Distortion sensing, distortion prevention, and distortion-aware bass enhancement

Abstract

Distortion sensing, distortion prevention, and distortion aware bass enhancement are disclosed. Systems and methods that provide distortion sensing, prevention, and / or distortion aware bass enhancement in an audio system can be implemented in a variety of applications. Sensing circuitry can generate statistics based on a received input signal for which an acoustic output is generated. In various embodiments, the statistics can be used such that a multi-notch filter can be used to provide an input to a speaker to generate the acoustic output. In various embodiments, the statistics from the sensing circuitry can be provided to a bass parameter controller coupled to a bass enhancement circuit to operatively provide a parameter to the bass enhancement circuit. The bass enhancement circuit can provide a bass enhanced signal for generation of the acoustic output based on the parameter. Various combinations of bass enhancement circuits and multi-notch filters using statistics from the sensing circuitry can be implemented to provide an enhanced acoustic output.

Description

[0001] This application is a divisional application of patent application No. 201780074419.3, filed on October 20, 2017, entitled "Distortion Sensing, Anti-Distortion, and Distortion Detection Bass Enhancement".

[0002] Related applications

[0003] This application claims priority to U.S. Provisional Application No. 62 / 411,415, filed October 21, 2016, pursuant to 35U.SC119(e), which is incorporated herein by reference in its entirety. Technical Field

[0004] This invention generally relates to apparatus and methods for processing audio signals. Background Technology

[0005] Over the years, the physical size of consumer devices has decreased significantly, a trend that is inconsistent with ideal acoustic environments. This reduction in device size necessitates the use of microspeakers, whose design differs from their typical loudspeaker counterparts. For example... Figure 1 As shown, a significant drawback of miniature speakers is their reduced low-frequency response, which necessitates extensive flattening equalization (EQ) for more acceptable audio quality. Figure 1 Here is an example of the frequency response of a miniature loudspeaker, where the loudspeaker resonance (f0) is 845 Hz and the quality factor (Qtc) of the resonance in the closed box is 30.

[0006] The shape of such an equalization curve below 2kHz typically contains at least three components: high-pass, low-frequency enhancement, and peak resonance attenuation, such as... Figure 2 As shown. Figure 2 The example shown is the frequency response curve 139 of the resulting miniature loudspeaker, which has been flattened using a flattened EQ curve 137 that includes high-pass, low-shelf, and peak filtering. While static compensated equalization can significantly improve subjective and objective performance at low or even medium sound pressure levels (SPL), flattened equalization can cause distortion at higher listening levels due to excessive loudspeaker offset. For this reason, several solutions on the market offer dynamic flattened equalization algorithms guided by loudspeaker offset models. These solutions essentially attempt to apply flattened equalization as much as possible without causing excessive offset of the diaphragm or diaphragm within the loudspeaker. When these solutions are measured using common synthetic log-sweep sine (LSS) signals, still tones, or even noise, these systems appear to work quite well, but the measured total harmonic distortion (THD) is high.

[0007] Alternatively, it has been observed in several portable devices using miniature loudspeakers that multi-tonal signals with only moderate amounts of energy under loudspeaker resonance can exhibit significant distortion, often referred to as “friction and hum.” However, it is unclear whether the distortion is caused by excessive offset within the loudspeaker and enclosure or by other acoustic / mechanical issues. Determining this requires measuring the offset using a loudspeaker enclosure modified with a transparent panel exposing the internal loudspeaker and a laser hysterometer to observe the loudspeaker offset during distortion events. Furthermore, low-frequency enhancement in typical loudspeaker flattening solutions exacerbates the problem because the offset model guiding the time-varying equalization may overlook factors contributing to distortion, resulting in over-engaging equalization curves. Attached Figure Description

[0008] Figure 1 This is an example of the frequency response of a miniature loudspeaker.

[0009] Figure 2 This is an example of the frequency response of a miniature loudspeaker that has been flattened using equalization, which includes high-pass filtering, low-shelf filtering, and peak filtering.

[0010] Figure 3 This is a block diagram of features of example architectures for providing dynamic distortion sensing and adaptive distortion reduction according to various embodiments.

[0011] Figure 4 This is a block diagram of the features of an example architecture for providing dynamic distortion sensing and adaptive multi-band distortion reduction according to various embodiments.

[0012] Figure 5 This is a block diagram depicting an example architecture for integrating a multi-notch filter with external audio hardware and signal processing components according to various embodiments.

[0013] Figure 6 This is a block diagram illustrating an example notch filter bank designer according to various embodiments, which configures the coefficients of each of a series of multiple notch band filters between the input signal and the output signal.

[0014] Figure 7 The diagram illustrates adjustments made according to various embodiments to determine the lowest frequency signal notch band that can be used to reduce distortion.

[0015] Figure 8 The diagram illustrates, according to various embodiments, how the number of notch bands used can be increased to further reduce distortion by using linear or octave frequency spacing.

[0016] Figure 9 The diagram illustrates how the effectiveness of each frequency band is weighted using a slope parameter, according to various embodiments.

[0017] Figure 10 The trade-off between distortion and timbre is illustrated according to various embodiments, where the parameters of the multi-notch filter are readjusted to reduce attenuation.

[0018] Figure 11 This is a schematic diagram illustrating an exemplary single-notch band filter implemented as a linear combination of an input signal and an all-pass response, according to various embodiments.

[0019] Figure 12A This is a block diagram of an example device according to various embodiments, which provides audio output to a speaker in response to a device output signal and uses sensing circuitry to analyze the device output signal.

[0020] Figure 12B This is a block diagram of an example device according to various embodiments, which provides audio output to a speaker in response to a device output signal and uses sensing circuitry to analyze the device output signal.

[0021] Figure 13 This is a block diagram of example dynamic gain control for adjusting notch depth using measured distortion sensing energy according to various embodiments.

[0022] Figure 14 This is a block diagram of a bass enhancement circuit according to various embodiments.

[0023] Figure 15 The diagram shows a system according to various embodiments that integrates dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction with distortion-aware harmonic bass enhancement for audio signals.

[0024] Figure 16 This is a block diagram of an example adaptive dynamic sensing and multi-notch module according to various embodiments.

[0025] Figure 17 This is a block diagram of an example adaptive dynamic sensing and multi-notch module according to various embodiments.

[0026] Figure 18 This is a block diagram of an example adaptive dynamic sensing and multi-notch module configured in a feedforward arrangement according to various embodiments.

[0027] Figure 19 This is a block diagram of an example adaptive dynamic sensing and multi-notch module configured for feedback arrangement according to various embodiments.

[0028] Figure 20This is a block diagram of an example adaptive dynamic sensing and multi-notch module according to various embodiments, which combines feedforward and feedback audio signals with feedback information derived from or equivalent to real-time voltage and / or current sensing information from external transducer amplifier circuitry to sense the likelihood of distortion, or the degree of offensive, perceptible or measurable distortion.

[0029] Figure 21 This is a flowchart illustrating the features of example methods for sensing and reducing distortion in sound output generated from a speaker, according to various embodiments.

[0030] Figure 22 This is a flowchart illustrating the features of example methods for sensing and reducing distortion in sound output generated from a speaker, according to various embodiments.

[0031] Figure 23 This is a flowchart illustrating the features of example methods for achieving distortion-aware harmonic bass enhancement of audio signals according to various embodiments.

[0032] Figure 24 This is a flowchart illustrating the features of example methods for implementing distortion sensing, distortion reduction, and distortion-aware harmonic bass enhancement of audio signals according to various embodiments.

[0033] Figure 25 This is a block diagram of an example audio system according to various embodiments, which is arranged to operate for distortion sensing and distortion reduction in order to sense distortion in sound output and reduce distortion in sound output. Detailed Implementation

[0034] In the following description of embodiments of the audio system and method, reference is made to the accompanying drawings. These drawings illustrate, by way of illustration and not limitation, specific examples of how various embodiments can be implemented. These embodiments are described in sufficient detail to enable those skilled in the art to implement these and other embodiments. Other embodiments may be utilized, and structural, logical, electrical, and mechanical changes may be made to these embodiments. The various embodiments are not necessarily mutually exclusive, as some embodiments may be combined with one or more other embodiments to form new embodiments. Therefore, the following detailed description should not be considered limiting.

[0035] In various embodiments, dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction can be achieved for audio signals. The device can be configured to sense the quantitative degree of empirical nonlinear distortion produced by a loudspeaker for a given sound input and reduce the distortion using a set of notch or band-stop filters. A loudspeaker is any electroacoustic output transducer intended for audio listening. This includes, but is not limited to, loudspeakers, miniature loudspeakers, handsets, headset drivers, headphone drivers, or headphones. Such a device can use multiple features obtained from the audio or a range of other possible audio features. Statistics can be calculated based on the energy from the bandpass filter at discrete frequencies or bands. The notch or band-stop filter banks can be arranged in cascade or parallel.

[0036] Dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction can be applied to improve the sound quality generated by loudspeakers. Distortion-triggered spectral components of the signal used for audio output can be reduced or removed, keeping the number of removed components to a minimum. Simultaneously, different sets of features or spectral components can be used to sense distortion; the analyzed components or features do not need to directly correspond to the reduced / removed components.

[0037] In various embodiments, the system and the method of using the system may include receiving a signal for audio output and processing the received signal using a processing module including at least one notch filter. A module is one or more tools that perform a specific function, wherein the tools may include hardware, a processor executing instructions stored in a memory device, firmware, logic circuitry, or a combination thereof. Parameters of the notch filter may be determined based on sensing to detect the likelihood of distortion in the audio output. The output of the notch filter may be fed to a device, process, or apparatus that contains observable audio distortion. The device, process, or apparatus may be acoustic or digital, or any device capable of outputting audio. Furthermore, at least one parameter of at least one notch filter may be determined by measuring the likelihood of distortion at the output or the degree of perceptible, objectionable, or measurable distortion at the output using sensors, measurements, statistics, or a combination thereof.

[0038] In various embodiments, distortion-aware harmonic bass enhancement can be implemented for audio signals. The quantitative degree of sensing empirical nonlinear distortion can be combined with flexible harmonic bass enhancement algorithms. Adaptive harmonic bass enhancement modules can be used in a variety of applications. Enhancement can be avoided when the enhanced harmonics would cause unpleasant distortion. The module can employ tunable asymmetric nonlinearity, as well as pre-cutoff and post-cutoff filtering, to vary the amount of harmonic energy added to the signal.

[0039] In various embodiments, the system and the method of using the system may include receiving a signal for sound output and processing the received signal using a processing module that performs bass enhancement. The execution of bass enhancement may include, but is not limited to, bass enhancement based on linear filtering, bass enhancement based on the insertion and / or generation of additional harmonics of the signal carrying audio content (also known as psychoacoustic bass enhancement), or combinations thereof. Parameters of the bass enhancement may be determined based on sensing to detect the likelihood of distortion. The output of the bass enhancement module may be fed to a device, process, or apparatus that contains observable audio distortion. The device, process, or apparatus may be acoustic or digital or any device capable of outputting audio. Furthermore, at least one parameter of the bass enhancement module may be determined by a sensor, measurement, statistics, or a combination thereof that measures the likelihood of distortion at the output. Bass enhancement may be provided to maintain bass and impact intensity in the output audio in the presence of distortion.

[0040] Embodiments of dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction can operate independently of embodiments of distortion-aware harmonic bass enhancement. Embodiments of distortion-aware harmonic bass enhancement can operate independently of embodiments of dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction. In various embodiments, dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction can be combined with distortion-aware harmonic bass enhancement to best avoid distortion from bass enhancement.

[0041] This paper describes a processing algorithm for reducing the aforementioned distortions, including friction and hum distortion, by reducing energy below and associated with a critical harmonic interval of the microspeaker's resonant frequency. The algorithm may include a multi-notch filter applied to the speaker input signal, multi-tone signal analysis for measuring the amount of energy contributing to distortion, and a dynamic gain unit for controlling the time-varying depth of the notch filters. Parameters of the notch filters, such as the notch frequency, can be configured offline via dynamic control of the notch depth performed in real time. In signal processing, real-time refers to completing some signal / data processing within a time sufficient to keep up with external processes, such as generating a sound output from a serially processed input audio signal.

[0042] Figure 3This is a block diagram of an embodiment of example system 200, which includes a speaker 230 that generates sound output in response to a signal from an audio signal source 205. The audio signal source may include not only audio from direct playback of a recorded or designed soundtrack or sound stream, but also audio signals from other processed (e.g., spatial processing) sources of the soundtrack or sound stream, or other sources used to generate the speaker's sound output. System 200 may include sensing circuitry 210, which may be coupled to receive the signal received by speaker 230. Sensing circuitry 210 generates statistics about the signal to modify the signal received by speaker 230. Speaker 230 may generate sound output based on the modified received signal in response to the statistics generated by sensing circuitry 210. Sensing circuitry 210 may be implemented by circuitry including processing devices and stored instructions to process the signal provided to sensing circuitry 210. Sensing circuitry 210 may be implemented in a digital signal processor.

[0043] System 200 or variations of system 300 may include a number of different embodiments that can be combined depending on the application of such a system and / or the architecture of a system that can implement multiple methods. Such a system may include sensing circuitry operable to generate statistics that measure the probability of distortion at the speaker output or measure the degree of offensive, perceptible, or measurable distortion at the speaker output. The sensing circuitry may be operable to compute a binary indicator of distortion at the speaker output using techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence. The techniques may use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, Naive Bayes classification, linear discriminant analysis, and random forests. The sensing circuitry may also be operable to compute a soft indicator corresponding to the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion using techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence. A soft indicator may also be referred to as a fuzzy indicator. A soft indicator is a value selected from one or more ranges of values. A hard indicator may be defined as a value selected from a set of discrete values.

[0044] System 200 or variations thereof may include sensing circuitry 210 comprising a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to speaker distortion. Sensing circuitry 210 may include circuitry for estimating the amount of distortion-activated energy in the filtered signal from the plurality of IIR or FIR bandpass filters and / or dynamic spectrum analysis filters as one or more statistics calculated from the energy across all frequency bands of the plurality of bandpass and / or dynamic spectrum analysis filters, and includes circuitry for measuring the total energy of the filtered signal. The one or more statistics may include one or more statistics from a set of statistics including the median, average, arithmetic mean, geometric mean, maximum energy value, minimum energy value, and the energy of the nth largest energy, where n is a chosen positive integer. Sensing circuitry 210 may be arranged to calculate the energy as time-varying energy or to calculate the envelope of a full-band audio signal.

[0045] Figure 4 This is a block diagram of an embodiment of an example architecture 300 for providing dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction. Architecture 300 can be incorporated into multiple different systems providing sound output. Architecture 300 may include a multi-notch filter 305, sensing circuitry 310, and dynamic gain unit 315 that can operate with a speaker 330 providing sound output. The speaker 330 may be implemented as a miniature speaker, but alternatively may be implemented as any electroacoustic output transducer intended for audio listening. This includes, but is not limited to, a speaker, a miniature speaker, a headset, an earphone driver, a headphone driver, or a headset. The multi-notch filter 305 may be arranged to receive a signal used to generate the sound output. Here, the signal used to generate the sound output is a signal representing physical sound and may be referred to as an "audio signal". The sensing circuitry 310 may be coupled to receive a filtered signal corresponding to the signal received and processed by the multi-notch filter 305 and to generate statistics about the filtered signal.

[0046] The sensing circuit 310 can be arranged in various ways. The sensing circuit 310 may include a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to distortion of the loudspeaker 330. The sensing circuit 310 may include circuitry for estimating the amount of distortion-activated energy in the filtered signal using one or more of various statistics calculated from energy across all frequency bands of the plurality of IIR or FIR bandpass filters and / or dynamic spectrum analysis filters, and includes circuitry for measuring the total energy of the filtered signal. The sensing circuit 310 is operable to generate statistics that measure the probability of distortion or the degree of perceptible, objectionable, or measurable distortion at the output of the loudspeaker 330. The sensing circuit 310 may be implemented using circuitry including processing means and stored instructions for processing the signal provided to the sensing circuit 310.

[0047] Sensing circuitry 310 can be configured to operate on using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence (AI) to calculate a measure of the probability of distortion or the degree of perceptible, objectionable, or measurable distortion at the speaker output. This technique can use one or more processes selected from the group consisting of classification and regression trees (aka decision trees), ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression. Artificial neural networks can include, but are not limited to, perceptrons, multilayer perceptrons, deep belief networks, and deep neural networks. Nonlinear regression can include, but is not limited to, multinomial regression.

[0048] Sensing circuitry 310 can be configured to operate to compute a binary indicator of distortion at the speaker output using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, or artificial intelligence. The binary indicator can provide a yes / no distortion indication. The technique can use one or more processes selected from the group consisting of classification and regression trees, support vector machines, artificial neural networks, logistic regression, Naive Bayes classification, linear discriminant analysis, and random forests. Support vector machines can include support vector classification. Artificial neural networks can include, but are not limited to, perceptrons, multilayer perceptrons, deep belief networks, and deep neural networks.

[0049] Sensing circuit 310 can be configured to compute a binary indicator by derivation from a soft indicator using a threshold or comparator device. Alternatively, a semi-binary or semi-hard indicator can be derived from the soft indicator using a threshold, generating discrete indicators for input values ​​above / below a specific threshold, and variable indicators for input values ​​below / above a certain threshold. A semi-hard indicator can be defined as a value selected from a combination of discrete values ​​and one or more ranges of values. For example, for a specific task, a semi-hard indicator could be a value selected from a combination of discrete values ​​0.0 and values ​​in the range (0.5, 1.0). The combination of one or more discrete values ​​with one or more ranges of values ​​depends on the application for which it generates the semi-hard value. A semi-binary indicator is a semi-hard indicator having a combination of values ​​selected from discrete values ​​0 and 1 and one or more ranges of values ​​between 0 and 1. In the case of deriving a binary or semi-binary indicator in this way, a hysteresis can be employed to avoid issuing a discrete indicator for only a brief offset of the soft indicator above a threshold.

[0050] The supervised learning paradigm, common in the field, can be used to train techniques selected from machine learning, statistical learning, predictive learning, and artificial intelligence (AI), as described in several features below. First, a representative set of audio recordings can be collected and compiled, including content from music, film, and speech (and other genres). Second, these recordings can be played back using an audio signal processing cascade detailed above, but with multiple notch filters bypassed. They can be played back at various volume or gain levels typical of real-world playback scenarios, with particular focus on maximum volume playback. During playback, the acoustic output of the loudspeakers (including miniature loudspeakers) can be recorded in a quiet or anechoic environment using a calibrated measurement microphone at a fixed distance from the loudspeaker.

[0051] Third, each measurement record can be analyzed with reference to the original content record to identify and label the presence, absence, or degree of perceptible, objectionable, or measurable distortion in portions of the measurement record. To characterize portions of perceptible or objectionable distortion, hearing tests can be performed on human subjects, or a computational psychoacoustic model of distortion can be employed, chosen from several such models common in the field, in which the measurement record and the original content record are compared subjectively or digitally. To characterize portions of measurable distortion, a level-matching gain can be applied to the original content record such that its low-level portions (chosen to be sufficiently low to be assumed undistorted) match the levels of the corresponding measurement record. Furthermore, the level-matched original content record can be time-aligned with the corresponding measurement record. The time-varying amplitude of the measurable distortion can be calculated as the envelope or time-varying RMS energy of the difference between the aligned original content record and the corresponding measurement record.

[0052] Fourth, the labeled portions of the measured audio recordings can be subdivided into training and test sets, where the balance between the undistorted and distorted portions is approximately equal in both sets. Fifth, multiple sets of audio features can be computed from consecutive frames or blocks of distorted and undistorted portions. Sixth, a classification or regression technique selected from common machine learning, statistical learning, predictive learning, and artificial intelligence (AI) techniques in the field can be trained on the training set, and the metric for regression or classification performance on the training set can be monitored as training progresses. Seventh, a metric for regression performance can be computed based on the predictions of the trained model computed on audio features from the test set. Steps 4 through 7 are known as cross-validation, an established paradigm in the fields of machine learning, statistical learning, predictive learning, and artificial intelligence (AI). These steps can be iterated to improve classification or regression performance.

[0053] The dynamic gain unit 315 can be coupled to receive statistics from the sensing circuit 310 and coupled to the multi-notch filter 305 to operatively modify the depth of the multi-notch filter 305. The loudspeaker 330 can be arranged to receive the filtered signal to generate a sound output. The dynamic gain unit 315 can be arranged to provide gain based on subsequent statistics to adjust the depth of the multi-notch filter, these subsequent statistics being based on the aforementioned bandpass energy-based statistics and total energy, wherein the bandpass energy-based statistics and the measured total energy are provided by the sensing circuit 310.

[0054] Figure 5 This is a block diagram of an embodiment of example architecture 400, depicting the integration of a multi-notch filter 405 with external audio hardware and signal processing components. Architecture 400 includes an audio channel source 401 that feeds spatial, loudness, and dynamics processing 402, the output of which is input to a transducer equalizer 403. The output of the transducer equalizer 403 is input to the multi-notch filter 405, whose output is directed to a loudspeaker 430 via a limiter 407. The loudspeaker 430 may be implemented as a miniature loudspeaker. The limiter 407 allows signals below a specified input threshold level to pass unaffected, while attenuating the peak values ​​of stronger signals above that threshold. In addition to the output of the limiter 407 being coupled to the loudspeaker 430, the output of the limiter 407 may also be coupled to a sensing circuit 410. The sensing circuit 410 may use multiple features obtained from the audio signal or a range of other possible audio signal features to provide statistics to a dynamic gain unit 415. Based on statistics or other information about the audio signal that causes the sound output, the dynamic gain unit 415 can provide input to the multi-notch filter 405 to operatively modify the depth of the multi-annotation filter 405. The multi-notch filter 405, the sensing circuit 410, and the dynamic gain unit 415 can be coupled with… Figure 4The multi-notch filter 305, sensing circuit 310, and dynamic gain unit 315 are implemented in a similar or identical manner.

[0055] When driving a miniature loudspeaker with certain multi-tone signals, friction and hum distortion may be activated. Using a tuned static multi-notch filter, such as... Figure 6 As shown, the energy that activates distortion can be reduced to below the speaker distortion activation threshold. Figure 6 This is a block diagram illustrating an example embodiment of a notch filter bank designer 506, which configures the coefficients of each of a series of N notch band filters 505-1, 505-2…505-N between the input signal 501 and the output signal 508. The number of notches used and the configuration of each notch can be determined from a set of design controls.

[0056] There are essentially two goals in designing multi-notch filters: eliminating distortion and minimizing the impact on the perceived timbre of the sound. Timbre is a characteristic quality of sound, independent of pitch and loudness, from which its source or manner of generation can be inferred. Typically, best practice is to begin tuning with a test content aimed at eliminating distortion, known to be using a static multi-notch filter to activate the distortion of interest.

[0057] Multiple parameters can be used in tuning, with each speaker channel having its own set of parameters. These parameters can include the number of notch bands, fundamental frequency, reference Q, reference notch depth, reference slope, bandwidth scale, and frequency scale. The number of notch bands is the total number of notch bands. The fundamental frequency is the cutoff frequency of the lowest frequency. The reference Q is the bandwidth of the lowest frequency notch. The reference notch depth (dB) is the attenuation applied to the lowest frequency notch band when fully engaged. The reference slope (dB / band) is the amount gradually added to the notch depth of each additional band. Regarding the bandwidth scale, starting from the first low-frequency band, the additional bands can maintain a constant Q factor or a constant bandwidth. Regarding the frequency scale, the notch frequency can maintain a linear scaling relative to the fundamental frequency (e.g., [100, 200, 300] Hz) or an octave scaling (e.g., [100, 200, 400] Hz).

[0058] The first step in tuning may involve determining the fundamental frequency by adjusting, for example, a single notch band with a relatively low reference Q of approximately 4 and a depth of approximately -15 dB, and scanning the cutoff frequency from approximately 100 Hz up to frequencies where significant distortion is eliminated. Although heuristic methods are used to determine the parameters of the multi-notch filter, the optimal cutoff frequency tends to be near the speaker resonant frequency or has a harmonic relationship at the resonator resonant frequency. Figure 7As an example, given a miniature loudspeaker resonant frequency of 800Hz, the possible optimal fundamental frequency would be close to 100 or 200Hz, depending on the loudspeaker's available low-frequency response. Figure 7 A single notch band is shown, which was adjusted to determine the lowest frequency that can be used to reduce distortion. Modification of the reference Q parameter can help discover the fundamental frequency.

[0059] The optimal frequency scale and the number of notch bands can be determined. After establishing the fundamental frequency, additional frequencies above the fundamental frequency and below the speaker resonant frequency can be identified to reduce distortion. These frequencies should have a linear or octave scaling relative to the fundamental frequency, such as... Figure 8 As shown. Figure 8 The diagram shows that the number of notch bands can be increased to further reduce distortion using linear or octave frequency spacing. The number of notches indicated in the figure (three (3)) is not a limitation or constraint on the number of notches to that particular value; it can be arbitrary (e.g., ...). Figure 6 The variable N is shown. After establishing a set of target notch frequencies, the design parameters for the number of notch bands can be adjusted based on the total number of frequencies, the fundamental frequency based on the lowest target frequency, and the additional frequencies above the fundamental frequency. The reference notch depth, reference slope, and bandwidth scale can be adjusted. Within a given set of target frequencies, some notch bands are more effective than others in eliminating unwanted distortion. When the multi-notch filter is fully engaged, the notch depth and reference slope parameters determine the amount of attenuation at each target frequency. Generally, the notch depth decreases as the notch frequency increases. The reference slope parameter will increase or decrease the subsequent notch depth from the reference frequency. For example, in Figure 9 In this context, the reference notch depth is -15 dB, the reference slope is 3 dB, and the specific notch depths are [-15, -12, -9, ...]. Figure 9 The diagram illustrates the weighting of effectiveness for each frequency band using a slope parameter. Similarly, the number of notches is not limited or constrained to the number shown in the example in this figure. Positive notch depths can be prevented.

[0060] A trade-off can be struck between distortion removal and tone preservation. This step can be based on the current parameter settings checked to minimize the impact on subjective tone. For example, the reference Q should be adjusted to the highest possible setting (narrowest bandwidth) while still maintaining distortion protection. Figure 10 This demonstrates the trade-off between distortion and timbre when readjusting the parameters of multiple notch filters.

[0061] Such as Figure 4 305 multi-notch filter and Figure 4The multi-notch filter 405 can be implemented as a series of independent IIR notch filter bands. Each notch band can be achieved by using a linear combination of a signal and an all-pass filter set to the cutoff frequency and Q of that band. Using this method, the filter coefficients can be statically determined at initialization using the cutoff frequency and bandwidth parameters, and the time-varying depth of the notch can be efficiently adjusted, allowing the notch depth to be moved up and down in real time, such as... Figure 10 As shown by the double-headed arrow in the image.

[0062] Figure 11 This is a schematic diagram illustrating an example single-notch band filter 1005 implemented as a linear combination of input signal 1001 and all-pass response 1013. Notch depth gain can be provided to linear interpolation 1011, which provides a first factor such as (1 - gain) * .5 to multiply the input signal 1001 to multiplier 1014-1, and a second factor such as (1 + gain) * .5 to multiply the all-pass response 1013 to multiplier 1014-2. The multiplication results are provided to summer 1012 for linear combination, and an output signal 1008 is provided from summer 1012.

[0063] It is possible to construct, such as Figure 4 The sensing circuit 310 and Figure 5 The sensing module of the sensing circuit 410 is used to sense distortion to estimate the amount of distortion activation energy within the signal driving the loudspeaker. Therefore, ideally, there is no further processing of the loudspeaker signal downstream of the sensing module. Sensing analysis can then be used in either a feedforward or feedback architecture to inform the dynamic gain unit to finally determine the signal. Figure 4 and Figure 5 Independent depth within the multi-notch filter in the associated architecture. Sensing analysis can consist of multiple IIR or FIR bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies identified as contributing to friction and hum or other unpleasant distortion in a particular microspeaker. The bandwidth of each frequency band is ideally very narrow, so two or more cascaded second-order IIR or FIR analysis filters and / or dynamic spectrum analysis filters for each band can be used to achieve good signal energy measurement. Because the energy within a single frequency band does not activate distortion, the estimated amount of distortion-activating energy can be calculated as an aggregate statistic, such as the arithmetic or geometric mean of the energy across all frequency bands. This ensures that only multi-tone frequency band energy within a considerable number of frequency bands will result in a high estimate of the distortion-activating energy. Statistical methods other than arithmetic or geometric means can be used. The sensing module also measures the total energy of the input signal.

[0064] Figure 12AThis is a block diagram of an embodiment of example device 1100A, which provides sound output from speaker 1130A in response to device output signal 1101A and uses sensing circuit 1110A to analyze device output signal 1101A. Device output signal 1101A can be directed to speaker 1130A via digital-to-analog converter (DAC) 1122A, where it is amplified using power amplifier 1124A after digital-to-analog conversion.

[0065] Sensing circuit 1110A may include bandpass filters 1120-1, 1120-2…1120-M, which are coupled to corresponding root mean square (RMS) circuits 1123-1, 1123-2…1123-M. The output of each RMS circuit 1123-1, 1123-2…1123-M may be coupled to a circuit to calculate a statistic 1126. The statistic 1126 may be provided as the bandpass energy (dB) of the output of sensing circuit 1110A. The statistic 1126 may include one or more statistics from a set of statistics, including the median, average, arithmetic mean, geometric mean, maximum energy value, minimum energy value, and the energy of the nth largest energy among all calculated bandpass energies, where n is a selected positive integer. Sensing circuit 1110A may also include another RMS circuit 1121 to receive the output signal 1101A and generate a signal energy (dB) 1129, which may be output from sensing circuit 1110A. These RMS circuits and the circuits used to calculate statistics can include various types of memory devices and one or more processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), and other computing devices. The bandpass energy (dB) 1128 and signal energy (dB) 1129 output from the sensing circuit 1110A can be input to the dynamic gain unit.

[0066] Figure 12B This is a block diagram of an embodiment of example device 1100B, which provides sound output from speaker 1130B in response to device output signal 1101B and uses sensing circuitry 110B to analyze device output signal 1101B. Device output signal 1101B can be directed to speaker 1130B via digital-to-analog converter (DAC) 1122B, where it is amplified by power amplifier 1124B after digital-to-analog conversion.

[0067] Sensing circuitry 1110B may include circuitry for generating statistical feature 1125-1, circuitry for generating statistical feature 1125-2, ..., circuitry for generating statistical feature 1125-L, wherein each circuitry is coupled to distortion prediction computation 1127, which generates a predicted distortion probability associated with the sound output from speaker 1130B. Distortion prediction computation 1127 may be implemented as one or more computational circuits for measurements based on machine learning, statistical learning, predictive learning, or AI, which can provide a calculation of the distortion probability at speaker output 1130B. Conventional circuitry used for such learning and AI techniques can be used. These circuits may include various types of memory devices and one or more processors, DSPs, ASICs, and other computing devices. The predicted distortion probability may be output from sensing circuitry 1110B to dynamic gain unit.

[0068] Dynamic gain unit (e.g.) Figure 4 The dynamic gain unit 315, Figure 5 The dynamic gain unit 415 or a similar dynamic gain unit can be variably coupled to a multi-notch filter, for example... Figure 4 305 multi-notch filter Figure 5 The multi-notch filter 405 or similar multi-notch filter is applied to the speaker signal path to reduce friction and hum or other unpleasant distortion, such as... Figure 13 As shown. In various embodiments, the goal is to engage the multi-notch filter only when distortion is otherwise activated, minimizing the impact on perceived timbre when listening to a speaker system. Similar to a dynamic range compressor, a sensed analysis estimate of the distortion activation energy can be compared to a tunable threshold, such that an initial notch depth is determined when the estimate exceeds the threshold. The initial notch depth can then be scaled based on the significance of the sensed analysis estimate relative to the total signal energy.

[0069] Figure 13 This is a block diagram of an exemplary embodiment of dynamic gain control 1215, which can be used by, for example, in... Figure 12A and 12BThe sensing circuitry determines the measured distortion sensing energy to adjust the depth of the notch filter. The notch filter is engaged when the distortion sensing energy exceeds a threshold. Dynamic gain control 1215 includes inputs for receiving a measure of the bandpass energy 1228 and inputs for receiving a measure of the signal energy 1229 from the sensing circuitry 1210. The measure of the bandpass energy 1228 can be passed to comparator 1231, which generates the difference between a threshold (dB) and the measure of the bandpass energy 1228. The threshold is a parameter that provides a threshold of distortion activation energy at which the multiple notch filter will begin to engage. The output x of comparator 1231 can be input to comparator 1234, which selects the minimum value between x and 0. The output of comparator 1234 is input to multiplier 1238.

[0070] The measurement of bandpass energy 1228 can also be passed to adder 1232, which adds sensitivity (dB) to the measurement of bandpass energy 1228. The measurement of bandpass energy 1228 with added sensitivity can then be passed from adder 1232 to comparator 1233, which generates the difference between the measurement of bandpass energy 1228 with added sensitivity and the measurement of signal energy 1229. The output of comparator 1233 can be converted from dB to a linear measurement by converter 1236, the output of which is provided to range [x, 0, 10] 1237, which provides a ratio to multiplier 1238. Range [x, 0, 10] clamps the value of output x to the range between 0 and 10. It executes a function to output x: a value of 0 if x is less than zero; a value of 1 if x is greater than 10; and a value of x if x is greater than or equal to zero and less than or equal to 10. Sensitivity parameters can be used to adjust the sensitivity and range of a dynamic ratio from the range [x, 0, 10] 1237. The ratio from the range [x, 0, 10] 1237 is multiplied by the output of comparator 1234 and applied to the linear converter 1239 in decibels. A smoothing operator 1241 can be applied to the output of converter 1239. From the smoothing operator 1241, a multi-notch depth gain 1243 can be provided via dynamic gain control 1215.

[0071] While this is somewhat similar to compressor ratio control, compressor ratios are typically static and limited to gain reduction ratios ranging from 1:1 to infinity:1. Dynamic gain units with, for example, dynamic gain control 1215 can also apply an inverse ratio, meaning the multi-notch filter depth can be greater than the difference between the distortion activation energy and the threshold parameter. The dynamic unit also provides a smoothing parameter to mitigate multi-notch filter depth modulation artifacts. Other parameters that can be used with dynamic gain control can include mode, attack time, release time, and attack threshold. Mode relates to whether dynamic gain control statically and fully engages the multi-notch filter or dynamically and variably engages it. Attack time is the speed at which the notch will engage. Release time is the speed at which the notch will disengage. The attack threshold is the amount of differential notch depth engagement required for the desired attack time trajectory. The ratio of combined bandpass energy to total channel energy, measured in the sensing module, can be used to dynamically adjust the multi-notch filter depth.

[0072] Figure 14 This is a block diagram of a bass enhancement circuit 1400. The bass enhancement circuit 1400 can be implemented by one or more processors and a memory with instructions for bass enhancement. The one or more processors can be implemented as an ASIC, a DSP, or a combination thereof. Typically, bass enhancement has two main functions: increasing the amplitude of the bass frequency range and inserting harmonics. The bass enhancement circuit 1400 may include an input gain 1446 that operates in response to a received audio signal input. The output of the input gain 1446 can be input to a dithering process 1447. The output of the dithering process 1447 can be input to a positive harmonic cutoff filter 1448, which can be disabled if needed. The output of the positive harmonic cutoff filter 1448 can be routed to two paths. The output of the positive harmonic cutoff filter 1448 can be coupled to a summer 1449. The output of the positive harmonic cutoff filter 1448 can also be coupled to a pre-low-pass filter 1451.

[0073] The output of the pre-low-pass filter 1451 can be coupled to the harmonic generator 1452. The output of the harmonic generator 1452 can be coupled to the low-pass filter 1453, the output of which can optionally be operated by the gain filter 1454. The output of the low-pass filter 1453 or (if implemented) the gain filter 1454 can be coupled to the summer 1449. The summer 1449 can add the output of the positive harmonic cutoff filter 1448 (or the output of the dithering process 1447 if the positive harmonic cutoff filter 1448 is disabled) to the output of the low-pass filter 1453 or (if implemented) the output of the gain filter 1454.

[0074] The output of summer 1449 can be coupled to parametric equalizer (PEQ) filter 1456. The output of PEQ filter 1456 can be coupled to dynamic processing module 1457. Dynamic processing module 1457 may include a shelving filter whose output can be operated by make-up gain. The output of dynamic processing module 1457 can be coupled to high-pass filter 1459. High-pass filter 1459 can provide an output that is directed along a path to a speaker. Figure 14 Audio signal (not shown in the image).

[0075] Harmonic generation 1452 can utilize active nonlinearity. For such harmonic generation, a low-pass filter 1451 can be inserted before harmonic generation 1452, which avoids the generation of harmonics from higher frequency components. Since the low-pass filter 1451 is positioned before harmonic generation 1452, it can be referred to as a pre-low-pass filter 1451. The pre-low-pass filter 1451 can have an adjustable cutoff frequency. The pre-low-pass filter 1451 can be an eighth-order filter, which can have coefficients for a single second-order section, which can be repeated four times per block to achieve an eighth-order transfer function. The cutoff frequency of each second-order section can be fixed, so the four copies of the second-order section generate a frequency response four times the gain.

[0076] Harmonic generator 1452 can employ active nonlinearity to generate a richer harmonic spectrum. Many different nonlinear and harmonic generation methods common in the art can be selected to generate harmonics. Higher gain and wider bandwidth can be provided to compensate for the increased energy in the low-mid frequency band region. To compensate for the increased harmonic energy in the low-mid frequency band region due to the updated nonlinear insertion in harmonic generator 1452, the height and width of the PEQ filter 1456 in the low-frequency band region can be increased to provide an improved sense of depth relative to the increased low-mid frequency band energy. Bass level adjustment is applied to the harmonics and the equalization of the mix. Additionally, the PEQ filter 1456 can be selected and tuned according to techniques common in the art to enhance the low-frequency region, wherein the amplitude and bandwidth of the low-frequency enhancement are adjustable.

[0077] If this process is enabled, input gain 1446 can provide gain to the input audio signal. This gain can be used as a headroom gain for this process, which can be set between two selected gain levels. Headroom allows the internal signal to not overshoot or undershoot below the maximum or minimum values ​​supported in the processing module. Once the input gain is applied, the algorithm checks for jitter, which can be set by a silence detector invoked in the algorithm before applying the input gain. If silence is detected, jitter can be added to the signal. Jitter can be added only to the algorithm's analysis chain, and it helps the dynamic processing applied later to minimize any boost effects that may be noticeable when switching between silence and signal segments.

[0078] After the input gain and jitter stages, the main process can be called for further processing. On the other hand, if this process is bypassed (… Figure 14 (not shown in the diagram), then the signal is first delayed by an amount equal to the delay introduced by the algorithm when the process is enabled, and then a gain equal to the bypass gain is applied to the signal. The delay introduced in the bypass path causes the effect of minimizing the visible discontinuity between the on and off states of the algorithm, and vice versa.

[0079] The main process generates a 1452 monitoring signal from harmonics and calculates how much headroom is available. This can be done by finding the maximum values ​​of both the L and R channels block by block. These maximum values ​​are stored in a statistical history buffer of a selected length, and they can be averaged to give the average maximum value for each channel in the history of the buffer of the selected length.

[0080] Low-pass filter 1453 is used to limit the highest frequency of the injected harmonic content, and the generated harmonics can optionally be processed by an additional filter, gain filter 1454, which has low-pass gain characteristics. This can be performed to shape the generated harmonics and give more emphasis to the lower harmonics.

[0081] Once the PEQ filter 1456 has been applied to the signal, dynamic processing 1457 can be applied, which may include applying compensation gain across the entire frequency range of the signal. A preprocessing function can be invoked, which calculates the maximum value of the enhanced signal for both the L and R channels. Additionally, the preprocessing function can identify the number of signal peaks above a threshold. The preprocessing function can manage a look-ahead buffer for the protected gain applied to the signal in the next stage of the process. Many dynamic processing techniques common in the art can be implemented. The compensation gain provided by this dynamic processing is adjustable.

[0082] The boost applied to the signal can be increased or decreased incrementally based on all the checks described previously. The maximum boost that can be applied can be calculated based on the exposed controls. Dynamic processing can be used internally to drive the gain line equation, which generates the maximum boost allowed for a given dynamic setting. This maximum value can be compared to the calculated available headroom. In other words, the maximum boost to be applied does not exceed the previously calculated available headroom. Furthermore, in various embodiments, the maximum gain applied is not greater than the inverted, tracked average maximum value of the signal, and correspondingly, the boost gain is not greater than that maximum gain. Dynamic processing can dynamically adjust the amount of gain added to the signal based on the signal's dynamics. In other words, the stronger the signal, the less gain is added.

[0083] After applying dynamic processing 1457, the signal can pass through high-pass filter 1459. The order of high-pass filter 1459 can be varied, for example, from 1 to 8. The cutoff frequency of high-pass filter 1459 can be user-exposed controlled and can be tuned to reduce signal energy below the speaker size setting. If the high-pass filter is applied outside of bass enhancement circuitry 1400, high-pass filter 1459 can be disabled.

[0084] Figure 15 This is a block diagram of an embodiment of example system 2000, which features dynamic multi-feature distortion sensing and adaptive multi-band distortion reduction for audio signals, combined with distortion-aware harmonic bass enhancement. System 2000 may include a bass enhancer 2050, a multi-band or multi-notch dynamic attenuator 2005, a distortion sensing circuit 2010, a dynamic gain unit 2015, and a dynamic bass parameter control 2055, all operable with a speaker 2030 providing sound output. The speaker 2030 may be implemented as a miniature speaker. The dynamic bass parameter control 2055 may modify one or more adjustable bass enhancement parameters, which may be obtained from a non-exhaustive list of the following:

[0085] 1. The peak amplitude of the swept-tone envelope response of the bass enhancement module can be provided as the amplitude of the PEQ filter 1456.

[0086] 2. The center frequency of the PEQ filter 1456.

[0087] 3. Bandwidth of the PEQ filter 1456

[0088] 4. The amount of harmonic energy inserted by the bass enhancement module, which can achieve an adjustable gain for a summation of 1449.

[0089] 5. The cutoff frequency of the pre-low-pass filter 1451, which provides the maximum frequency at which energy is supplied to the harmonic generator 1452.

[0090] 6. The cutoff frequency of the low-pass filter 1453, which provides the maximum frequency at which harmonic energy of the low-frequency enhancement signal injected through summation 1449 is obtained.

[0091] 7. The threshold value of 1452 is used in the nonlinearity of harmonic generation.

[0092] 8. The ratio of odd-order harmonics to even-order harmonics generated by the 1452 insertion.

[0093] 9. Dynamically process the gain of 1457, and / or

[0094] 10. The slope of the harmonic generated by the 1452-interpolation.

[0095] System 2000 may optionally include an audio preprocessing 2004 operable on preprocessed audio input. The output from audio preprocessing 2004 may be input to a bass booster 2050, the output of which may optionally be processed by intermediate audio processing 2009 before being input to a multi-band or multi-notch dynamic attenuator 2005. The output from multi-band or multi-notch dynamic attenuator 2005 may be coupled to a distortion sensing circuit 2010 and a path coupled to a speaker 2030. Optionally, the output from multi-band or multi-notch dynamic attenuator 2005 may be processed by audio amplification and other post-processing before the speaker 2030. Audio amplification and other post-processing may include speaker offset protection, bass boost, digital-to-analog conversion, and analog gain.

[0096] The distortion sensing circuit 2010 can be arranged in a variety of ways. As taught herein, the sensing circuit 2010 can be operated to generate a statistic that measures the probability of distortion or the degree of perceptible, objectionable, or measurable distortion at the output of the speaker 2030. The distortion sensing circuit 2010 can acquire its single input signal from various points in the signal serial chain, or it can acquire multiple input signals from various points in the signal serial chain. For multiple input signals of the distortion sensing circuit 2010, the distortion sensing circuit 2010 can include a multivariate statistical or multivariate machine learning regression mechanism to determine the output of the distortion sensing circuit 2010.

[0097] The distortion sensing circuit 2010 can be configured to operate using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence (AI) to calculate a measure of the probability of distortion or the degree of perceptible, objectionable, or measurable distortion at the speaker output. This technique can use one or more processes selected from the group consisting of classification and regression trees (aka decision trees), ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression. Artificial neural networks can include, but are not limited to, perceptrons, multilayer perceptrons, deep belief networks, and deep neural networks. Nonlinear regression can include, but is not limited to, multinomial regression.

[0098] The distortion sensing circuit 2010 can be configured to operate to compute a binary indicator of distortion at the speaker output using techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence. This technique can use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, Naive Bayes classification, linear discriminant analysis, and random forests. Support vector machines can include support vector classification. Artificial neural networks can include, but are not limited to, perceptrons, multilayer perceptrons, deep belief networks, and deep neural networks.

[0099] The sensing circuit 2010 can be configured to operate to compute a binary indicator by derivation from a soft indicator using a threshold or comparator device. Alternatively, a semi-binary or semi-hard indicator can be derived from the soft indicator using a threshold, thereby generating discrete indicators for input values ​​above / below a certain threshold, and variable indicators for input values ​​below / above a certain threshold. In the case of deriving binary or semi-binary indicators in this way, hysteresis can be employed to avoid issuing discrete indicators for only brief offsets of the soft indicator exceeding the threshold.

[0100] The dynamic gain unit 2015 can be coupled to receive the output of the distortion sensing circuit 2010 and coupled to the multi-band or multi-notch dynamic attenuator 2005 to operatively provide parameters or control signals to modify the attenuation caused by the multi-band or multi-notch dynamic attenuator 2005. The output of the distortion sensing circuit 2010 can be input to the dynamic bass parameter control 2055, which can provide input to the bass enhancement 2050 to adjust or modify the bass enhancement parameters in the bass enhancement 2050.

[0101] The Dynamic Bass Parameter Control 2055 can dynamically modify multiple bass enhancement parameters. These parameters can be selected from a set that may include the peak amplitude of the sweep envelope response of the bass enhancer 2050 (which can be represented by a "bump" around the amplified bass frequency region), the center frequency of the amplified bass frequency region, the frequency width of the amplified bass frequency region, the amount of harmonic energy inserted by the bass enhancer 2050, the maximum harmonic order inserted by the bass enhancer 2050, the maximum frequency of the harmonic energy inserted by the bass enhancer 2050, the shape of the continuous harmonics inserted by the bass enhancer 2050, including the slope of the continuous harmonics inserted by the bass enhancer 2050, the ratio of even to odd harmonics inserted by the bass enhancer 2055, or any linear or nonlinear crossfading characteristic curve that can adjust all the above parameters from a relative "low bass" setting to a "high bass" setting.

[0102] It can realize various variations of the architecture for distortion sensing and distortion reduction, as well as the architecture for distortion-aware harmonic bass enhancement. Figure 16 This is a block diagram of an embodiment of an adaptive and dynamic sensing and multi-notch filter module 2100. The sensing and multi-notch filter module 2100 may include a multi-band analysis filter bank 2116 that receives an audio signal and provides an output to each of dynamic processing 2118-1, dynamic processing 2118-2, and dynamic processing 2118-3. Like the multi-band analysis filter bank 2116, the multi-band analysis filter bank can decompose the input signal into different spectral ranges, such as low-frequency band, mid-frequency band, and high-frequency band. The dynamic module is essentially a form of dynamic compression processing that can be performed over a selected frequency range. Each of dynamic processing 2118-1, dynamic processing 2118-2, and dynamic processing 2118-3 can be input to a multi-band synthesis filter bank 2119. The multi-band synthesis filter bank 2119 can provide an input to an adaptive notch filter module 2105, which provides a filtered audio signal to a limiter 2107 to provide a signal for sound output.

[0103] The multi-band synthesized filter bank 2119 can also provide input to the distortion sensing module 2110. The distortion sensing module 2110 can also receive feedback from the limiter 2107. The signals from the limiter 2107 and the multi-band synthesized filter bank 2119 can be used by the distortion sensing module 2110 to determine the distortion and generation parameters used as input to the adaptive notch filter module 2105 to reduce distortion. The distortion sensing module 2110 and the adaptive notch filter module 2105 may include functions that perform AND and AND operations. Figure 3-14 The associated notch filter and sensing circuit have similar or identical functions and / or characteristics.

[0104] Figure 17 This is a block diagram of an embodiment of an adaptive and dynamic sensing and multi-notch filter module 2200. The sensing and multi-notch filter module 2200 may include a multi-band analysis filter bank 2216 that receives an audio signal and provides an output to each of dynamic processing 2218-1, dynamic processing 2218-2, and dynamic processing 2218-3. The outputs of each of dynamic processing 2218-1, dynamic processing 2218-2, and dynamic processing 2218-3 may be input to a multi-band synthesis filter bank 2119. The multi-band synthesis filter bank 2219 may provide an input to an adaptive notch filter module 2205 that provides a filtered audio signal to a limiter 2207 to provide a signal for sound output.

[0105] The multi-band synthesis filter bank 2219 can also provide input to the multi-band level management and distortion sensing module 2210. The multi-band level management and distortion sensing module 2210 can also receive outputs from the multi-band analysis filter bank 2216 that are directed to each of the dynamic processes 2218-1, 2218-2, and 2218-3. Additionally, the multi-band level management and distortion sensing module 2210 can receive feedback from the limiter 2207. The signals from the limiter 2207, the multi-band synthesis filter bank 2219, and the output from the multi-band analysis filter bank 2216 can be used by the multi-band level management and distortion sensing module 2210 to determine the distortion and generation parameters for input to the adaptive notch filter module 2205 to reduce distortion. Furthermore, the multi-band level management and distortion sensing module 2210 can use these inputs to provide parameters for the dynamic processes 2218-1, 2218-2, and 2218-3. The multi-band horizontal management and distortion sensing module 2210 and the adaptive notch filter module 2205 may include execution and Figure 3-16 The associated notch filter and sensing circuit have similar or identical functions and / or characteristics.

[0106] Figure 18 This is a block diagram of an embodiment of an adaptive and dynamic sensing and multi-notch filter module 2400. The sensing and multi-notch filter module 2400 may include a multi-band analysis filter bank 2416 that receives an audio signal and provides an output to each of dynamic processing 2418-1, dynamic processing 2418-2, and dynamic processing 2418-3. The outputs of each of dynamic processing 2418-1, dynamic processing 2418-2, and dynamic processing 2418-3 are inputs to a multi-band synthesis filter bank 2419. The multi-band synthesis filter bank 2419 may provide an input to an adaptive notch filter module 2405 that provides a filtered audio signal to a limiter 2407 to provide a signal for sound output.

[0107] The multi-band synthesized filter bank 2419 can also provide input to the distortion sensing module 2410. The signal from the multi-band synthesized filter bank 2419 can be used by the distortion sensing module 2410 to determine distortion and generation parameters for input to the adaptive notch filter module 2405 to reduce distortion. In this architecture, the signal from the limiter 2407 is not fed back to the distortion sensing module 2410. The distortion sensing module 2410 and the adaptive notch filter module 2405 may include functions that perform AND and AND operations. Figure 3-17 The associated notch filter and sensing circuit have similar or identical functions and / or characteristics.

[0108] Figure 19 This is a block diagram of an embodiment of an adaptive and dynamic sensing and multi-notch filter module 2500. The sensing and multi-notch filter module 2500 may include a multi-band analysis filter bank 2516 that receives an audio signal and provides an output to each of dynamic processing 2518-1, dynamic processing 2518-2, and dynamic processing 2518-3. The outputs of each of dynamic processing 2518-1, dynamic processing 2518-2, and dynamic processing 2518-3 are inputs to a multi-band synthesis filter bank 2519. The multi-band synthesis filter bank 2519 may provide an input to an adaptive notch filter module 2505, which provides a filtered audio signal to a limiter 2507 to provide a signal for sound output.

[0109] The output of limiter 2507 can also be fed back to distortion sensing module 2510. The feedback signal from limiter 2507 can be used by distortion sensing module 2510 to determine the distortion and generation parameters to be input to adaptive notch filter module 2505 to reduce distortion. Figure 3-18 The associated notch filter and sensing circuit have similar or identical functions and / or characteristics.

[0110] Figure 20This is a block diagram of an embodiment of an adaptive and dynamic sensing and multi-notch filter module 2600. The sensing and multi-notch filter module 2600 may include a multi-band analysis filter bank 2616 that receives an audio signal and provides an output to each of dynamic processing 2618-1, dynamic processing 2618-2, and dynamic processing 2618-3. The outputs of each of dynamic processing 2618-1, dynamic processing 2618-2, and dynamic processing 2618-3 serve as inputs to a multi-band synthesis filter bank 2619. The multi-band synthesis filter bank 2619 may provide an input to an adaptive notch filter module 2605, which provides a filtered audio signal to a limiter 2607. The output of the limiter 2607 may be input to an amplifier 2669, the output of which provides a signal for sound output.

[0111] The multi-band synthesizer filter bank 2619 can also provide input to the distortion sensing module 2610. The distortion sensing module 2610 can also receive feedback from the limiter 2607 and feedback from the amplifier 2669. The signals from the amplifier 2669, limiter 2607, and multi-band synthesizer filter bank 2619 can be used by the distortion sensing module 2610 to determine the distortion and generation parameters for input to the adaptive notch filter module 2605 to reduce distortion. The distortion sensing module 2610 and the adaptive notch filter module 2605 may include functions that perform AND and AND operations. Figure 3-19 The related notch filter and sensing circuit have similar or identical functions and / or characteristics.

[0112] Figure 21 This is a flowchart illustrating the features of an embodiment of an example method 2101 for sensing and reducing distortion in sound output generated from a speaker. At 2102, a signal from an audio signal source is received at the speaker. At 2103, a signal is received at a sensing circuit. At 2104, statistics about the signal are generated in the sensing circuit. The generation of the statistics of the signal in the sensing circuit can be implemented in a digital signal processor. Generating the statistics may include generating a statistic that measures the probability of distortion at the speaker output, or measures the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0113] At 2106, the signal is modified based on the generated statistics, and the modified signal is directed to the speaker. At 2108, sound output is generated from the speaker based on the received modified signal.

[0114] Variations of method 2101 or methods similar to method 2101 may include a number of different embodiments that can be combined depending on the application of these methods and / or the architecture of the system in which these methods can be implemented. These methods may include calculating a binary indicator of distortion at the speaker output in the sensing circuit using techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence. Using this technique includes using one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, Naive Bayes classification, linear discriminant analysis, and random forests. Variations of method 2101 or methods similar to method 2101 may include calculating a soft indicator in the sensing circuit corresponding to the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion using techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence.

[0115] Variations of or similar to method 2101 may include generating statistics about the signal in a sensing circuit, including using a sensing circuit having a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to speaker distortion. Such a method may include estimating, in the sensing circuit, the amount of distortion-activated energy in the filtered signal from the plurality of IIR or FIR bandpass filters and / or dynamic spectrum analysis filters as one or more statistics calculated from the energy across all frequency bands of the plurality of bandpass and / or dynamic spectrum analysis filters, and measuring the total energy of the filtered signal. The one or more statistics may include one or more statistics from a set of statistics including the median, average, arithmetic mean, geometric mean, maximum energy value, minimum energy value, and the energy of the nth largest energy, where n is a chosen positive integer. Variations of method 2101 or methods similar to method 2101 may include calculating energy as time-varying energy in the sensing circuit, or calculating the envelope of a full-band audio signal.

[0116] Figure 22This is a flowchart illustrating the features of an embodiment of an example method 2201 for sensing and reducing distortion in sound output generated from a loudspeaker. At 2202, a multiple notch filter is applied to a signal, which is provided to generate sound output. At 2204, a filtered signal corresponding to the signal received and processed by the multiple notch filter is received at a sensing circuit, and the sensing circuit is used to generate statistics about the filtered signal. Generating statistics may include generating statistics for measuring the probability of distortion at the loudspeaker output. Using the sensing circuit may include using multiple infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters, which are tuned to frequencies determined to contribute to distortion in the loudspeaker. Using the sensing circuit may include estimating the amount of distortion activation energy in the filtered signal as an average of the energy across all frequency bands of the multiple IIR or FIR bandpass filters and / or dynamic spectrum analysis filters, and measuring the total energy of the filtered signal.

[0117] At 2206, a dynamic gain unit coupled to receive statistics from the sensing circuit is used to modify the depth of the multi-notch filter based on the statistics. Using the dynamic gain unit includes providing gain based on one or more statistics to adjust the depth of the multi-notch filter. Modifying the depth of the multi-notch filter using the dynamic gain unit may include adjusting the depth of the multi-notch filter based on the ratio of the average value to the measured total energy. At 2208, a filtered signal is received at the speaker, and an acoustic output is generated from the speaker. Receiving the filtered signal processed by the multi-notch filter, generating statistics about the filtered signal, and modifying the depth of the multi-notch filter can be implemented in a digital signal processor.

[0118] Variations of method 2201, or methods similar to method 2201, may include using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence in the sensing circuit to calculate a measure of the probability of distortion at the speaker output. Such variations may include using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, or artificial intelligence in the sensing circuit to calculate a binary indicator of distortion at the speaker output.

[0119] Variations of method 2201 or methods similar to method 2201 may include applying a multiple notch filter to a signal, including providing the signal as a bass-processed signal obtained by processing of a bass enhancement circuit to the multiple notch filter. The bass enhancement circuit processing may include generating statistics about the bass-processed signal using sensing circuitry; generating parameters based on the statistics about the bass-processed signal; and using these parameters to provide bass enhancement to the signal applied to the multiple notch filter.

[0120] Figure 23This is a flowchart illustrating the features of an embodiment of an example method 2301 for realizing distortion-aware harmonic bass enhancement of an audio signal. At 2302, a signal is received at a bass enhancement circuit, wherein the signal is provided for generating an acoustic output. At 2304, a bass-enhanced signal corresponding to the signal received and processed by the bass enhancement circuit is received at a sensing circuit, and statistics about the bass-enhanced signal are generated using the sensing circuit. The processing of the bass enhancement circuit may include performing one or more bass enhancements selected from the group including bass enhancement based on linear filtering and bass enhancement based on the insertion and / or generation of additional harmonics of the signal carrying audio content. The processing of the bass enhancement circuit may include using an adaptive harmonic bass enhancement circuit to vary the amount of harmonic energy added to the received signal, the adaptive harmonic bass enhancement circuit employing tunable asymmetric nonlinearity and pre-cutoff filtering and post-cutoff filtering. Generating statistics may include generating statistics that measure the probability of distortion at the speaker output, or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0121] At 2306, a parameter is generated based on statistics about the bass-processed signal, and this parameter is used to provide bass enhancement to the signal applied to the bass enhancement circuitry. Using this parameter to provide bass enhancement may include providing parameters for controlling bass amplitude and bass harmonic levels. At 2308, the bass-enhanced signal is received at the speaker, and an acoustic output is generated from the speaker.

[0122] Variations of method 2301 or methods similar to method 2301 may include a number of different embodiments that can be combined depending on the application of these methods and / or the architecture of the system implementing these methods. Variations of method 2301 or methods similar to method 2301 may include calculating a binary indicator of distortion at the speaker output in the sensing circuit using techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence. Using the techniques may include using one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, Naive Bayes classification, linear discriminant analysis, and random forests.

[0123] Variations of method 2301 or methods similar to method 2301 may include using techniques selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence to calculate a soft indicator corresponding to the probability of distortion or the degree of offensive, perceptible, or measurable distortion at the speaker output during sensing. Using this technique may include employing one or more processes selected from the group consisting of classification and regression trees, ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression. These methods may include deriving a binary indicator from the soft indicator using a thresholding device or a comparator device.

[0124] Variations of method 2301 or methods similar to method 2301 may include using a sensing circuit having a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to speaker distortion. Such a method may include estimating, in the sensing circuit, the amount of distortion-activated energy in the filtered signal from the plurality of IIR or FIR bandpass filters and / or dynamic spectrum analysis filters as one or more statistics calculated from the energy across all frequency bands of the plurality of bandpass filters and / or dynamic spectrum analysis filters, and measuring the total energy of the filtered signal. The one or more statistics may include one or more statistics from a set of statistics including the median, average, arithmetic mean, geometric mean, maximum energy value, minimum energy value, and the energy of the nth largest energy value, where n is a chosen positive integer.

[0125] Figure 24This is a flowchart illustrating the features of an embodiment of an example method 2401 for realizing distortion sensing, distortion reduction, and distortion-aware harmonic bass enhancement of an audio signal. At 2402, a signal is received at a bass enhancement circuit and a bass-enhanced signal is generated, which is provided to generate a sound output. At 2403, the bass-enhanced signal or a processed form of the bass-enhanced signal is received in a multi-notch filter, which is arranged to provide a filtered bass-enhanced signal. At 2404, a sensing circuit receives a signal corresponding to the signal received and processed by the bass enhancement circuit and the multi-notch filter, or a filtered bass-enhanced signal corresponding to the sound output, and uses the sensing circuit to generate statistics regarding the filtered bass-enhanced signal. At 2406, a parameter based on the statistics regarding the filtered bass-processed signal is generated in a bass parameter controller, and this parameter is used to provide bass enhancement to the signal applied to the bass enhancement circuit. Statistics from the sensing circuit can be received in a dynamic gain unit, and the depth of the multi-notch filter can be modified using the dynamic gain unit based on the received statistics. A filtered, bass-enhanced signal can be generated from a multi-notch filter with modified depth. At 2408, the filtered, bass-enhanced signal is received at the speaker, and an acoustic output is generated from the speaker.

[0126] Variations of method 2401 or methods similar to method 2401 may include a number of different embodiments that can be combined according to the application of these methods and / or the architecture of a system in which these methods can be implemented. These methods may include operating bass enhancement circuitry, sensing circuitry, bass parameter controller, multi-notch filter, and dynamic gain unit in a digital signal processor. These methods may include amplifying and / or post-processing the output of the multi-notch filter and providing the amplified and / or post-processed output of the multi-notch filter to a loudspeaker. These methods may also include audio processing of the bass-enhanced signal and providing the audio-processed bass-enhanced signal to the multi-notch filter. Features of method 2401 may be combined with features of methods 2101, 2201, or 2301. Additionally, features as taught herein may be combined with methods 2101, 2201, 2301, 2401, methods 2101, 2201, 2301, and 2401, and variations thereof.

[0127] Figure 25This is a block diagram of an embodiment of an example audio system 2900 arranged to operate with respect to distortion sensing, prevention, and distortion-aware bass enhancement according to the techniques taught herein. The audio system 2900 may include an input device 2901 and a processing device 2912, operatively coupled to the input device 2901 to receive and manipulate the input audio signal to control the generation of a sound output signal. The input device 2901 may include one or more audio sources, such as a DVD player, Blu-ray device, TV tuner, CD player, handheld player, internet audio / video, game console, or other device providing signals for audio production. The input device 2901 may be a node or set of nodes of an input source providing one or more signals to the processing device 2912. The output audio signal may be sent to an audio output device 2925, such as a speaker. The speaker may be a stereo speaker, surround sound speaker, headset speaker, miniature speaker, or other similar audio output device.

[0128] Processing device 2912 can be implemented using various machine tools to perform and / or control functions associated with manipulating signals for audio production as taught herein. Such machine tools may include one or more devices designed to perform the functions described herein, such as, but not limited to, general-purpose processors, computing devices having one or more processing devices, DSPs, ASICs, field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. General-purpose processors and processing devices may be microprocessors, but alternatively, the processor may be a controller, microcontroller, or state machine, combinations thereof, etc. The processor may also be implemented as a combination of computing devices, such as a combination of DSPs and microprocessors, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration. Processing device 2912 may be implemented by a set of processing devices having one or more processing devices. This set of processing devices may include a set of integrated circuits. Processing device 2912 may be operatively coupled to one or more storage devices 2517 storing instructions and data for processing as taught herein. Storage devices 2517 may be implemented by multiple machine-readable storage devices.

[0129] Processing device 2912 can be configured to apply a signal to a multi-notch filter, wherein the signal is provided for generating sound output; receive a filtered signal corresponding to the signal received and processed by the multi-notch filter at a sensing circuit, and use the sensing circuit to generate statistics about the filtered signal; modify the depth of the multi-notch filter based on the statistics using a dynamic gain unit coupled to receive the statistics from the sensing circuit; and receive the filtered signal at a loudspeaker and generate sound output from the loudspeaker. The generation of statistics may include generating statistics for measuring the probability of distortion at the loudspeaker output. The use of the sensing circuit may include using multiple infinite impulse response (IIR) or finite impulse response bandpass filters and / or dynamic spectrum analysis filters, which are tuned to frequencies determined to contribute to loudspeaker distortion. The use of the sensing circuit may include estimating the amount of distortion activation energy in the filtered signal as an average of the energy across all frequency bands of the multiple IIR or FIR filters and / or dynamic spectrum analysis filters, and measuring the total energy of the filtered signal.

[0130] In various embodiments, the processing device 2912 may be configured to apply a signal as a bass-processed signal obtained from the processing of the bass enhancement circuit to a multi-notch filter. The processing of the bass enhancement circuit may include: generating statistics about the bass-processed signal using sensing circuitry; generating parameters based on the statistics about the bass-processed signal; and using these parameters to provide bass enhancement for the signal applied to the multi-notch filter.

[0131] In various embodiments, processing device 2912 may be configured to receive a signal at a bass enhancement circuit, wherein the signal is provided for generating sound output; receive a bass-enhanced signal corresponding to the signal received and processed by the bass enhancement circuit at a sensing circuit, and use the sensing circuit to generate statistics about the bass-enhanced signal; generate parameters based on the statistics of the bass-processed signal, and use these parameters to provide bass enhancement for the signal applied to the bass enhancement circuit; and receive the bass-enhanced signal at a speaker and generate sound output from the speaker. A set of parameters for providing bass enhancement may include parameters for controlling bass amplitude and bass harmonic levels. The generation of statistics may include generating statistics measuring the probability of distortion at the speaker output.

[0132] Processing of the bass enhancement circuit may include performing one or more bass enhancements selected from the group consisting of bass enhancement based on linear filtering and bass enhancement based on the insertion and / or generation of additional harmonics of the signal carrying audio content. Processing of the bass enhancement circuit may include using an adaptive harmonic bass enhancement circuit to vary the amount of harmonic energy added to the received signal, the adaptive harmonic bass enhancement circuit employing tunable asymmetric nonlinearity as well as pre-cutoff filtering and post-cutoff filtering.

[0133] Processing device 2912 can be configured to provide distortion sensing, prevention, and / or distortion-aware bass enhancement. Processing device 2912 may include features as taught herein and variations thereof to provide functionality for sensing and reducing distortion. Processing device 2912 may include features as taught herein and variations thereof to provide functionality for distortion-aware bass enhancement. Combinations of features may be incorporated into processing device 2912.

[0134] According to various embodiments, the first example system may include: a speaker configured to receive a signal from an audio signal source and generate a sound output based on the received signal; and a sensing circuit coupled to receive a signal and generate statistics about the signal to modify the signal received by the speaker.

[0135] Alternatively, according to the aforementioned first example system, another embodiment provides that the sensing circuit is implemented in a digital signal processor.

[0136] Optionally, in any of the foregoing examples of the first example system, further embodiments provide sensing circuitry operable to generate statistics that measure the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0137] Optionally, in any of the foregoing examples of the first example system, further embodiments provide sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence to calculate a binary indicator of distortion at the speaker output.

[0138] Optionally, in any of the foregoing examples of the first example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, naive Bayes classification, linear discriminant analysis, and random forests.

[0139] Optionally, in any of the foregoing examples of the first example system, further embodiments provide sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence to calculate a soft indicator corresponding to the probability of distortion or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0140] Optionally, in any of the foregoing examples of the first example system, further embodiments provide that the sensing circuitry may include a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to distortion of the loudspeaker.

[0141] Optionally, in any of the foregoing examples of the first example system, a further embodiment provides a sensing circuit that may include circuitry for estimating the amount of distortion activation energy in a filtered signal from a plurality of IIR or FIR bandpass filters and / or dynamic spectrum analysis filters as one or more statistics calculated from the energy across all frequency bands of the plurality of bandpass filters, and the sensing circuitry includes circuitry for measuring the total energy of the filtered signal.

[0142] Optionally, in any of the foregoing examples according to the first example system, a further implementation provides one or more statistics that may include one or more statistics from a set of statistics, including the median, mean, arithmetic mean, geometric mean, maximum energy value, minimum energy value, and the energy of the nth largest energy among all calculated bandpass energies, where n is a selected positive integer.

[0143] Alternatively, in any of the foregoing examples of the first example system, further embodiments provide that the sensing circuitry can be arranged to calculate energy as time-varying energy or to calculate the envelope of a full-band audio signal.

[0144] According to various embodiments, a first example method may include: receiving a signal from an audio signal source at a speaker; receiving the signal at a sensing circuit; generating statistics about the signal in the sensing circuit; modifying the signal based on the generated statistics and directing the modified signal to the speaker; and generating sound output from the speaker based on the received modified signal.

[0145] Alternatively, based on the first example method described above, other implementations may provide operations associated with performing any of the aforementioned examples based on the first example system.

[0146] According to various embodiments, a first example non-transitory computer-readable medium is provided storing computer instructions that, when executed by one or more processors, cause a system to perform the following operations: receiving a signal from an audio signal source at a speaker; receiving the signal at a sensing circuit; generating statistics about the signal in the sensing circuit; modifying the signal based on the generated statistics and directing the modified signal to the speaker; and generating sound output from the speaker based on the received modified signal.

[0147] Optionally, based on the aforementioned first example non-transitory computer-readable medium storing computer instructions, other implementations may provide operations associated with performing any of the aforementioned examples according to the first example method and / or the first example system.

[0148] According to various embodiments, the second example system may include: a multi-notch filter configured to receive a signal for generating sound output; a sensing circuit coupled to receive the filtered signal and generate statistics about the filtered signal, the filtered signal corresponding to a signal received and processed by the multi-notch filter or a signal corresponding to a speaker output; a dynamic gain unit coupled to receive statistics from the sensing circuit and coupled to the multi-notch filter to operatively modify the depth of the multi-notch filter; and a speaker for receiving the filtered signal to generate sound output.

[0149] Alternatively, according to the aforementioned second example system, another implementation provides that sensing circuitry and other modules can be implemented on a DSP.

[0150] Optionally, in any of the foregoing examples of the second example system, further embodiments provide sensing circuitry operable to generate statistics that measure the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0151] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence to calculate a binary indicator of distortion at the speaker output.

[0152] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides that a variable hangover time can be provided to the binary indicator to avoid the situation where the tail of the distorted audio segment is lost.

[0153] Alternatively, in any of the foregoing examples according to the second example system, a further implementation provides that a variable hang-before time (with a corresponding look-ahead or processing delay combined for multi-notch processing) can be provided to the binary indicator to avoid the occurrence of a leading portion of a lost distorted audio segment.

[0154] Optionally, in any of the foregoing examples according to the second example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, naive Bayes classification, linear discriminant analysis, and random forests.

[0155] Optionally, in any of the foregoing examples of the second example system, a further implementation provides sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence to calculate a soft indicator corresponding to the probability of distortion or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0156] Optionally, in any of the foregoing examples according to the second example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression.

[0157] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides that a binary indicator can be derived from the soft indicator by using a threshold or comparator device.

[0158] Optionally, in any of the foregoing examples of the second example system, a further implementation provides that a semi-binary or semi-hard indicator can be derived from a soft indicator using a threshold, thereby generating a constant indicator for input indicator values ​​above / below a certain threshold, and a variable indicator for input indicator values ​​below / above a certain threshold.

[0159] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides that the hysteresis can be incorporated into the derivation of the hard value of the binary or semi-binary indicator to avoid issuing only a brief binary indication for a soft indicator exceeding a threshold.

[0160] Optionally, in any of the foregoing examples of the second example system, a further embodiment provides a sensing circuit including a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to the distortion of the loudspeaker.

[0161] Optionally, in any of the foregoing examples of the second example system, a further embodiment provides that the sensing circuit may include circuitry for estimating the amount of distortion activation energy in the filtered signal as a statistic calculated from the energy across all frequency bands of a plurality of bandpass filters, and the sensing circuitry includes circuitry for measuring the total energy of the filtered signal.

[0162] Optionally, in any of the foregoing examples based on the second example system, a further implementation provides that the statistic can be the median.

[0163] Optionally, in any of the foregoing examples based on the second example system, a further implementation provides that the statistic can be the average.

[0164] Optionally, in any of the foregoing examples according to the second example system, a further implementation provides that the average value can be an arithmetic average.

[0165] Optionally, in any of the foregoing examples according to the second example system, a further implementation provides that the average value may be a geometric average.

[0166] Optionally, in any of the foregoing examples based on the second example system, a further implementation provides that the statistical value can be the maximum value.

[0167] Optionally, in any of the foregoing examples based on the second example system, a further implementation provides that the statistical value can be the minimum value.

[0168] Alternatively, in any of the foregoing examples according to the second example system, a further implementation specifies that the statistic may be the energy of the nth largest energy among all calculated bandpass energies.

[0169] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides a sensing circuit capable of calculating the time-varying energy or envelope of a full-band audio signal.

[0170] Alternatively, in any of the foregoing examples of the second example system, a further implementation provides that the dynamic gain unit can be arranged to provide gain based on subsequent statistics calculated from the time-varying energy or envelope and statistics of the full-band audio signal to adjust the depth of the multi-notch filter.

[0171] According to various embodiments, the second example method may include: applying a multiple notch filter to a signal provided for generating a sound output; receiving a filtered signal corresponding to the signal received and processed by the multiple notch filter or corresponding to the sound output in a sensing circuit, and generating statistics about the filtered signal using the sensing circuit; modifying the depth of the multiple notch filter based on the statistics using a dynamic gain unit coupled to receive the statistics from the sensing circuit; and receiving the filtered signal and generating a sound output from the speaker in a loudspeaker.

[0172] Alternatively, based on the aforementioned second example method, other implementations may provide operations associated with performing any of the aforementioned examples based on the second example system.

[0173] According to various embodiments, a second example non-transitory computer-readable medium is provided storing computer instructions that, when executed by one or more processors, cause the system to perform the following operations: applying a multiple notch filter to a signal provided for generating a sound output; receiving, in sensing circuitry, a filtered signal corresponding to the signal received and processed by the multiple notch filter or corresponding to the sound output, and using the sensing circuitry to generate statistics about the filtered signal; modifying the depth of the multiple notch filter based on the statistics using a dynamic gain unit coupled to receive the statistics from the sensing circuitry; and receiving the filtered signal and generating a sound output from the speaker.

[0174] Optionally, based on the aforementioned second example non-transitory computer-readable medium storing computer instructions, other implementations may provide operations associated with performing any of the aforementioned examples according to the second example method and / or the second example system.

[0175] According to various embodiments, a third example system may include: a multi-notch filter configured to receive a signal for generating a sound output; a sensing circuit coupled to receive the filtered signal and generate statistics about the filtered signal, the filtered signal corresponding to a signal received and processed by the multi-notch filter or a signal corresponding to a speaker output; a dynamic gain unit coupled to receive statistics from the sensing circuit and coupled to the multi-notch filter to operatively modify the depth of the multi-notch filter; a speaker for receiving the filtered signal to generate a sound output; a bass enhancement circuit arranged to provide a signal as a bass-processed signal to the multi-notch filter, the sensing circuit coupled to receive the bass-processed signal and generate statistics about the bass-processed signal; and a bass parameter controller coupled to receive statistics from the sensing circuit and coupled to the bass enhancement circuit to operatively provide parameters to the bass enhancement circuit.

[0176] Alternatively, in the foregoing example according to the third example system, another implementation provides that the sensing circuit and other modules can be implemented in the DSP.

[0177] Optionally, in any of the foregoing examples of the third example system, further embodiments provide sensing circuitry operable to generate statistics that measure the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0178] Alternatively, in any of the foregoing examples of the third example system, a further implementation provides sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence to calculate a binary indicator of distortion at the speaker output.

[0179] Alternatively, in any of the foregoing examples according to the third example system, a further implementation provides that a variable hangover time can be provided to the binary indicator to avoid the situation where the tail of the lost distorted audio segment occurs.

[0180] Alternatively, in any of the foregoing examples according to the third example system, a further implementation provides that a variable hang-before time (with a corresponding look-ahead or processing delay combined for multi-notch processing) can be provided to the binary indicator to avoid the occurrence of a leading portion of a lost distorted audio segment.

[0181] Optionally, in any of the foregoing examples according to the third example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, naive Bayes classification, linear discriminant analysis, and random forests.

[0182] Optionally, in any of the foregoing examples of the third example system, a further implementation provides sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence to calculate a soft indicator corresponding to the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0183] Optionally, in any of the foregoing examples according to the third example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression.

[0184] Alternatively, in any of the foregoing examples based on the third example system, a further implementation provides that a binary indicator can be derived from a soft indicator by using a threshold or comparator device.

[0185] Optionally, in any of the foregoing examples according to the third example system, a further implementation provides that a semi-binary or semi-hard indicator can be derived from a soft indicator by using a threshold, thereby generating a constant indicator for input indicator values ​​above / below a certain threshold, and a variable indicator for input indicator values ​​below / above a certain threshold.

[0186] Optionally, in any of the foregoing examples according to the third example system, a further implementation provides that the hysteresis can be incorporated into the derivation of the hard value of the binary or semi-binary indicator to avoid issuing a binary indication for only a brief offset of the soft indicator exceeding the threshold.

[0187] Optionally, in any of the foregoing examples of the third example system, further embodiments provide that the sensing circuit may include a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to distortion of the loudspeaker.

[0188] Optionally, in any of the foregoing examples of the third example system, a further embodiment provides that the sensing circuit may include circuitry for estimating the amount of distortion activation energy in the filtered signal as a statistic calculated from the energy across all frequency bands of a plurality of bandpass filters, and the sensing circuitry includes circuitry for measuring the total energy of the filtered signal.

[0189] Optionally, in any of the foregoing examples based on the third example system, a further implementation provides that the statistic can be the median.

[0190] Optionally, in any of the foregoing examples based on the third example system, a further implementation provides that the statistic can be the average.

[0191] Optionally, in any of the foregoing examples based on the third example system, a further implementation provides that the average value can be an arithmetic average.

[0192] Optionally, in any of the foregoing examples according to the third example system, a further implementation provides that the average value may be a geometric average.

[0193] Optionally, in any of the foregoing examples based on the third example system, a further implementation provides that the statistical value can be the maximum value.

[0194] Optionally, in any of the foregoing examples based on the third example system, a further implementation provides that the statistical value can be the minimum value.

[0195] Alternatively, in any of the foregoing examples according to the third example system, a further implementation specifies that the statistic may be the energy of the nth largest energy among all calculated bandpass energies.

[0196] Alternatively, in any of the foregoing examples of the third example system, a further implementation provides a sensing circuit capable of calculating the time-varying energy or envelope of a full-band audio signal.

[0197] Optionally, in any of the foregoing examples of the third example system, a further implementation provides a bass enhancement circuit operable to perform one or more bass enhancements selected from the group comprising bass enhancement based on linear filtering and bass enhancement based on the insertion and / or generation of additional harmonics of the signal carrying the audio content.

[0198] Alternatively, in any of the foregoing examples of the third example system, a further implementation provides that the bass enhancement circuit may be an adaptive harmonic bass enhancement circuit that employs tunable asymmetric nonlinearity and pre-cutoff filtering and post-cutoff filtering to change the amount of harmonic energy added to the received signal.

[0199] Alternatively, in any of the foregoing examples according to the third example system, a further implementation provides that the bass parameter controller can be arranged to operatively provide parameters to control the bass amplitude and bass harmonic level.

[0200] Optionally, in any of the foregoing examples of the third example system, a further implementation provides that the signal is a digital signal.

[0201] According to various embodiments, a third example method may include: applying a multiple notch filter to a signal provided for generating a sound output; receiving the filtered signal in a sensing circuit and generating statistics about the filtered signal using the sensing circuit, the filtered signal corresponding to a signal received and processed by the multiple notch filter; modifying the depth of the multiple notch filter based on the statistics using a dynamic gain unit coupled to receive the statistics from the sensing circuit; and receiving the filtered signal in a speaker and generating a sound output from the speaker, wherein applying the signal to the multiple notch filter may include providing the signal as a bass-processed signal obtained through processing by a bass enhancement circuit, the bass enhancement circuit processing may include generating statistics about the bass-processed signal using the sensing circuit; generating parameters based on the statistics about the bass-processed signal; and using the parameters to provide bass enhancement to the signal applied to the multiple notch filter.

[0202] Alternatively, based on the aforementioned third example method, other implementations may provide operations associated with performing any of the aforementioned examples based on the third example system.

[0203] According to various embodiments, a third example non-transitory computer-readable medium is provided storing computer instructions that, when executed by one or more processors, cause a system to perform the following operations: applying a multi-notch filter to a signal provided for generating a sound output; receiving a filtered signal corresponding to the signal received and processed by the multi-notch filter in a sensing circuit and generating statistics about the filtered signal using the sensing circuit; modifying the depth of the multi-notch filter based on the statistics using a dynamic gain unit coupled to receive the statistics from the sensing circuit; and receiving the filtered signal in a speaker and generating a sound output from the speaker, wherein applying the signal to the multi-notch filter may include providing the signal as a bass-processed signal obtained by processing by a bass enhancement circuit, the processing of which may include generating statistics about the bass-processed signal using the sensing circuit; generating parameters based on the statistics about the bass-processed signal; and using the parameters to provide bass enhancement to the signal applied to the multi-notch filter.

[0204] Optionally, based on the aforementioned third example non-transitory computer-readable medium storing computer instructions, other implementations may provide operations associated with performing any of the aforementioned examples according to the third example method and / or the third example system.

[0205] According to various embodiments, the fourth example system may include: a bass enhancement circuit configured to receive a signal for generating a sound output; a sensing circuit coupled to receive a bass-enhanced signal corresponding to the signal received and processed by the bass enhancement circuit or corresponding to the sound output, and generating statistics about the bass-enhanced signal or the speaker output signal; a bass parameter controller coupled to receive statistics from the sensing circuit and coupled to the bass enhancement circuit to operatively provide parameters to the bass enhancement circuit; and a speaker that receives the bass-enhanced signal to generate a sound output.

[0206] Alternatively, in the foregoing example according to the fourth example system, another implementation provides that the sensing circuitry and other modules can be implemented in the DSP.

[0207] Optionally, in any of the foregoing examples of the fourth example system, further embodiments provide sensing circuitry operable to generate statistics that measure the probability of distortion at the speaker output or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0208] Alternatively, in any of the foregoing examples of the fourth example system, a further implementation provides sensing circuitry operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, or artificial intelligence to calculate a binary indicator of distortion at the speaker output.

[0209] Alternatively, in any of the foregoing examples according to the fourth example system, a further implementation provides that a variable hangover time can be provided to the binary indicator to avoid the situation where the tail of the distorted audio segment is lost.

[0210] Alternatively, in any of the foregoing examples according to the fourth example system, a further implementation provides that a variable hang-before time (with a corresponding look-ahead or processing delay combined for multi-notch processing) can be provided to the binary indicator to avoid the occurrence of a leading portion of a lost distorted audio segment.

[0211] Optionally, in any of the foregoing examples according to the fourth example system, further implementations provide that the technique can use one or more processes selected from the group including classification and regression trees, support vector machines, artificial neural networks, logistic regression, naive Bayes classification, linear discriminant analysis, and random forests.

[0212] Optionally, in any of the foregoing examples of the fourth example system, a further implementation specifies that the sensing circuitry is operable to use techniques selected from the group including machine learning, statistical learning, predictive learning, and artificial intelligence to calculate a soft indicator corresponding to the probability of distortion or the degree of offensive, perceptible, or measurable distortion at the speaker output.

[0213] Optionally, in any of the foregoing examples according to the fourth example system, further implementations provide that the technique can use one or more processes selected from the group consisting of classification and regression trees, ordinary least squares regression, weighted least squares regression, support vector regression, artificial neural networks, piecewise linear regression, ridge regression, lasso regression, elastic network regression, and nonlinear regression.

[0214] Alternatively, in any of the foregoing examples of the fourth example system, a further implementation provides that a binary indicator can be derived from a soft indicator by using a threshold or comparator device.

[0215] Optionally, in any of the foregoing examples of the fourth example system, a further implementation provides that a semi-binary or semi-hard indicator can be derived from a soft indicator using a threshold, thereby generating a constant indicator for input indicator values ​​above / below a certain threshold, and a variable indicator for input indicator values ​​below / above a certain threshold.

[0216] Alternatively, in any of the foregoing examples according to the fourth example system, a further implementation provides that the hysteresis can be incorporated into the derivation of the hard value of the binary or semi-binary indicator to avoid issuing a binary indication for only a brief offset of the soft indicator exceeding the threshold.

[0217] Optionally, in any of the foregoing examples of the fourth example system, further embodiments provide that the sensing circuitry may include a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectrum analysis filters tuned to frequencies determined to contribute to distortion of the loudspeaker.

[0218] Optionally, in any of the foregoing examples of the fourth example system, a further embodiment provides that the sensing circuit may include circuitry for estimating the amount of distortion activation energy in the filtered signal as a statistic calculated from the energy across all frequency bands of multiple bandpass filters, and the sensing circuitry includes circuitry for measuring the total energy of the filtered signal.

[0219] Optionally, in any of the foregoing examples based on the fourth example system, a further implementation provides that the statistic can be the median.

[0220] Optionally, in any of the foregoing examples based on the fourth example system, a further implementation provides that the statistic can be the average.

[0221] Optionally, in any of the foregoing examples according to the fourth example system, a further implementation provides that the average value can be an arithmetic average.

[0222] Optionally, in any of the foregoing examples according to the fourth example system, a further implementation provides that the average value may be a geometric average.

[0223] Optionally, in any of the foregoing examples based on the fourth example system, a further implementation provides that the statistical value can be the maximum value.

[0224] Optionally, in any of the foregoing examples based on the fourth example system, a further implementation provides that the statistical value can be the minimum value.

[0225] Alternatively, in any of the foregoing examples according to the fourth example system, a further implementation specifies that the statistic may be the energy of the nth largest energy among all calculated bandpass energies.

[0226] Optionally, in any of the foregoing examples of the fourth example system, further embodiments provide a bass enhancement circuit operable to perform one or more bass enhancements selected from the group comprising bass enhancement based on linear filtering and bass enhancement based on the insertion and / or generation of additional harmonics of the signal carrying the audio content.

[0227] Alternatively, in any of the foregoing examples of the fourth example system, a further implementation provides that the bass enhancement circuit may be an adaptive harmonic bass enhancement circuit that employs tunable asymmetric nonlinearity and pre-cutoff filtering and post-cutoff filtering to change the amount of harmonic energy added to the received signal.

[0228] Alternatively, in any of the foregoing examples of the fourth example system, a further implementation provides that the bass parameter controller can be arranged to operatively provide parameters to control the bass amplitude and bass harmonic level.

[0229] According to various embodiments, the fourth example method may include: receiving a signal at a bass enhancement circuit, the signal being provided for generating a sound output; receiving a bass-enhanced signal at a sensing circuit corresponding to a signal received and processed by the bass enhancement circuit or corresponding to the sound output, and using the sensing circuit to generate statistics about the bass-enhanced signal; generating parameters in a bass parameter controller based on the statistics about the bass-processed signal, and using the parameters to provide bass enhancement for the signal applied to the bass enhancement circuit; and receiving the bass-enhanced signal at a speaker and generating a sound output from the speaker.

[0230] Alternatively, based on the aforementioned fourth example method, other implementations may provide operations associated with performing any of the aforementioned examples according to the fourth example system.

[0231] According to various embodiments, a fourth example non-transitory computer-readable medium is provided storing computer instructions that, when executed by one or more processors, cause a system to perform the following operations: receiving a signal at a bass enhancement circuit, the signal being provided for generating a sound output; receiving a bass-enhanced signal at a sensing circuit corresponding to a signal received and processed by the bass enhancement circuit or corresponding to the sound output, and using the sensing circuit to generate statistics about the bass-enhanced signal; generating parameters in a bass parameter controller based on the statistics about the bass-processed signal, and using the parameters to provide bass enhancement for a signal applied to the bass enhancement circuit; and receiving the bass-enhanced signal at a speaker and generating a sound output from the speaker.

[0232] Optionally, based on the aforementioned fourth example non-transitory computer-readable medium storing computer instructions, other implementations may provide operations associated with performing any of the aforementioned examples according to the fourth example method and / or the fourth example system.

[0233] According to various embodiments, the fifth example system may include: a bass enhancement circuit configured to receive a signal for generating a sound output; a multi-notch filter configured to receive the bass-enhanced signal or a processed form of the bass-enhanced signal, and configured to provide a filtered bass-enhanced signal; a sensing circuit coupled to receive the output of the multi-notch filter to receive the filtered bass-enhanced signal corresponding to or corresponding to the sound output of the signal received and processed by the bass enhancement circuit and the multi-notch filter, and to generate statistics about the filtered bass-enhanced signal; a bass parameter controller coupled to receive statistics from the sensing circuit and coupled to the bass enhancement circuit to operatively provide parameters to the bass enhancement circuit to enhance the signal; a dynamic gain unit coupled to receive statistics from the sensing circuit and coupled to the multi-notch filter to operatively modify the depth of the multi-notch filter; and a speaker that receives the filtered bass-enhanced signal to generate a sound output.

[0234] Alternatively, in the foregoing example according to the fifth example system, another implementation provides that the sensing circuitry and other modules can be implemented in the DSP.

[0235] Alternatively, in any of the foregoing examples of the fifth example system, a further implementation provides that the amplifier and / or post-processor can be coupled between the multi-notch filter and the speaker.

[0236] Alternatively, in any of the foregoing examples of the fifth example system, a further implementation provides that the audio processor can be coupled between the bass enhancement circuitry and the multi-notch filter.

[0237] According to various embodiments, a fifth example method may include: receiving a signal at a bass enhancement circuit and generating a bass-enhanced signal, the signal being provided for generating a sound output; receiving the bass-enhanced signal or a processed form of the bass-enhanced signal in a multi-notch filter, the multi-notch filter being configured to provide a filtered bass-enhanced signal; receiving at a sensing circuit a signal corresponding to the signal received and processed by the bass enhancement circuit and the multi-notch filter or a filtered bass-enhanced signal corresponding to the sound output, and using the sensing circuit to generate statistics about the filtered bass-enhanced signal; generating parameters in a bass parameter controller based on the statistics of the filtered bass-processed signal, and using the parameters to provide bass enhancement to a signal applied to the bass enhancement circuit; receiving statistics from the sensing circuit in a dynamic gain unit, and modifying the depth of the multi-notch filter using the dynamic gain unit based on the received statistics; generating the filtered bass-enhanced signal from the multi-notch filter having the modified depth; and receiving the filtered bass-enhanced signal at a speaker and generating a sound output from the speaker.

[0238] Alternatively, based on the aforementioned fifth example method, other implementations may provide operations associated with performing any of the aforementioned examples according to the fifth example system.

[0239] According to various embodiments, a fifth example non-transitory computer-readable medium is provided storing computer instructions that, when executed by one or more processors, cause a system to perform the following operations: receiving a signal at a bass enhancement circuit and generating a bass-enhanced signal provided for generating a sound output; receiving the bass-enhanced signal or a processed form of the bass-enhanced signal in a multi-notch filter configured to provide a filtered bass-enhanced signal; receiving at a sensing circuit a signal corresponding to the signal received and processed by the bass enhancement circuit and the multi-notch filter or a filtered bass-enhanced signal corresponding to the sound output, and using the sensing circuit to generate statistics about the filtered bass-enhanced signal; generating parameters in a bass parameter controller based on the statistics about the filtered bass-processed signal, and using the parameters to provide bass enhancement to a signal applied to the bass enhancement circuit; receiving statistics from the sensing circuit in a dynamic gain unit and modifying the depth of the multi-notch filter using the dynamic gain unit based on the received statistics; generating the filtered bass-enhanced signal from the multi-notch filter having the modified depth; and receiving the filtered bass-enhanced signal at a speaker and generating a sound output from the speaker.

[0240] Optionally, based on the aforementioned fifth example non-transitory computer-readable medium storing computer instructions, other implementations may provide operations associated with performing any of the aforementioned examples according to the fifth example method and / or the fifth example system.

[0241] By studying this document, many other variations of the content described herein will become apparent. For example, depending on the embodiment, certain actions, events, or functions of any methods and algorithms described herein may be performed in a different order, may be added, combined, or omitted together (not all described actions or events are necessary for the implementation of the methods and algorithms). Furthermore, in some embodiments, actions or events may be performed concurrently, for example, through multithreading, interrupt handling, or on multiple processors or processor cores or other parallel architectures, rather than sequentially. Additionally, different tasks or processes may be performed by different machines and computing systems that can work together.

[0242] The various illustrative logic blocks, modules, methods, and algorithmic processes and sequences described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, and process actions have been generally described above in terms of functionality. Whether this functionality is implemented as hardware or software depends on the specific application and design constraints for the overall system. The described functionality may be implemented differently for each specific application, but such implementation decisions should not be construed as departing from the scope of this document.

[0243] Embodiments of the systems and methods described herein can operate in a variety of general-purpose or special-purpose computing system environments or configurations. Typically, a computing environment can include any type of computer system, including but not limited to computer systems based on one or more microprocessors, mainframe computers, digital signal processors, portable computing devices, personal organizers, device controllers, computing engines within a facility, mobile phones, desktop computers, mobile computers, tablet computers, smartphones, and facilities with embedded computers, to name just a few.

[0244] Such computing devices can typically be found in devices with at least some minimum computing capabilities, including but not limited to personal computers, server computers, handheld computing devices, laptop or mobile computers, communication devices such as cellular phones and PDAs, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, audio or video media players, and so on. In some embodiments, the computing device will include one or more processors. Each processor may be a dedicated microprocessor, such as a DSP, a very long instruction word (VUW) or other microcontroller, or may be a conventional central processing unit (CPU) with one or more processing cores, including dedicated graphics processing unit (GPU)-based cores in a multi-core CPU.

[0245] The procedural actions of the methods, processes, or algorithms described in conjunction with the embodiments disclosed herein may be directly embodied in hardware, in a software module executed by a processor, or in any combination of both. The software module may be contained in a computer-readable medium accessible by a computing device. The computer-readable medium may include volatile and non-volatile media, which may be removable, non-removable, or some combination thereof. The computer-readable medium is used to store information such as computer-readable or computer-executable instructions, data structures, program modules, or other data. By way of example and not limitation, the computer-readable medium may include computer storage media and communication media.

[0246] Computer storage media include, but are not limited to, computer or machine-readable media or storage devices such as Blu-ray Disc (BD), Digital Universal Disc (DVD), Compact Disc (CD), floppy disk, magnetic tape drive, hard disk drive, optical drive, solid-state storage device, RAM memory, ROM memory, EPROM memory, EEPROM memory, flash memory or other memory technology, magnetic tape cassette, magnetic tape, disk storage or other magnetic storage device, or any other device that can be used to store desired information and can be accessed by one or more computing devices.

[0247] Software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of non-transitory computer-readable storage medium, media, or physical computer memory known in the art. An exemplary storage medium may be coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. Alternatively, the processor and storage medium may reside as discrete components in the user terminal.

[0248] As used in this document, the phrase “non-transitory” means “persistent or long-lasting.” The phrase “non-transitory computer-readable medium” includes any and all computer-readable media, with the sole exception of transient propagation signals. By way of example and not limitation, this includes non-transitory computer-readable media such as register memory, processor cache, and random access memory (RAM).

[0249] The retention of information such as computer-readable or computer-executable instructions, data structures, and program modules can also be accomplished by encoding one or more modulated data signals, electromagnetic waves (e.g., carrier waves), or other transmission mechanisms or communication protocols using various communication media, including any wired or wireless information transmission mechanism. Generally, these communication media refer to signals in which one or more characteristics are set or altered to encode information or instructions. For example, communication media include wired media, such as wired networks or direct wired connections carrying one or more modulated data signals, and wireless media, such as acoustic, radio frequency (RF), infrared, laser, and other wireless media for transmitting, receiving, or both of these modulated data signals or electromagnetic waves. Any combination of the foregoing should also be included within the scope of communication media.

[0250] Furthermore, one or more of the software, programs, computer program products, or any combination thereof embodying the various embodiments of the distortion sensing / distortion reduction system and / or bass management system and methods or portions thereof described herein may be stored, received, transmitted, or read from any desired combination of computer or machine-readable media or storage devices and communication media in the form of computer-executable instructions or other data structures.

[0251] Embodiments of the systems and methods described herein can be further described in the general context of computer-executable instructions (e.g., program modules) that are executed by computing devices. Typically, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The embodiments described herein can also be implemented in a distributed computing environment, where the task is performed by one or more remote processing devices, or within a cloud of one or more devices linked through one or more communication networks. In a distributed computing environment, program modules can reside on both local and remote computer storage media, including media storage devices. Furthermore, the aforementioned instructions can be implemented, in part or in whole, as hardware logic circuitry, which may include a processor.

[0252] Unless otherwise stated or otherwise understood in the context in which they are used, the conditional language used herein, particularly words such as “may,” “possibly,” “can,” “e.g.,” etc., is generally intended to indicate that certain embodiments include certain features, elements, and / or states. Therefore, such conditional language is not generally intended to imply that one or more embodiments require features, elements, and / or states in any way, or that one or more embodiments must include logic for determining whether such features, elements, and / or states are included or to be performed in any particular embodiment. The terms “comprising,” “including,” “having,” etc., are synonymous and used inclusively in an open-ended manner and do not exclude other elements, features, actions, operations, etc. Furthermore, the term “or” is used in its inclusive sense (but not in its exclusive sense), and thus, when used, for example, to connect a list of elements, the term “or” refers to one, some, or all of the elements in the list. While the above detailed description has shown, described, and pointed out novel features applicable to various embodiments, it should be understood that various omissions, substitutions, and changes may be made in the form and detail of the described devices or algorithms without departing from the scope of this disclosure.

[0253] As will be appreciated, certain embodiments of the invention described herein may be embodied in a form that does not provide all the features and benefits set forth herein, as some features may be used or implemented separately from other features. Furthermore, although the subject matter has been described in language specific to structural features and methodological actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions disclosed above are disclosed as exemplary forms of implementing the claims.

Claims

1. A system for processing an audio signal, comprising: a multi-notch filter configured to receive a signal for generating a sound output; a sensing circuit coupled to receive a filtered signal corresponding to the signal received and processed by the multi-notch filter or to the sound output and to generate statistics about the filtered signal; a dynamic gain unit coupled to receive the statistics from the sensing circuit and coupled to the multi-notch filter to operatively modify a depth of the multi-notch filter, the dynamic gain unit having inputs to receive a measure of bandpass energy and a measure of signal energy from the sensing circuit and having a threshold of bandpass energy at which the multi-notch filter begins to engage, such that the operative modification of the depth includes operatively modifying the depth by a difference between the received measure of bandpass energy and the threshold; and a loudspeaker to receive the filtered signal to generate the sound output.

2. The system of claim 1, wherein, The multi-notch filter, the sensing circuit, the dynamic gain unit are disposed in a digital signal processor.

3. The system of claim 1, wherein, The sensing circuit is operable to generate statistics that measure a likelihood of distortion at an output of the loudspeaker.

4. The system of claim 3, wherein, The sensing circuit is operable to compute a measure of the likelihood of distortion at the output of the loudspeaker using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence.

5. The system of claim 3, wherein, The sensing circuit is operable to compute a binary indicator of distortion at the output of the loudspeaker using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, or artificial intelligence.

6. The system of claim 1, wherein, The sensing circuit includes a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectral analysis filters tuned to frequencies determined to contribute to distortion of the loudspeaker, and the sensing circuit includes circuitry to estimate an amount of distortion active energy in the filtered signal from the plurality of IIR or FIR bandpass filters and / or dynamic spectral analysis filters as one or more statistics of energy across all frequency bands of the plurality of IIR or FIR bandpass filters and / or dynamic spectral analysis filters, and the sensing circuit includes circuitry to measure a total energy of the filtered signal.

7. The system of claim 6, wherein, The dynamic gain unit is configured to provide a gain to adjust the depth of the multi-notch filter based on the one or more statistics.

8. A method for processing an audio signal, comprising: applying a multi-notch filter to a signal provided for generating a sound output; receiving at a sensing circuit a filtered signal corresponding to the signal received and processed by the multi-notch filter or to the sound output and generating statistics about the filtered signal using the sensing circuit; modifying the depth of the multi-notch filter based on statistics, including a measure of bandpass energy and a measure of signal energy, using a dynamic gain unit coupled to receive the statistics from the sensing circuit, the dynamic gain unit having a sensitivity parameter and a threshold of bandpass energy at which the multi-notch filter begins to engage, such that modifying the depth includes operatively modifying the depth by a difference between the measure of received bandpass energy and the threshold; and receiving the filtered signal at a speaker and generating the sound output from the speaker.

9. The method of claim 8, wherein, receiving a filtered signal processed by the multi-notch filter, generating statistics about the filtered signal, and modifying the depth of the multi-notch filter are implemented in a digital signal processor.

10. The method of claim 8, wherein, Generating statistics includes generating statistics that measure a likelihood of distortion at an output of the speaker.

11. The method of claim 10, wherein, The method includes using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence in the sensing circuit to compute a measure of a likelihood of distortion at an output of the speaker.

12. The method of claim 10, wherein, The method includes using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, or artificial intelligence in the sensing circuit to compute a binary indicator of distortion at an output of the speaker.

13. The method of claim 8, wherein, Using the sensing circuit to generate statistics includes using a sensing circuit that includes a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectral analysis filters tuned to frequencies determined to contribute to distortion of the speaker, estimating an amount of distortion activation energy in filtered signals from the plurality of IIR bandpass filters and / or dynamic spectral analysis filters as one or more statistics of energy across all frequency bands of the plurality of IIR or FIR bandpass filters and / or dynamic spectral analysis filters, and measuring a total energy of the filtered signals.

14. The method of claim 13, wherein, Using the dynamic gain unit includes providing a gain to adjust the depth of the multi-notch filter based on the one or more statistics. modifying the depth of the multi-notch filter based on statistics, including a measure of bandpass energy and a measure of signal energy, using a dynamic gain unit coupled to receive the statistics from the sensing circuit, the dynamic gain unit having a sensitivity parameter and a threshold of bandpass energy at which the multi-notch filter begins to engage, such that modifying the depth includes operatively modifying the depth by a difference between the measure of received bandpass energy and the threshold; and receiving the filtered signal at a speaker and generating the sound output from the speaker. receiving a filtered signal processed by the multi-notch filter, generating statistics about the filtered signal, and modifying the depth of the multi-notch filter are implemented in a digital signal processor. Generating statistics includes generating statistics that measure a likelihood of distortion at an output of the speaker. The method includes using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, and artificial intelligence in the sensing circuit to compute a measure of a likelihood of distortion at an output of the speaker. The method includes using a technique selected from the group consisting of machine learning, statistical learning, predictive learning, or artificial intelligence in the sensing circuit to compute a binary indicator of distortion at an output of the speaker. Using the sensing circuit to generate statistics includes using a sensing circuit that includes a plurality of infinite impulse response (IIR) or finite impulse response (FIR) bandpass filters and / or dynamic spectral analysis filters tuned to frequencies determined to contribute to distortion of the speaker, estimating an amount of distortion activation energy in filtered signals from the plurality of IIR bandpass filters and / or dynamic spectral analysis filters as one or more statistics of energy across all frequency bands of the plurality of IIR or FIR bandpass filters and / or dynamic spectral analysis filters, and measuring a total energy of the filtered signals. Using the dynamic gain unit includes providing a gain to adjust the depth of the multi-notch filter based on the one or more statistics.

Images