Org) and approximated by efficient Linear Predictive Coding (LPC);

[0026] Figure 15 generally depicts a system having a spectrum compressor according to one embodiment;

[0027] 16A to 16C generally depicting a spectrum diagram of a loudspeaker with different settings of a spectrum compressor;

[0028] Figure 17 generally depicts a system for providing a current-based feedback linearizer according to one embodiment;

[0029] Figure 18 various graphs generally depicting a magnitude frequency response, a phase frequency response, a sensitivity function, and a smoothed sensitivity function for a feedback filter according to one embodiment;

[0030] Figure 19 Roughly depicts the approximation of a loudspeaker's admittance curve by biquad and warped (FIR) filters;

[0031] Figure 20 Roughly depicts the overall quality (difference) of the approximation of the loudspeaker's admittance curve by biquad and warped (FIR) filters;

[0032] Figures 21A to 21B generally describing real-time test examples / results of the functionality of a current-based feedback linearizer with the linearizer turned off and with the linearizer turned on, respectively;

[0033] Figure 22 generally depicts a system combining a current-based feedback linearizer with an advanced system for protecting a loudspeaker from thermal and overexcursion overload, according to one embodiment; and

[0034] Figure 23 An overall system providing advanced loudspeaker protection against thermal and over-excursion overload and an adaptive spectral compressor as a feedback-controlled based linearizer according to one embodiment is generally depicted. DETAILED DESCRIPTION

[0035] As required, detailed embodiments of the present invention are disclosed herein; however, it should be understood that the disclosed embodiments are merely exemplary of the invention, which may be embodied in various alternative forms. The drawings are not necessarily drawn to scale; some features may be exaggerated or minimized to illustrate details of particular components. Therefore, the specific structural and functional details disclosed herein should not be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to employ the present invention in various ways.

[0036] It should be appreciated that the controllers / devices disclosed herein and in the appendix hereto may include any number of microprocessors, integrated circuits, memory devices (e.g., flash memory, random access memory (RAM), read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or other suitable variants thereof), and software that cooperate to perform the operations disclosed herein. In addition, such controllers disclosed utilize 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 the functions disclosed. In addition, the controllers provided herein include a housing and various numbers of microprocessors, integrated circuits, and memory devices (e.g., flash memory, random access memory (RAM), read-only memory (ROM), electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM)) located within the housing. The disclosed controllers also include hardware-based inputs and outputs for receiving data from and transmitting data to, respectively, other hardware-based devices discussed herein. Although the various systems, blocks, and / or flow diagrams described herein refer to the time domain, frequency domain, etc., it should be appreciated that such systems, blocks, and / or flow diagrams may be implemented in any one or more of the time domain, frequency domain, etc.

[0037] The aspects described herein may provide for, but are not limited to, optimization of speakers via electronic processing. The term "optimization" may be understood in many different ways. For example, optimization may correspond to using an existing speaker to get the most out of it (e.g., power) without causing the speaker to malfunction or be damaged. Optimization may also correspond to meeting a specific goal, such as keeping the speaker as small or as light as possible, as provided within a customer's specifications (e.g., minimum power). This approach may be of interest for all sorts of different applications, such as speakers in cars (e.g., making the overall weight as light as possible without compromising sound quality and / or sound pressure level (SPL)), and particularly for devices using small speakers (e.g., smartphones, laptops, smart speakers, sound bars, etc.).

[0038] Now, having established the need for such a system, it is possible to imagine implementing such an optimizer and considering the constraints. A natural place to start is to explore options for protecting the loudspeaker from malfunction or damage. Various methods of protecting the loudspeaker may include utilizing one or more of the following:

[0039] Limiters for protection: may cause acoustic artifacts and fail to exploit the physical potential of the loudspeaker, e.g. at low frequencies;

[0040] Multi-band limiter (MBL): By dividing the input signal into sub-bands and applying a (protection) limiter to each sub-band in a specific way to reflect the characteristics of a given loudspeaker, better performance can be achieved (especially) at low frequencies (bass). However, the MBL must be adjusted / tuned, which may require some effort.

[0041] Thermal protection: For example, based on speaker-specific tuning protection against short-term and long-term thermal damage, the measured impedance curve can be used to estimate the speaker's current and long-term power consumption. However, tuning may have to be performed (e.g., manually). In addition, the speaker's impedance may change over time due to heat and due to the volume / structure to which it is actually coupled (e.g., if installed in a vehicle door).

[0042] Current solutions may be insufficient or may require extensive tuning efforts to achieve more from a given loudspeaker. Even well-tuned, systems that may be based on a single measurement or condition may not be optimal, as various loudspeaker properties may change under certain conditions. As mentioned above, such conditions may not be anticipated. Online measurements may be required to continuously update certain loudspeaker parameters to enable adaptive control of these protection elements.

[0043] The embodiments as described herein may be based on the implementation of an online parameter extraction implementation that may be used in conjunction with the '138 application as described above.

[0044] 1. Online parameter extraction

[0045] There are several possibilities for extracting desired parameters (small-signal or large-signal parameters) from an unknown loudspeaker. For example, Klippel provides a measurement system, such as a laser, for measuring the deflection of a loudspeaker's diaphragm, from which various loudspeaker-specific parameters can be extracted. In this regard, systems based on acoustic measurements using microphones and / or accelerometers mounted very close to or directly on the loudspeaker (e.g., the diaphragm) are also known. In addition, current-voltage measurements of the current driving the loudspeaker and its corresponding current may be another way to measure certain loudspeaker parameters.

[0046] To implement the overexcursion limiter and the thermal limiter, the necessary loudspeaker parameters can be obtained from the current-impedance curve and its admittance curve, respectively. This can be easily obtained by measuring the actual drive signals (voltage and current) of the loudspeaker, which is probably the simplest and most cost-effective of all the possibilities mentioned.

[0047] Despite the fact that the impedance curve is probably the most predominantly used curve, it has been found that parameter extraction based on the adaptively estimated admittance curve may be more effective (see Figure 1 ). Figure 1 A first graph 100 depicts a magnitude frequency response for an admittance curve of a loudspeaker and a second graph 102 for its corresponding impulse response. Figure 2 A first graph 104 of a magnitude frequency response of the impedance of a loudspeaker and a second graph 106 of a group delay frequency response of the impedance are depicted.

[0048] refer to Figure 1 (See the first graph 100), it can be seen that the admittance curve has a wider spectrum characteristic with a small but deep notch at the resonant frequency of the speaker. Figure 2 The corresponding impedance curve shown (see first graph 104 ) naturally has a peak at the resonant frequency. This can be similar to the shape of a peaking filter. This spectral characteristic makes it easier to estimate the admittance curve, especially when using an adaptive FIR filter of finite length to directly estimate the impedance curve. This is possible, but on the one hand, a longer FIR filter may be required. On the other hand, however, due to the lower energy of the desired signal, a longer lead time may be required for convergence. In addition, the latter may also lead to a higher sensitivity to noise in the measured signal and, therefore, a less robust estimate of the curve.

[0049] exist Figure 2 In the corresponding impedance curve in the first graph 104, some desired loudspeaker parameters are shown, such as the currently estimated resonant frequency (f res ), the loudspeaker parameters can be extracted from its group delay, which is Figure 2 By examining the second graph 106, it can be appreciated that, in addition to the resistance at the resonant frequency (R res ), mechanical system quality (Q ms ), electrical system quality (Q ES ) and the total (complete) system quality (Q TS ) (e.g., online estimation of parameters), the group delay frequency response also includes a clearer and therefore easier to detect peak. Figure 1Other loudspeaker parameters shown in may include the DC resistance, the corresponding frequency at which the DC resistance is taken, and the estimated inductance of the loudspeaker.

[0050] The inductance can be estimated by the slew rate of the impedance curve. For example, between the frequency point where the DC resistance is obtained (i.e., the second zero) and a second spectral point at a higher frequency, it can be defined by 2 to 10 times the given resonant frequency, so that in this case, the reactance of the loudspeaker is guaranteed to dominate its resistance. Alternatively, if the resistance at a higher frequency where the inductive reactance dominates is used, the inductance can also be estimated. Therefore, the DC resistance can be subtracted from its (absolute) value, and the difference can then be used to calculate the inductance, as shown below (for example, inductance estimation):

[0051]

[0052] Using this set of extracted parameters, the aforementioned over-excursion limiter and thermal limiter can be operated, wherein the thermal limiter may also require an actual current signal for correct operation.

[0053] Figure 3 A system 150 for performing online extraction of speaker parameters is depicted, according to one embodiment. System 150 generally includes at least one controller 152 (hereinafter referred to as "controller 152") and at least one speaker 154 (hereinafter referred to as "speaker 154"). Although not shown, it should be appreciated that controller 152 may be operatively coupled to any number of memory devices storing instructions that enable controller 152 to perform any number of the operations described herein. Controller 152 is configured to transmit an audio signal from an audio source 156 to speaker 154 for playback of the audio data in a listening environment 158. System 150 is configured to prevent speaker 154 from overexcursion, in which the cone (not shown) of speaker 154 may travel too far along a first axis 160. This condition can minimize distortion and artifacts in the audio played in the listening environment. Similarly, system 150 can also prevent speaker 154 from experiencing overheating conditions. This aspect can improve the quality of audio playback in listening environment 158.

[0054] The controller 152 includes a signal processing block 170 (e.g., a single gain stage), an online parameter estimation block 172, a thermal model gain estimation block 174, an over-excursion limiter gain calculation block 176, and a speaker control and protection block 178. In general, the over-excursion limiter gain calculation block 176 receives a signal x 最大 , which corresponds to the maximum allowed excursion of the loudspeaker 154. The over-excursion limiter gain calculation block 176 is responsive to the signal x from the online parameter estimation block 172. 最大and the signal PARAMETER to generate an over-excursion limiter gain signal (eg, a gain OEL ). It will be appreciated that, as described in the '0138 application, any one or more of the adaptively extracted parameters on the signal PARAMETER (e.g., Rdc, fres, Res, Qts, impedance, etc.) can be provided to one or more audio amplifiers to limit the excursion of the voice coil of the loudspeaker and to limit the temperature of the loudspeaker. The various extracted parameters on the signal PARAMETER transmitted from the online parameter estimation block 172 are discussed in more detail below.

[0055] The thermal model gain estimation block 174 receives a signal τ corresponding to the maximum allowable operating temperature of the speaker 154 最大 and the change signal (i(t)) output by the speaker 154 (eg, via the current sensor ( Figure 3 (not shown) The measured current signal is output by the speaker 154. Signal signal x 最大 and τ 最大 The value of may be stored in a memory (not shown) of the controller 152 and may be provided via a data table of the speaker 154. The thermal model gain estimation block 174 is responsive to the signal τ 最大 , the current change signal i(t) and the DC resistance value of the voice coil (not shown) of the speaker 154 (eg, R DC ) to generate a thermal limiter gain signal (eg, gain TM , to keep the speaker 154 within the maximum allowable temperature range τ 最大 Over-excursion limiter gain signal (e.g., gain OEL ) generally corresponds to a control signal indicating an amount of deflection that the cone of the speaker 154 may travel along the first axis 160 without experiencing deflection. The thermal limiter gain signal (eg, gain TM ) generally corresponds to a control signal that indicates the thermal limit at which the speaker 154 will operate. The speaker control and protection block 178 is responsive to the over-excursion limiter gain signal gain OEL , and the thermal limiter gain signal Gain TM to generate a gain signal (e.g., Gain), which is transmitted to the signal processing block 170. The signal processing block 170 transmits a signal u(t) (or drive signal) corresponding to the varying input voltage signal provided to the speaker 154 in response to the gain signal from the speaker power and control block 178. The varying input voltage signal u(t) controls the speaker 146 to travel to a maximum linear position x on the axis 160. 最大 (For example, the speaker 154 will not travel beyond a maximum position x 最大 ), and can also control the speaker 154 to operate within an operating temperature range (eg, up to a maximum temperature τ 最大), so as not to exceed the given maximum value τ 最大 In addition to controlling the excursion and power consumption of speaker 154, signal processing block 170 (or controller 150) can also control the varying input voltage signal u(t) to generally control the volume (or SPL) of speaker 154, which can directly affect the temperature of the voice coil of speaker 154. Thus, controller 152, along with signal u(t), can prevent short-term overexcursion of the voice coil and long-term overtemperature. These aspects can prevent damage to speaker 154.

[0056] Adaptive filter

[0057] Figure 4A A high-level system is generally depicted according to one embodiment, comprising a controller 152, and a Figure 3 The online parameter estimation block 172 includes at least one adaptive filter 190 (hereinafter referred to as "adaptive filter 190") and a small signal estimation block 192. The adaptive filter 190 is generally configured to estimate the admittance of the speaker 154 (i.e., the inverse of the desired impedance curve), thereby determining the desired parameters. The adaptive filter 190 receives a control signal u(t) and a varying current signal i(t) at the speaker 154 to generate a signal g(n). The signal g(n) generally corresponds to the desired impedance of the speaker 154, either directly or through a transformation (e.g., inversion). By analyzing the impedance of the speaker 154 (e.g., its amplitude-frequency response, its group delay frequency response of the speaker 154, its impulse response, etc.), the controller 150 can determine the parameters of the speaker 154 (e.g., Rdc, fres, Res, Qts, impedance, etc.).

[0058] Figure 4B Describes the detailed implementation of the online parameter estimation block 172 of the controller 152, which includes the adaptive filter 190 and the small signal estimation block 192. Figure 4B As shown, the online parameter estimation block 172 can be implemented in the subband domain. Specifically, according to one embodiment, the online parameter estimation block 172 includes an adaptive filter 190 in the subband domain (or frequency domain). The online parameter estimation block 172 includes an input block 200, a first fast Fourier transform (FFT) block 202, a calculation power block 204, an inverse FFT (IFFT) block 206, a first frame block 208, a second frame block 210, a second FFT block 212, and an adder 214.

[0059] The first FFT block 202 converts the input signal of the speaker 154 from the time domain to the sub-band domain (or frequency domain) (ie, u(z) or U(e) jΩ,n)), the input signal is provided as an input to the calculation power block 202 and the adaptive filter 190. The calculation power block 204 calculates the power of the signal u(z) transmitted to the adaptive filter 190. In general, the least mean square (LMS) algorithm can be used to control the adaptive filter 190. Thus, the adaptive step size of the adaptive filter 190 can be normalized by the power of the signal u(z). The second FFT block 210 is configured to convert the error signal e(n) from the time domain to the subband domain (or frequency domain (i.e., e(z) or E(e)) j′Ω ,n)), which is provided as an input to the adaptive filter 190. The error signal e(n) corresponds to the difference between the output of the adaptive filter 190 and the time-varying current signal i(n) from the speaker 154. For example, the adaptive filter 190 provides a signal g(n) in the time domain, which is fed to the small signal estimation block 192. Similarly, the adaptive filter 190 generates a signal d(z) (or D(e j′Ω ,n)). The signal d(z) roughly corresponds to the given / measured current signal i est (n). The first frame block 208 represents the output frame signal, whereby only the second half comprises valid signals / values ​​(e.g., if a 50% frame shift is applied). Similarly, the second frame block 210 also represents the output frame block, where the first half is padded with zeros to avoid interfering by-products of the circular convolution if a 50% frame shift is applied. The IFFT block 206 converts the signal d(z) (or i est (z), which corresponds to the estimated current output from speaker 154) is converted from the frequency domain to the time domain as signal d(n). Adder 214 subtracts the estimated desired signal d(n) from the varying current signal i(n) of speaker 154 to generate an error signal e(n). Generally speaking, adaptive filter 190 can be implemented as a multi-rate signal processing framework.

[0060] Figure 5 Depicts a detailed implementation of an adaptive filter 190 implemented in the frequency domain (FD) according to one embodiment. The adaptive filter 190 may be part of the controller 152 and include a complex conjugate block 220, a first multiplier circuit 222, a second multiplier circuit 224, a divider circuit 226, an adder circuit 228, and a third multiplier circuit 230. The filter 190 may utilize least mean squares (LMS), recursive least squares (RLS), or any other suitable updating scheme. In general, in conjunction with Figure 5The adaptive filter 190 shown illustrates how a new set of filter coefficients G(z) can be calculated over time. A complex conjugate block 220, a first multiplier circuit 222, a second multiplier circuit 224, a divider circuit 226, an adder circuit 228, and a third multiplier circuit 230 are formed to simulate an equation that provides a signal d(z) corresponding to a given / measured current signal i(n).

[0061] Adaptive filter 190 depicts a normalized LMS ("NLMS") based adaptive filter, which provides a high degree of flexibility, for example to implement certain constraints and / or control tasks. Additionally, adaptive filter 190 may represent an efficient method (at least in terms of processing power consumption) for implementing general system identification.

[0062] Unlike other system identification tasks (such as those known from microphone-based systems, such as acoustic echo cancellers (AECs)), embodiments herein may not require demanding adaptive adaptation steps represented by the current signal i(n) (or i(t) in the time domain) as the desired signal, and may not include unintended interference (other than sensor noise) as would be the case with microphone signals (e.g., tapping the microphone, blowing into the microphone, speech signals from a near-end speaker, etc.). This approach may simplify the adaptive filter. Furthermore, a residual echo suppressor may not be needed to further reduce the error signal, since the current set of filter coefficients representing the linear portion of the estimated admittance curve may be of interest.

[0063] Adaptive control

[0064] Figure 6A A high-level system 350 is generally depicted on the controller 152 for providing a spectrum system identifier and adaptive control to the online parameter estimation block 172 according to one embodiment. The online parameter estimation block 172 includes the adaptive filter 190, the small signal estimation block 192, and the adaptive control block 352. The adaptive control block 352 controls the adaptive filter 190 to obtain an estimate of the admittance g(n) when the following conditions are met:

[0065] (i) the drive signal u(t) (or u(n)) exceeds a certain minimum level (e.g., the power level of the drive signal u(t) or u(n) exceeds a predetermined minimum level), which is typically set to at least a few [dB] (e.g., 1 to 6 [dB]) above a given (current) sensor noise; and

[0066] (ii) The input signal spectrum (e.g., the spectrum of the varying drive signal u(n) arriving at the speaker 154) contains sufficient energy at or near the resonant frequency of the speaker 154, otherwise there is a risk that the adaptive filter 190 will operate but will not deliver a valid curve at and near the resonant frequency of the speaker 154 (e.g., by using a narrowband signal such as a sine wave, whose frequency is set to be independent of the resonant frequency of the speaker 154). This aspect may lead to invalid parameter extraction, which should be avoided. It may be necessary to determine whether the drive signal u(n) contains sufficient energy at the resonant frequency of the speaker 154, because such a signal may not be able to successfully perform adaptation if it does not have sufficient energy (i.e., if the signal-to-noise ratio (SNR) is too low) (e.g., see (i) above). This condition specifically takes into account the portion of the spectrum required to extract / estimate the desired small signal parameters (e.g., Rdc, fres, Res, Qts, impedance, etc.) at and near the resonant frequency of the speaker 154. For at least this reason, adaptation is allowed if (a) sufficient energy is present and even provided, (b) the minimum amount of energy possible is present at and near the estimated resonant frequency of the speaker 154 .

[0067] Adaptive control block 352 is configured to transmit a flag signal (i.e., a flag) that is set to zero or one. If conditions (i) and (ii) are met, adaptive control block 352 sets the flag signal to one. If the flag signal is set to one, the filter coefficients of adaptive filter 190 are adapted. The flag signal (if set to one) may indicate whether a new set of parameters (e.g., Rdc, fres, Res, Qts, impedance, etc.) is to be determined and used, or whether the previously estimated set of parameters should be used instead. If the flag signal is set to one, adaptive filter 190 is adapted to generate a new signal for g(n) that corresponds generally directly or through a transformation (e.g., inversion) to the desired impedance of speaker 154, as described above. By analyzing the impedance of speaker 154 (e.g., its amplitude-frequency response, its group delay-frequency response, its impulse response, etc.), controller 150 may determine the new parameters (e.g., Rdc, fres, Res, Qts, impedance, etc.) of speaker 154. In this case, the small signal estimation block 192 extracts the parameter R from the new signal g(n) generated by the adaptive filter 190 dc 、f res 、R es , Q ts , impedance, etc.

[0068] If the flag signal is set to zero, the adaptive filter 190 may be deactivated. In this case, the system 350 delivers a previously determined set of parameters based on the previously adapted admittance curve and the speaker parameters extracted from such curve. Thus, in this case, the adaptive control may be used as a fail-safe mechanism. Generally speaking, the flag condition controls the adaptation of the adaptive filter 190, which indicates whether the currently available signal g(n) from the adaptive filter 190 is valid (i.e., whether the currently available signal g(n) can be used for current parameter extraction). If the adaptive filter 190 cannot be adapted based on the flag signal (e.g., the flag signal is set to zero), the small signal estimation block 354 does not update the parameters (i.e., the previously calculated parameters remain frozen and / or based on an older, previous signal of g(n)).

[0069] Figure 6B Another implementation of a system 350 for providing a spectral system identifier and adaptive control to the online parameter estimation block 172 of the controller 152 according to one embodiment is generally depicted. The system 350 includes an adaptive filter 190, a small signal estimation block 192, an adaptive control block 352, and a computational weighting block 354. The computational weighting block 354 is configured to provide a weighting function to emphasize regions at and near the resonant frequency of the loudspeaker 154 to ultimately allow adaptation of the filter 190, even in the presence of narrowband signals. If the adaptive control block 352 is already controlling the adaptive filter 190, the system 350 may apply weighting to modify or adjust the input signal i(n) and the signal u(n) from the loudspeaker 154 at least once.

[0070] Figure 6C Generally depicting a Figure 6B The detailed implementation of the system 350 is shown in FIG. The system 350 includes an input block 200, a first FFT block 202, a calculation power block 204, an IFFT block 206, a first block 208, a second block 210, a second FFT block 212, and an adder 214, as shown above. Figure 4B The operation of these features has been described above.

[0071] The system 350 also includes an adaptive control block 352 and a computational weighting block 354. The adaptive control block 352 includes a first determination block 400 that provides a flag signal. The computational weighting block 354 includes a windowing block 402, an FFT block 404, an absolute value block 406, a first smoothing block 408, a first average block 410, a weighting block 412, a second average block 414, a second smoothing block 416, a spectrum limiting block 418, a limiting block 420, and a normalization block 422, a threshold block 424, and a threshold calculation block 426. The windowing block 402 receives an input signal u(n) generated from the signal processing block 170 to a speaker. The windowing block 402 applies a windowing function (e.g., Von-Hann (or Hann window)) to u(n) to avoid a picket fence effect. For example, Figure 7A The picket fence effect of signal u(n) on waveform 403 is generally depicted. Figure 7B Removal of the picket fence effect on the signal u(n) 403 is shown.

[0072] Without the windowing block 402, the signal levels of the test tone frequencies appear higher than they actually are due to the picket fence effect (see, for example, Figure 7A ), but for the applied window (e.g., Von-Hann), this negative effect disappears and the lobe at the frequency becomes wider (see, e.g., Figure 7B ). However, this situation may not adversely affect the system 350. The FFT block 404 converts the drive signal u(n) from the time domain to the frequency domain. The absolute value block 406 takes the absolute value of the signal u(z) and then feeds it to the first smoothing block 408. The first smoothing block 408 performs nonlinear smoothing from high frequency to low frequency (e.g., "up / down") and nonlinear smoothing from low frequency to high frequency (e.g., "down / up"). In other words, the first smoothing block 408 performs two smoothing operations. The smoothing is typically performed in parallel. Generally speaking, when the absolute value block 406 takes the absolute value of the signal u(z), the first smoothing block 408 performs nonlinear smoothing on the power spectral density (PSD) of the signal u(z). For example, by taking the absolute value of the complex spectrum of the signal u(z), this condition enables the first smoothing block 408 to perform nonlinear smoothing of the PSD.

[0073] The first mean block 410 obtains the mean of two smoothed versions of the signal u(z). In this case, the spectrum deviation of the nonlinear smoothed signal can be successfully avoided. For example, Figure 7B Waveform 403 in FIG. 4 also illustrates the spectral offset of a nonlinearly smoothed signal. In this case, it can be seen that the energy of a 100 Hz sinusoidal tone, for example, can be almost completely removed from the spectrum. Thus, system 350 can render adaptation insensitive to narrowband signals because adaptation can deliver effective values ​​at frequencies where the SNR is high enough to allow convergence, but which do not necessarily coincide with the resonant frequency of speaker 154.

[0074] Generally speaking, once the system 350 has been successfully adapted (e.g., the adaptive filter 190 is activated in response to the set flag signal, and the small signal estimation block 192 determines the new parameters (e.g., R dc 、f res 、R es , Q ts , impedance, etc.), weighting can be performed by the system 350. The weighting can be determined as follows: the threshold calculation block 426 receives the changing current signal i(n) from the speaker 154 and calculates the current error return loss enhancement signal (ERLE(n)). Despite the fact that Figure 6C The ERLE(n) in is determined in the time domain, but it is recognized that ERLE(n) can also / alternatively be calculated in the frequency domain. In this case, the error signal e(z), i.e., the corresponding E(e^jw,n) already available, and the frequency domain transformed version of the current time signal i(n) (i.e., i(z)) can also be used for this purpose. The current error return loss enhancement signal generally corresponds to the ratio between the desired current to be provided to the speaker 154 and the error. For example, the ratio between the current signal i(n) and the error signal e(n) serves as an indicator of how good the adaptive filter 190 has been covered, as represented by the most recent ERLE(n) measurement. If the current error return loss enhancement signal (ERLE(n)) exceeds the threshold ERLE TH , the threshold block 424 activates the normalization block 422 to utilize the currently existing impedance curve (eg, impedance) provided by the small signal estimation block 192 as a basis for determining a weighting function (Weight(n)).

[0075] To obtain a desired weighting amount from the impedance curve, the normalization block 422 may first obtain an absolute value of the impedance curve, and then the normalization block 422 may set a lower limit of the absolute value to a normalization value, ie, 0 dB.

[0076] The limit block 420 limits the normalized value to a tunable maximum value. Thereafter, the spectrum limit block 418 limits the spectrum region of the tunable maximum value to be below a certain tunable lower frequency and above a certain tunable upper frequency (f 最大) can be set to 1 (0 [dB]) (i.e., to a neutral value). The spectrum limiting block 418 ensures that spectral regions that would be overemphasized by the corresponding trajectory of the impedance curve used as a weighting function are avoided. Therefore, the purpose of the weighting is to emphasize the region at and near the resonant frequency of the loudspeaker 54 to ultimately allow adaptation via the adaptation block 352 and the adaptive filter 190 even in the presence of narrowband signals. This operation can be performed if the narrowband signal includes sufficient energy in the desired spectral region (e.g., at and near the resonant frequency of the loudspeaker 154), once the system 350 has been successfully adapted (e.g., if the current ERLE measurement value (ERLE(n)) exceeds a given threshold ERLE TH , this is the case), which is known.

[0077] Figure 8 An example weighting function generated based on the impedance curve of the speaker 154 is generally depicted. For example, Figure 8 Generally shown is an emphasis on the region at and near the resonant frequency of 100 Hz, which corresponds to the resonant frequency of the speaker 154. As shown, frequencies greater than 100 Hz are removed and not considered for weighting.

[0078] Return Reference Figure 6C , after activating the weighting block 412 to apply the weighting function (weight(n)), the second averaging block 412 obtains the average frequency to obtain a single energy value. For example, the average frequency can be obtained by a time domain infinite impulse response (IIR) smoothing filter (or second smoothing block 416) with a separately adjustable rise time constant τ 上 and individually adjustable fall time constant τ 下 The single energy value is continuously smoothed. Since the second averaging block 414 calculates the mean frequency, a single value that varies with time is retained, which is then smoothed by the smoothing filter 416. The attack time can typically be shortened to the decay time constant to avoid unnecessary freezing of the adaptation once a broadband signal with sufficient energy is present. Broadband signals with sufficient energy are generally Figure 7A and Figure 7B The first determination block 400 takes the smoothed energy value of the signal u(z) (e.g., as output from the second smoothing block 416 or (e.g., as Figure 6C The signal M)) shown is the same as that in Figure 7A and 7B The adjustable threshold level shown as 407 TH If the smoothed energy level of the signal u(z) is greater than the adjustable threshold level TH , the first determination block 400 sets the flag signal to one to activate the adaptive filter 190. As described above, if the flag signal is set to one, this condition indicates that the small signal estimation block 192 will determine new parameters (e.g., Rdc 、f res 、R es , Q ts and impedance). If the flag signal is set to zero (e.g., the smoothed energy level of the signal u(z) is less than the adjustable threshold level TH ), then this condition indicates that the previously determined parameters established by the small signal estimation block 192 are to be used. As described above, an optional weighting function may be performed prior to activating the adaptive filter 190. Thus, in this case, the weighting may be performed using the weighting block 412 in conjunction with the spectrum limiting block 418, the limiting block 420, the normalization block 422, the threshold block 424, and the threshold calculation block 426.

[0079] Returning to the threshold block 424, the current error return loss enhancement signal (ERLE(n)) is less than the threshold ERLE TH This condition corresponds to the first time the system 350 is started up and no previously stored admittance / impedance curve g(n) is available. Therefore, it can be assumed that the adaptive filter 190 (and the adaptive control block 352) are blind. Therefore, the system 350 has no information about the impedance (i.e., this condition also means that there is no estimate about the resonant frequency of the loudspeaker 154). In this case, the threshold block 424 sets the weighting function equal to one and initializes the weighting with a plurality of ones, which will not prevent the adaptive filter 190 from adapting. In general, the threshold block 424 may not be considered to be blind by setting the weighting function equal to one when ERLE(n) is less than ERLE TH Setting the signal FLAG signal (or setting the weight to 1) only indirectly affects the adaptive control (e.g., the adaptive filter 190) in the absence of a threshold value. This is necessary because the weight is not the only criterion that affects setting the signal FLAG (or setting the flag signal). Other or independent criteria may also be considered, such as the overall current SNR of the input signal (u(z)) examined or evaluated at the first determination block 400. This condition indicates that even if the weight is set to one via the threshold block 424, the current SNR of the input signal u(n) (i.e., the smoothed output from the second smoothing block 416 or (e.g., signal M)) is greater than or less than the level TH , the flag will still become one or zero.

[0080] Once the adaptive filter 190 has sufficiently estimated the system 350 after startup (e.g., after the weighting function has been set to one or initialized as described above), and the SNR of the input signal u(z) (e.g., the output of block 416) remains above the threshold level TH For a sufficient duration, the system 350 will operate as expected. For example, once the unknown system 350 is adequately estimated by the adaptive filter coefficient set g(n), then in this case, once the ERLE nThe measured value exceeds the given threshold ERLE TH , the currently estimated admittance / impedance curve g(n) is used to generate a weighting function "weight(n)", which will affect the signal FLAG, thereby controlling the adaptation of the adaptive system. Generally, once a valid version of the admittance / impedance is estimated by g(n), the adaptation controlled by the signal FLAG will be performed, and if there is not enough SNR available at and near the resonant frequency of the loudspeaker 154, the valid set will not be destroyed over time, thereby preventing adaptation.

[0081] Enhanced robustness of parameter extraction

[0082] Once the system 350 employs adaptive control (ie, the adaptation is more or less fail-safe), it is possible to extract parameters from the coefficients representing the admittance of the adaptive filter 190 (eg, by performing an inversion of the impedance curve).

[0083] Generally speaking, as mentioned above, one of the loudspeaker parameters of interest is the resonant frequency f Res As mentioned above, the resonant frequency f Res It can be extracted neither from the admittance curve nor from the impedance curve (which is possible in general), but for robustness reasons it can be extracted from the group delay frequency response of the impedance curve using, for example, the Smith method for group delay frequency response calculation.

[0084] Another speaker parameter of interest may be the DC resistance R DC The online parameter estimation block 170 (or the small signal estimation block 192) can determine the DC resistance R DC This value can be extracted from the impedance curve by searching for a minimum below the resonant frequency. In some cases, this determination may be erroneous, primarily because the estimated curve does not represent the true trajectory, since, typically, the input audio signal may not contain enough energy in those very low spectral regions. To this end, the online parameter estimation block 170 (or small signal estimation block 192) can search for a second minimum in the impedance curve that resides at the resonant frequency of the loudspeaker 154. In this region, there may be enough energy to make a good estimate of the impedance curve.

[0085] Due to high sensor noise from the current sensor (not shown) measuring the varying current signal i(t) from the loudspeaker 154, the estimated admittance curve, and hence the impedance curve, can often appear very noisy. A typical reason for this is that the input signal (e.g., the input audio signal x(t) or any other typical playback signal) does not contain sufficient energy at higher frequencies, but the sensor noise (e.g., the current sensor noise) is nearly white. Therefore, the signal-to-noise ratio (SNR) in those higher frequency regions may not be optimal, which inevitably leads to interference in the adaptation. If the noise is too high, the impedance curve may also become too noisy, which can lead to erroneous parameter extraction because high peaks are often misinterpreted as the loudspeaker's resonant frequency due to the noisy trace. To safely avoid this misinterpretation, the admittance and impedance curves should be nonlinearly smoothed, for example using octave smoothing. In this case, the higher spectral regions are smoothed, while the lower spectral regions are gently smoothed. This can be desirable because the resonant frequency of the loudspeaker 154 is typically located at low frequencies, and therefore, the resonant frequency value of the loudspeaker 154 is not negatively affected by the smoothing.

[0086] Spectrum Compressor

[0087] As described above, certain parameters can be reliably extracted from an unknown speaker (e.g., speaker 154) in an adaptive manner. Furthermore, such parameters can be securely protected from speaker 154, thereby improving the use of the speaker's physical capabilities. One aspect of the disclosed system can be to securely protect speaker 154. Optimization can be achieved by utilizing MBL.

[0088] One advantage of an MBL is that the MBL can limit different spectral regions separately, rather than in a broadband manner like conventional limiters and / or dynamic compressors. A benefit of dividing the spectrum into separate regions and limiting them individually may be that certain regions of the spectrum (usually given by their lower spectral portions) statistically tend to overdrive the loudspeaker 154 more often as mid-frequency or higher spectral portions. Thus, if the time signal exceeds a certain threshold, it may not be necessary to compress the full broadband signal, but rather to limit that portion of the input signal spectrum that would actually overdrive the loudspeaker 154. This may result in disturbing acoustic artifacts, which should be avoided. By properly tuning the MBL, the performance of the loudspeaker 154 may be optimized, as certain harmonic distortions may not produce disturbing acoustic artifacts from a subjective (psychoacoustic) point of view and are therefore allowed to remain in the output signal, which ultimately results in better performance at low frequencies. In practice, this aspect may enable things like Figure 9The method of estimating THD in an adaptive manner is shown. Moreover, this aspect can enable the loudspeaker 154 to sound better as if the loudspeaker 154 were operating purely at its linear limit. The automatic adjustment of the spectrum compressor to which the MBL belongs by taking into account psychoacoustic principles will be described in more detail below.

[0089] Estimation of nonlinear distortion

[0090] Figure 9 Another implementation of the online parameter estimation block 172 on the controller 152 is generally depicted, according to one embodiment, comprising a plurality of adaptive filters 190a to 190n in the spectral domain to provide an estimate of total harmonic distortion (THD). Figure 9 The online parameter estimation block 172 is shown to be generally similar to Figure 4B The online parameter estimation block 172 is shown. However, as Figure 9 The online parameter estimation block 172 is shown to include a plurality of stages 451a to 451n. The stages 451a to 451n include a plurality of first FFT blocks 202a to 202n, a plurality of computational power blocks 204a to 204n, a plurality of IFFT blocks 206a to 206n, a plurality of first blocks 208a to 208n, a plurality of second blocks 210a to 210n, a plurality of second FFT blocks 212a to 212b, and a plurality of adders 214a to 214n. Figure 9 The more general form of the online parameter estimation block 172 is generally shown, which is amplified by the estimated (spectral) THD. The estimated (spectral) THD is then used as input to calculate the spectrum compressor, which is not Figure 9 Generally speaking, the online parameter estimation block 172 may provide an estimate of the current nonlinear distortion, such as provided by a total harmonic distortion (THD) measurement, an intermodulation distortion (IMD) measurement (or a nonlinear fingerprint (NLF)), which includes all distortion of the loudspeaker 154, not just due to the harmonic portion, etc., which is then used as input to the spectral compressor calculations.

[0091] The online parameter estimation block 172 also includes a calculation scaling block 452, a calculation harmonics block 454, a THD estimation block 456, and a plurality of time-varying gain values ​​458a to 458n. The gain values ​​458a to 458n may reflect a special / simplified form of the filter (e.g., gain values) that may vary over time. The online parameter estimation block 172 may increase the signal processing workload (e.g., machine instructions per second (MIPS) and memory consumption) because a separate adaptive filter stage 190 may be required for each desired higher harmonic. Even though the second and third harmonics (K2 and K3) may be considered (which may be the most dominant harmonics of the speaker 154), the workload may be increased by at least three times when compared to, for example, a conventional adaptive filter or the first harmonic used to estimate the speaker.

[0092] The online parameter estimation block 172 can determine the THD of the speaker 154 in the following manner. The computational scaling block 452 (which can be optional) can scale the drive signal (or incoming audio signal) u(n) and then feed it to the filter 458a. As described above, the computational scaling block 452 is optional. If block 452 is not implemented, the gain values ​​458a to 458n are not needed. However, if scaling is applied, the gain values ​​458a-458n are necessary to correct for the scaling. Generally speaking, the computational scaling block 452 can improve performance and ensure that the system is stable for different types of input signals unknown to the system. The variable gain value 458a provides a filtered, scaled voltage of the signal u(n) to the computational harmonics block 454. The computational harmonics block 454 provides an output to each gain value 458b and 458n. Generally, the adaptive filters 190a to 190 are each similar to the adaptive filter 190 described above. However, due to the use of different reference signals (e.g., u1(z)-u2(n)), the gain values ​​458a to 458n are not required. n (z) as input, so there are adaptive filters 190a-190n respectively. The harmonic calculation block 454 generates reference signals u1(z)-u2 after converting the input signal u(n) into corresponding higher harmonic signals using trigonometric functions. n (z) to obtain the desired higher harmonic versions of u(n) (e.g., u2(w)=u(2*w), u3(w)=u(3*w), ..., un(w)=U(n*w)).

[0093] The THD estimation block 456 calculates the THD of the speaker 154 according to the following formula:

[0094]

[0095] In other words, the THD estimation block 456 divides the sum of the squared outputs from the adaptive filters 190b through 190n by the sum of the squared outputs from the adaptive filters 190a through 190n to provide a first value. The THD estimation block 456 takes the square root of the first value to provide the THD.

[0096] Figure 10

[0046] Another implementation of an online parameter estimation block 172 on the controller 152 that provides an estimate of the NLF in the spectral domain is generally depicted in accordance with one embodiment. The online parameter estimation block 172 is generally similar to the example shown in FIG. Figure 4B The online parameter estimation block 172 is shown. However, as Figure 9 The online parameter estimation block 172 is shown to include a single stage 451 and an NLF estimation block 470. The online parameter estimation block 172 can determine the NLF based on the drive signal u(n) and the varying current signal i(n) from the speaker, as the error signal of the adaptive filter 190 roughly estimates the sum of the linear portion of the speaker 154 and all nonlinear byproducts.

[0097] The NLF estimation block 470 calculates the NLF based on the following equation:

[0098]

[0099] In other words, the NLF estimation block 470 converts the squared error signal (e.g., E(e)) output by the FFT block 212 into JΩ ,n)) divided by the changing current signal from the speaker (e.g., I(e JΩ ,n)) to provide a first value. The NLF estimation block 470 takes the square root of the first value to provide the NLF. For example, the NLF estimation block 470 takes the square root of the ratio of the squared error signal to the squared current signal from the speaker 154 to obtain the NLF. In general, the NLF estimation block 470 may be based on the varying current signal i(n) and the spectral error signal E(e JW ,n) to calculate NLF. The formula indicates the calculation in the spectrum domain, although Figure 10 This is not shown in , but this requires that i(n) be converted to the spectral domain by the NLF estimation block 470 .

[0100] NLF can be interpreted as a spectrum-dependent distortion measure that provides a value between 0 and 1 (or between 0% and 100%). Typically, most nonlinear distortion can occur around the low frequencies and resonant frequencies of the speaker 154, such as in combination with Figure 11 Shown roughly. Figure 11 Generally showing a three-dimensional graph of spectrum-dependent THD over time based on measurements of a loudspeaker driven with pink noise brought to a gradually increasing volume over time. Figure 9 and Figure 10 With the features discussed, it can be seen that spectrally dependent nonlinearities, such as THD and / or NLF, of an unknown speaker (eg, speaker 154) can be continuously estimated.

[0101] Calculation of spectrum compressor equalization filter

[0102] After determining the THD and / or NLF of the speaker 154, which generally corresponds to the spectrally related nonlinearities of the speaker 154, aspects related to a spectral compressor, such as spectral weighting or an equalization (EQ) filter, may be determined. The following section discloses aspects related to the spectral compressor, which may correspond to spectral weighting of an equalization (EQ) filter. Figure 12 Generally depicts a system 500 (or spectrum compressor 500) on the controller 152, which can be used to generate a spectrum based on the signal h EQ (n), the error signal e(n) and the current signal i(n) to determine the desired EQ filter. eq (n) generally corresponds to the filter coefficients, regardless of whether an IIR or FIR filter is applied to the EQ filter 604, e.g. Figure 15 shown.

[0103] Reference again Figure 12 The spectrum compressor 500 includes first and second analysis window blocks 504a and 504b, first and second FFT blocks 506a and 506b, first and second absolute value blocks 508a and 508b, first and second multiplier blocks 510a and 510b, first and second smoothing blocks 512a and 512b, an NLF calculation block 514, a nonlinear smoothing block 516, a maximum search block 518, a first replacement block 520, a third multiplier block 522, a curve inversion block 524, a smoothing block 526, an optional HP filter 528, a limiting block 530, and a domain conversion block 532. In operation, the error signal e(n) is fed to the first analysis window block 504a, and the varying current signal i(n) from the speaker 154 is fed to the second analysis window block 504b. Each of the first analysis window block 504a and the second analysis window block 504a applies a window (e.g., a 300 ms long rectangular window) to the error signal e(n) and the current signal i(n), respectively. The first FFT block 506a and the second FFT block 506b convert the error signal e(n) and the current signal i(n) into frequency domain (or spectral domain) signals e(z) (or E(e JΩ ,n)) and i(z)(or I(e JΩ, n)). The first absolute value block 506a and the second absolute value block 506b take the absolute values ​​of the signals e(z) and i(z), respectively, and the first multiplier block 510a and the second multiplier block 510b square the signals e(z) and i(z) to calculate the power spectral density (PSD).

[0104] The first smoothing block 512a and the second smoothing block 512b may then smooth the signals e(z) and i(z) using, for example, infinite impulse response (IIR) smoothing filters applied from low frequency to high frequency to provide two smoothed spectra. and At this time, in a serial manner from high frequency to low frequency, spectrum deviation is avoided, and based on these two smoothed spectra and The NLF calculation block 514 considers small values ​​of Δ NLF The NLF of the speaker 154 is calculated to avoid division by zero. Thereafter, the nonlinear smoothing block 516 smoothes the NLF by using a nonlinear smoothing filter that passes the maximum value of the NLF (or max(NLF)) in a nonlinear smoothing form in a lower frequency spectrum range. The maximum value of the NLF is transmitted to the limiting block 520. The maximum value of the NLF may be closer to the resonant frequency (f res ) provides the maximum value.

[0105] Maximum search block 518 determines the maximum frequency (e.g., f 最大 ) and its corresponding amplitude value α 最大 The first replacement block 520 uses the value α 最大 From 0 to f 最大 The modified NLF signal is replaced by the third multiplier block 522 using a tuning parameter G. Thus, the smaller the tuning parameter G, the higher the bass that can be achieved, but the nonlinearity is still maintained, so potentially perceptible and annoying acoustic artifacts may also be present.

[0106] Curve inversion block 524 inverts the scaled NLF by subtracting the scaled NFL from one. At this point, the curve is at or near unity in regions of the spectrum where little nonlinear distortion occurs, and is at a neutral value less than unity at frequencies where speaker 154 exhibits non-negligible nonlinear distortion. This curve, representing the desired amplitude response of the first form of the EQ filter, can then be smoothed by smoothing block 526. Smoothing block 526 can be a conventional first-order IIR filter. Optional filter 528 can be a high-pass filter and can include an adjustable slope at low frequencies, which can be applied with a gradient of, for example, 6 [dB / octave]. This may reduce perceived bass performance, but at the same time, acoustic echo cancellation (AEC) performance may be improved by several decibels. Typically, HP filter 528 provides a slope of 0 [dB / octave] (e.g., a flat line, i.e., when HP filter 526 is inactive), which can be applied to maximize bass performance from speaker 154. However, for other applications, such as those requiring improved AEC performance, filter 528 provides an option for achieving this.

[0107] In the following, additional acoustic performance may be of interest and it may be desirable to enhance this aspect to the greatest extent possible. Therefore, in this case, filter 528 may be removed and a slope of 0 [dB / octave] applied. Next, the curve may be limited to an adjustable lower limit by a limit block 530 to prevent certain spectral regions from being severely degraded by the EQ filter. The lower limit imposed by the limit block 530 may be set to Δ NLF , but it will be appreciated that different tuning parameters may be used. Finally, the spectral EQ filter may be converted from the spectral or frequency domain to the time domain by a domain conversion block 532. Thus, different options are possible.

[0108] It is recognized that different options are possible. For example, one embodiment may include generating a conventional finite impulse response (FIR) filter of a certain length, such as by using a frequency sampling method to obtain a linear phase FIR filter or a more efficient minimum phase version of a linear phase FIR filter. In this regard, it should be noted that due to the fact that the EQ filter naturally reduces the level, especially at low frequencies, at and near the resonant frequency of the speaker 154. The FIR filter may have a certain minimum length; otherwise, the spectral resolution of the resulting FIR filter may inevitably be too coarse and therefore undesirable. Long FIR filters may be expensive to implement. Therefore, it may be preferable to use a linear predictive coding (LPC) filter to approximate the desired spectral trajectory of the EQ filter, which can be efficiently implemented by using a short FIR filter in the feedback loop. Another option may include implementing the desired EQ filter using an IIR filter, but estimating any desired trajectory using an IIR filter may be expensive. Testing has shown that the LPC version may be the most efficient in terms of filter length, but it is also the most efficient in terms of the computational effort of the LPC filter coefficients. Typically, any desired EQ filter can be implemented with half the coefficients of a minimum phase FIR filter and about a quarter the coefficients of a linear phase FIR filter. Some applications may require the use of a linear phase FIR filter, for example, if a time-varying spectrum compressor 500 (or control signal h) is also included. EQ The phase of the entire acoustic system (n)) is not allowed to change over time to avoid undesirable acoustic modifications, such as dynamic changes in localization, auditory source width, listener envelope, etc., which are all related to the overall phase and its stability over time. As an alternative, a constant phase (IIR) filter can also be used as a more efficient filtering version.

[0109] Figure 13 A first graph 550 and a second graph 560 are generally depicted. First graph 550 generally illustrates the magnitude responses of the original impedance curve 552 and the smoothed impedance curve 554. Second graph 560 generally illustrates waveform 556 corresponding to the NFP. As shown, waveform 556 exhibits a peak at 150 Hz due to the shape of the EQ filter. Waveforms 554 and 556 generally illustrate that, due to the effects of spectrum compressor 500, speaker 154 can play audio louder without disturbing distortion. Furthermore, waveforms 554 and 556 indicate the presence of more bass in the audio output.

[0110] Figure 14Graph 580 is generally depicted having a first waveform 582 corresponding to a THD function and a second waveform 584 showing an approximation of the magnitude frequency response for a 64-tap LPC FIR filter. The second waveform 584 shows that the limiting performed is sufficient to avoid artifacts. Graph 580 is an exemplary graph of a spectrum compressor.

[0111] Extension system with spectrum compressor

[0112] Figure 15 The overall system 600 for speaker optimization is depicted. The system 600 includes the controller 152, the speaker 154, the audio source 156, the online parameter estimation block 172, the over-excursion limiter gain calculation block 176, the THD estimation block 456 and the NLF estimation block 470 (e.g., see the distortion calculation block), and the spectrum compressor 500. The system 600 also includes a current sensor 602, an equalization filter 604, an adjustable gain block 606, and a control block (e.g., an adaptive filter control block (or a least mean square (LMS) control block)) 608, and an adder 610.

[0113] Generally speaking, the system 600 is composed of the desired parameters (R DC 、f DC 、f res , Q TS In addition to the thermal limiter (TL) driven by the online estimation of THD and L), advanced loudspeaker protection is also provided via the over-excursion limiter gain calculation block 176. As described above, the spectrum compressor 500 determines an estimate of the current nonlinear distortion of the loudspeaker 154 based generally on the THD and NLF from the distortion calculation blocks 456, 470, and outputs a signal h eq (n). The current nonlinear distortion of the speaker 154 includes the distortion of the speaker 154 not caused by harmonic components, etc. The signal h eq (n) corresponds to a real-time estimate of the current distortion of the speaker 154. The equalization filter 604 is configured to respond to the signal h eq (n) Consider the real-time distortion of the speaker 154.

[0114] Signal h EQ (n) provides a spectrum shape that varies with time n, which can be applied to the equalization filter 604. The equalization filter 604 is based on the signal h eq154. The equalization filter 604 is applied to the incoming audio signal x(t) from the audio source 156 before the over-excursion limiter gain calculation block 176 applies the gain G(n) to the adjustable gain block 606. The adjustable gain block 606 adjusts the gain output in response to the gain G(n) received from the over-excursion limiter gain calculation block 176. As described above, the over-excursion limiter gain calculation block 176 provides separate limiter gains for speaker over-excursion and for the temperature limiter of the speaker 154.

[0115] Generally speaking, the signal h eq The values ​​of u(n) and gain G(n) are adaptively changed, thereby modifying the loudspeaker drive signal (e.g., u(n)), and thus may directly affect the behavior and / or function of the loudspeaker 154 being tested and tuned in the closed loop. The analysis takes into account the real characteristics of the loudspeaker 154 (e.g., impedance). System 600 is implemented as a closed-loop hardware system because, in this case, the real physical loudspeaker can be part of the system 600 itself, or by using an accurate model of the loudspeaker being used, which is capable of simulating the loudspeaker in its complex form (e.g., also taking into account the nonlinear behavior of the loudspeaker 154 within the model). The preferred closed-loop hardware version may require hardware that can be connected to a simulation system operating on a personal computer (PC), thereby taking into account certain minimum delay requirements. Such an implementation may be expensive. To alleviate this problem, another approach can be adopted, such as using a detailed loudspeaker model that operates directly in the simulation instead of using the model to test the closed loop of the system. For example, the loudspeaker 154 is first measured via a Klippel measurement system to obtain loudspeaker parameters to model its complex behavior. These parameters are then used in a general loudspeaker model to simulate the behavior of the measured loudspeaker. The control block 608 generally specifies or serves as an adaptive control (e.g., LMS) of the adaptive filter 190. The adaptive filter 190 provides a signal g(n) (or g(z)) (e.g., admittance or impedance) that is used by the online parameter estimation block 192 to determine the above parameters. The adaptive filter 190 also provides a signal i est (t), which corresponds to the estimated signal (or estimated variation signal i(t)) output from the speaker 154. The adder 610 subtracts the signal i(t) from i(t). est (t) to provide the desired error signal e(t) necessary for the calculation of the estimated distortion (e.g., distortion calculation blocks 456, 470). The spectrum compressor 500 uses the distortion to generate the signal h eq (n).

[0116] Thus, the functionality of spectral compressor 500 can be verified, for example, by comparing the NLF before and after applying spectral compressor 500. If spectral compressor 500 is activated, a reduction in nonlinear behavior can be observed compared to when spectral compressor 500 is not activated. However, a desirable verification method may be to listen to the output file, because, compared to classical approaches to avoiding acoustically disturbing artifacts (e.g., by using corresponding HP crossover filters), with spectral compressor 500 activated, more bass can be perceived without perceptually disturbing acoustic artifacts once spectral compressor 500 is well tuned. It can also be verified by analyzing the output signal (e.g., see the signal in FIG. 16 , which shows the signal radiated by loudspeaker 154 (i.e., perceived by the listener)) that, with a tuned spectral compressor 500, some harmonic distortion may still remain in the spectrum of the output signal, despite the fact that it is not perceptually disturbing. Harmonic distortion that remains below the main spectral peaks will be successfully masked. This may determine that the spectrum compressor 500 is able to enhance the bass performance of the speaker 154 to achieve a maximum bass performance that exceeds the physical limits of the speaker 154 but still remains below a psychoacoustically acceptable limit.

[0117] Figure 16A 1 and 2 generally depict the spectrum of the speaker 154 when the spectrum compressor 500 is not used. Figure 16A In particular, it can be seen that between the formats of the speech signal (i.e. the audio output signal), there are other signals, which are due to the nonlinearities generated by the heavy bass (e.g. high energy content at (very) low frequencies), which may also be present in the signal.

[0118] Figure 16B Generally depicts the effect of the spectrum compressor 500 when it is activated and conservatively tuned (e.g., by using a frequency below f 最大 A biquad HP filter (e.g., combined with Figure 12 Spectrum diagram of speaker 154 when HP filter 528 is shown. Figure 16B The spectral region (horizontal spectral lines) generally showing between the formats of the speech signal is much less contaminated by the nonlinearity of the loudspeaker 154. This may be because the energy at the low frequencies has been reduced due to the HP filter 528. Consequently, bothersome acoustic artifacts may no longer be perceived, and bass performance has also been reduced.

[0119] Figure 16C 5. The spectrum of the speaker 154 is generally depicted when the spectral compressor 500 is activated and the HP filter 528 is deactivated. As shown, some of the nonlinear byproducts between the formants of the speech reappear but are less pronounced, just as when the spectral compressor 500 is not activated. On the other hand, compared to Figure 16B The bass content (or bass performance) has been improved compared to the conservatively tuned case shown. Also, in this case, no acoustic artifacts are perceived despite the fact that the bass performance is improved compared to the conservatively tuned case.

[0120] Linearizer

[0121] While spectrum compressor 500 can reduce certain spectral regions where nonlinear distortion becomes excessive, ultimately limiting the overall nonlinear distortion of loudspeaker 154 to a certain threshold, it may not represent a so-called "linearizer." If the drive signal of loudspeaker 154 is pre-distorted to compensate for the nonlinear distortion, ultimately linearizing the loudspeaker, the functional principle of a classic linearizer can be demonstrated.

[0122] This task can be achieved, for example, if the unknown loudspeaker 154 can be accurately modeled, including its nonlinear behavior. If such a model can be successfully estimated during normal operation, the predictable distortion of the loudspeaker 154 can also be estimated and thus also compensated for by a so-called image filter, thereby generating the aforementioned predistortion of the drive signal (e.g., u(t)) of the loudspeaker 154.

[0123] Figure 17 A system 700 for providing a current-based feedback linearizer, according to one embodiment, is generally depicted. System 700 includes controller 152, speaker 154, audio source 156, adaptive filter 190, current sensor 602, control block 608 (or adaptive filter controller), adder 610, another adder 702, and linearizer 704 (or feedback filter). System 700 can utilize linearizer 704 and the output of linearizer 704 as feedback signals. In operation, linearizer 704 receives an error signal e(n) from adaptive filter 190. Error signal e(n) generally provides information indicating the nonlinear portion of the time-varying admittance curve G(z), which also represents the sum of all nonlinear byproducts of speaker 154. It should be appreciated that G(z) represents the true system, which includes not only linear products but also the sum of all nonlinear byproducts. However, since a “normal” adaptive system can only estimate a linear system (LTI (Linear Time Invariant) system), it is obvious that the estimated current signal i output by the adaptive filter 190 is est (t) (e.g., the estimated current signal generated by the speaker 154) represents the linear portion (or linear product). Therefore, after subtracting the estimated value from the real current signal i(t) from the estimated value from the changing current i(t) via the adder 610, est After the estimation of (t), the resulting error signal e(t) represents the sum of all nonlinear byproducts. This signal is then used as the input to the linearization filter 704.

[0124] In this case, the linearizer 704 models the predictable distortion of the speaker 154 based on the error signal e(n) from the adaptive filter 190 and transmits the feedback control signal fb(t) to the adder 702, which subtracts the feedback control signal from the incoming audio signal x(t). Thus, the indication of the nonlinear byproducts of the speaker 154 via the signal fb(t) can be subtracted from the incoming audio signal x(t) to compensate for the nonlinear byproducts of the speaker 154.

[0125] Another way to implement such a linearizer is by means of feedback (FB) control, known for example from feedback active noise control (ANC) systems, such as Figure 15 Basically, the feedback loop is driven by the error signal e(n) of the adaptive filter 190, which estimates the linear part of the admittance curve / transfer function G(z) over time and also represents the sum of all nonlinear byproducts of the loudspeaker 154.

[0126] Example Design of Filter W(z) for Feedback Control Linearizer

[0127] Figure 18 Graphs 720, 722, 724, and 726 are generally shown, depicting the amplitude-frequency response, phase-frequency response, sensitivity function, and full (smoothed) sensitivity function, respectively, related to linearizer 704. Waveform 730, as shown in graphs 720 and 722, represents the underlying admittance function G(z) corresponding to the linear system. Waveform 732 represents the Bode plot of linearizer 704. Waveform 734 represents the open-loop system, given by HOL(z) = G(z) * W(z), and waveform 736 represents the limits, specifically the desired amplitude and phase margins. Waveform 740 in graphs 724 and 726 represents the sensitivity function, and waveform 742 in graph 724 represents the adjusted error margin. Waveform 740 in graph 726 represents the smoothed sensitivity function, which, in principle, can illustrate the frequency-dependent achievable reduction in the nonlinear byproducts of loudspeaker 154 and also provides a measure of the degree to which loudspeaker 154 can be linearized in the corresponding spectral region.

[0128] The data provided in curves 720, 722, 724 and 726 may indicate promising results. For example, a real-time system may be provided to verify the functionality of the feedback-controlled linearizer 704. In addition, the real-time system may need to meet requirements regarding latency, otherwise, correct operation may not be possible. Ultimately, such a low-latency real-time system may include an evaluation board for, for example, a Sigma 50 digital signal processor (DSP) from Analog Devices (ADI), which can also be programmed with acceptable effort. The DSP may not provide sufficient signal processing power to achieve an online estimation of the admittance function G(z) with sufficient spectral resolution (i.e., with a sufficiently long FIR filter). Therefore, for the verification test, the adaptively adjusted estimate of G(z) is replaced by a fixed filter, thereby approximating the linear characteristics of the fixed filter. The approximation by a pair of biquad IIR filters (biquad) is studied (see Figure 19 750 in FIG), and the use of short warped FIR (WFIR) filters (see Figure 19 752 in FIG. Figure 19 The waveform 754 in FIG shows the original admittance function G(z). Both versions (see waveforms 750 and 752) are able to simulate a given linear portion of the admittance function G(z), which is about 60 [dB] at most, as shown in FIG. Figure 19 and Figure 20 As shown. Therefore, the spectral region of interest to be approximated is at or near the resonant frequency of the loudspeaker 154 used, since here, as has been shown before, the maximum distortion is likely to occur. It will be appreciated that a fixed filter (not shown) may be used instead of the adaptive filter 190 and the control block 608. Figure 17 , the control block 608 can be removed and a fixed filter can be used instead of the adaptive filter 190. In short, the implementation of the adaptive filter 190 corresponds to a linearizer, and the use of a fixed filter (e.g., approximating a reference admittance or impedance) allows the feedback system 700 to match the loudspeaker 154 used to this reference admittance or impedance, which can be interpreted as an automatic matching system.

[0129] exist Figure 20 In FIG, waveforms 760 and 762 depict the difference between the underlying / original admittance function G(z) and its approximation by 10 biquad and 16Tap WFIR filters, respectively. In general, Figure 20It is shown that both approximations achieve acceptable results. One aspect of this system that may be considered is that the actual shape of G(z) may vary or change slightly over time, which may negatively affect the verification results. Therefore, it may be preferable to use the current error signal e(n) to feed the linearizer 704 based on an actual estimate of G(z) to achieve the highest linearization effect, but it may also be the case that the linear part of G(z) or the expected admittance function G of the reference loudspeaker is used instead. Ref (z). The linearizer 704 may automatically move the current loudspeaker 154 to this target. The linearizer 704 may attempt to automatically emulate the properties of a desired reference loudspeaker (e.g., at least the properties defined by the (complex) admittance curve), which may be another useful possibility for using the linearizer 704 in a beneficial manner.

[0130] Figures 21A to 21B 1 and 2 generally depict examples of real-time testing of the functionality of the current-based feedback linearizer 704 with the linearizer 704 closed and opened, respectively. Figure 21B As shown, a first real-time test may show that this system works in principle, since a reduction of about 20 [dB] of the first pair of higher harmonics (i.e., K2 and K3) can be achieved by using a low-frequency sine wave as the input signal, in which case these nonlinearities are produced. The higher harmonics may still reside at frequencies where the delay is still within an acceptable range to allow correct function. This is possible because during the tests, only a sampling rate fs=48 [kHz] was used, corresponding to fs=96 [kHz], which is usually too low for correct feedback control, as mentioned above. Using a higher sampling rate will result in a lower overall delay, which may allow control over a higher frequency spectrum.

[0131] Figures 21A to 21B Waveforms 780 and 782 (eg, current and voltage) illustrate that the linearizer 704 may modify the drive signal (eg, u(t)) driving the speaker 154 (eg, see Figure 21B Waveform 780). As shown above Figure 21A As noted in FIG. 7 , in the case where the linearizer 704 is deactivated, the waveform 780 shows that the drive signal contains only the input signal (eg, having a frequency f test =33[Hz] sine wave). However, Figure 21B Waveform 780 in FIG. 1 shows that waveform 782 (eg, the speaker drive signal) now has a sine wave in addition to the still dominant input signal (f test =33[Hz] sine wave) also includes a large number of additional signal parts, which are approximately up to f~250[Hz]. These additional signal parts are generated based on the error signal e(n) and filtered by the linearizer 704, which can ultimately reduce the nonlinear distortion of the system. This can be seen in the inspection Figures 21A to 21B The waveform 782 is seen. Figure 21A A typical plot of the current for a nonlinear system is shown, since a non-negligible portion of the signal exists at higher harmonics, such as K2 and K3, but there is also some intermediate nonlinearity, e.g., residing between K2 and K3. After the linearizer 704 is turned on, the resulting loudspeaker drive signal is predistorted (e.g., see waveform 780), as described above, resulting in Figure 21B The linearization of the effective current signal 782 shown in FIG. can reduce the harmonics K2 and K3 and the non-harmonic part between them. Therefore, the linear part may not be affected because Figure 21A and Figure 21B The current may not change between the graphs shown.Thus, the approximation of the linear portion of the admittance curve of the loudspeaker 154 (eg, achieved by 10 biquads) and the portion described above work as expected.

[0132] Speaker Optimization System

[0133] Figure 22 Generally depicted is a system 800 that combines the current-based feedback linearizer 704 with the over-excursion limiter block 176 according to one embodiment. System 800 includes the elements / features described in detail above (e.g., controller 152, speaker 154, audio source 156, online parameter estimation block 172, over-excursion limiter block 176, current sensor 602, adjustable gain block 606, control block 608 (or adaptive filter control block 608), adders 610, 702, adaptive filter 190, etc.). The descriptions of these elements / features described above also apply to system 800.

[0134] Figure 23 Generally depicted is a system 900 that combines the over-excursion limiter block 176, the spectrum compressor 500, and the linearizer 704 according to one embodiment. System 900 includes the elements / features described in detail above (e.g., the controller 152, the speaker 154, the audio source 156, the online parameter estimation block 172, the over-excursion limiter block 176, the adaptive filter 190, the spectrum compressor 500, the current sensor 602, the adjustable gain block 606, the control block 608, the adders 610, 702, the linearizer 704, etc.). The description of these elements / features described above also applies to system 800.

[0135] System 900 generally provides optimal performance for speaker 154 without damaging it. It is recognized that current sensor 602, which can be easily integrated into an amplifier's integrated circuit, may or may not increase hardware (HW) costs. From the perspective of controller 152 and memory, executing the instructions for the various features described herein may require additional effort, as adaptive filtering, along with, for example, at least two additional filters, may be required to estimate the current admittance curve G(z) in real time. The filters may implement the spectral shaping filter of spectral compressor 500 and the IIR-based feedback filter W(z), such as linearizer 704. Thus, the actual core of adaptive filter 190 (e.g., FIR filter G(z)) can be implemented at a high sampling rate, similar to linearizer 704, to maintain low latency, while the actual adaptation can be implemented at a lower sampling rate, most efficiently in the spectral domain (if desired) using efficient block processing. Furthermore, overexcursion limiter block 176 (including thermal limiter) can deliver a gain G(n) that is achievable at high frequencies, but since a single gain can be used, the real-time effort required to perform this portion is negligible.

[0136] To keep the software-related work low when implemented with the controller described herein, several measures have been proposed, such as implementing the spectrum compressor 500 using an efficient LPC FIR filter of reduced length (e.g., 64 taps) instead of a linear and / or minimum phase FIR filter. Adaptive FIR filters can be implemented using block processing (e.g., most efficient in the spectral domain) and / or downsampling, and feedback filter W(z) (or linearizer 704) can be implemented using a (minimum phase) IIR filter implemented using a pair (<=10) of ordinary biquads. With these efficiency-enhancing measures, such a system can be implemented using any processor with conventional performance.

[0137] Although exemplary embodiments have been described above, it is not intended that these embodiments describe all possible forms of the present invention. In fact, the words used in the specification are descriptive rather than restrictive, and it should be understood that various changes may be made without departing from the spirit and scope of the present invention. In addition, the features of the various embodiments may be combined to form other embodiments of the present invention.

Claims

1. An audio system for extracting online parameters, the system comprising: a loudspeaker for transmitting an audio signal in a listening environment; as well as At least one controller, the at least one controller comprising: a signal processing block programmed to provide a driving signal to drive the speaker to transmit the audio signal; and An adaptive filter programmed to perform: receiving the driving signal; receiving a first change signal from the speaker in response to the speaker transmitting an audio signal; and generating an admittance curve of the loudspeaker based at least on the drive signal and the first variation signal; The audio system also includes a distortion calculation block programmed to generate a first signal corresponding to nonlinear distortion of the loudspeaker based on at least one of higher harmonics of the drive signal and an error signal, wherein the higher harmonics include second and third harmonics, wherein the error signal corresponds to a difference between an estimated first variation signal and the first variation signal measured from the loudspeaker, wherein the first signal corresponds to one of a total harmonic distortion of the loudspeaker and a nonlinear fingerprint of the loudspeaker.

2. The audio system of claim 1, wherein the at least one controller is further programmed to determine an impedance curve of the loudspeaker based on the admittance curve.

3. The audio system of claim 2, wherein the at least one controller is further programmed to determine at least the quality of the entire system based at least on a magnitude-frequency response of the admittance curve or the impedance curve.

4. The audio system of claim 2, wherein the at least one controller is further programmed to determine a direct current (DC) resistance of a voice coil of the loudspeaker based at least on the impedance curve or the admittance curve of the loudspeaker.

5. The audio system of claim 2, wherein the at least one controller is further programmed to determine a resonant frequency of the speaker based at least on a group delay frequency response of the admittance curve or the impedance curve.

6. The audio system of claim 2, wherein the at least one controller is further programmed to determine the inductance of the speaker based on the admittance curve or the impedance curve of the speaker.

7. The audio system of claim 1 , wherein the drive signal is a voltage variation signal and the first variation signal is a current variation signal, and wherein the adaptive filter is programmed to generate the admittance curve based on the voltage variation signal and the current variation signal.

8. The audio system of claim 1, wherein the drive signal is a current variation signal and the first variation signal is a voltage variation signal, and wherein the adaptive filter is programmed to generate an impedance curve based on the voltage variation signal and the current variation signal.

9. The audio system of claim 1 , further comprising an over-excursion limiter gain calculation block programmed to limit the travel of the voice coil of the loudspeaker based at least on the admittance curve or the impedance curve of the loudspeaker.

10. The audio system of claim 1, further comprising a thermal model gain calculation block programmed to limit a temperature of a speaker based on a resistance of a voice coil of the speaker and the first variation signal.

11. The audio system of claim 1 , further comprising a linearizer programmed to receive an error signal indicative of a sum of all nonlinear byproducts of the admittance curve or impedance curve of the loudspeaker.

12. The audio system of claim 1, wherein the nonlinear portion of the admittance curve or impedance curve of the loudspeaker represents the sum of all nonlinear byproducts of the loudspeaker.

13. The audio system of claim 1 , further comprising a spectral compressor programmed to generate filter coefficients for an equalization filter applied to an incoming audio signal based on the estimated nonlinear distortion of the loudspeaker and to cancel the distortion associated with the loudspeaker.

14. The audio system of claim 1 , further comprising an adaptive control block programmed to control the adaptive filter to generate the admittance curve or impedance curve of the loudspeaker in response to at least the drive signal exceeding a minimum power level.

15. The audio system of claim 14, wherein the adaptive control block is further programmed to control the adaptive filter to generate the admittance curve or impedance curve of the loudspeaker in response to the input spectrum of the drive signal including energy at or near a resonant frequency of the loudspeaker.

16. A computer program product embodied in a non-transitory computer-readable medium programmed to extract online parameters associated with a loudspeaker, the computer program product comprising instructions for: providing a driving signal to drive the speaker to transmit an audio signal; receiving a change signal from the speaker in response to the speaker transmitting an audio signal; generating one of an admittance curve or an impedance curve of the speaker based at least on the drive signal and the variation signal; as well as An adaptive filter is controlled to generate an admittance curve or an impedance curve for the loudspeaker in response to at least the drive signal exceeding a minimum power level and an input spectrum of the drive signal including energy at or near a resonant frequency of the loudspeaker.

17. A method for extracting online parameters associated with a loudspeaker, the method comprising: providing a driving signal to drive the speaker to transmit an audio signal; receiving a change signal from the speaker in response to the speaker transmitting an audio signal; generating an admittance curve or an impedance curve of the loudspeaker based at least on the drive signal and the variation signal; as well as An adaptive filter is controlled to generate an admittance curve or an impedance curve for the loudspeaker in response to at least the drive signal exceeding a minimum power level and an input spectrum of the drive signal including energy at or near a resonant frequency of the loudspeaker.

Images