The invention relates to a method for filtering an input signal (3b, 4b, 5b) relative to a physical variable of a turbine engine (9), the input signal being digitised, the method implementing frequency filtering of said signal in a computer (6) of a control system (7) of said turbine engine (9), said signal being provided at the input of the computer, a digital derivative of said signal being intended for being used by the control system (7), characterised in that it involves: —detecting an amplitude variation of said variable on said input signal, by a step of generating a second derivative signal (S) of the input signal and a step of comparing a value of the second derivative value of the input signal with at least one predetermined threshold (S.sub.1 . . . S.sub.n); and —adapting the frequency filtering of said input signal as a function of the detected amplitude variation of said variable, by a step of controlling a controlled filter (PB.sub.11) capable of applying frequency filtering to the input signal, so that the controlled filter applies or does not apply the frequency filtering as a function of a result of the comparison step.

The present disclosure relates to a concept of nonlinear signal processing which may be used for predistortion for RF power amplifiers. The concept includes generating time variant filter coefficients for a linear filter circuit based on a nonlinear mapping of an input signal, and filtering the input signal with the linear filter circuit using the time variant filter coefficients in order to generate a filtered output signal. Thus, it is proposed to implement a non-linear filter by a time-varying linear filter where the time-varying coefficients are derived from the input signal.

The exemplary embodiments provide a parallel implementation of filters with recursive transfer functions. This can enable a filter to act as a frame filter that may process a frame of multiple samples of data in parallel rather than being limited to processing a single sample of data at a time. Each frame contains plural input samples of data values. The input samples are from a common source and have a time dependency. The exemplary embodiments are suitable for implementing various types of filters in parallel, such as cascaded integrator comb filters, biquad filters and other types of infinite impulse response (IIR) filters. The exemplary embodiments may use polyphase decomposition to decompose a filter with a recursive transfer function into multiple polyphase component filters. The polyphase component filters may be applied to respective samples of data in a parallel pipelined configuration to produce filtered output for the samples of data in parallel.

Example embodiments disclosed herein relate to filter coefficient updating in time domain filtering. A method of processing an audio signal is disclosed. The method includes obtaining a predetermined number of target gains for a first portion of the audio signal by analyzing the first portion of the audio signal. Each of the target gains is corresponding to a subband of the audio signal. The method also includes determining filter coefficients for time domain filtering the first portion of the audio signal so as to approximate a frequency response given by the target gains. The filter coefficients are determined by iteratively selecting at least one target gain from the target gains and updating the filter coefficient based on the selected at least one target gain. Corresponding system and computer program product for processing an audio signal are also disclosed.

This disclosure describes systems, methods, and apparatus for a digital-to-analog (DAC) converter, that can be part of a variable capacitor and/or a match network. The DAC can include a digital input, an analog output, N contributors (e.g., switched capacitors), and an interconnect topology connecting the N contributors, generating a sum of their contributions (e.g., sum of capacitances), and providing the sum to the analog output. The N contributors can form a sub-binary sequence when their contributions to the sum are ordered by average contribution. Also, the gap size between a maximum contribution of one contributor, and a minimum contribution of a subsequent contributor, is less than D, where D is less than or equal to two time a maximum contribution of the first or smallest of the N contributors.

A method for acoustic source separation comprises inputting acoustic data from a plurality of acoustic sensors, combined from a plurality of acoustic sources, converting the acoustic data to time-frequency domain data comprising time-frequency data frames, and constructing a multichannel filter for the time-frequency data frames to separate signals from the acoustic sources. The constructing comprises determining a set of de-mixing matrices (W.sub.f) to apply to each time-frequency data frame to determine a vector of separated outputs (y.sub.ft) by modifying each of the de-mixing matrices by a respective gradient value (G;G′) for a frequency dependent upon a gradient of a cost function measuring a separation of the sources by the respective de-mixing matrix. The respective gradient values for each frequency are each calculated from a stochastic selection of the time-frequency data frames.

A noise reduction system includes sensors configured to generate an input signal, an adaptive filter configured to represent a transfer function of a path traversed by the input signal, one or more processing devices, and one or more transducers. The processing devices receive the input signal and generate an updated set of filter coefficients of the adaptive filter by separating the input signal into frequency subbands; determining for each subband, coefficients of a corresponding subband adaptive module; and combining the coefficients of multiple subband adaptive modules. Determining the coefficients of the corresponding subband adaptive module includes selecting a subset of a precomputed set of filter coefficients of the adaptive filter. The processing devices process a portion of the input signal using the updated set of filter coefficients of the adaptive filter to generate an output that destructively interferes with another signal traversing the path represented by the transfer function.

An adaptive processor implements partial updates when it adjusts weights to optimize adaptation criteria in signal estimation, parameter estimation, or data dimensionality reduction algorithms. The adaptive processor designates some of the weights to be update weights and the other weights to be held weights. Unconstrained updates are performed on the update weights, whereas updates to the set of held weights are performed within a reduced-dimensionality subspace. Updates to the held weights and the update weights employ adapt-path operations for tuning the adaptive processor to process signal data during or after tuning.

An inference processing apparatus includes an input data storage unit that stores pieces X of input data, a learned NN storage unit that stores a piece W of weight data of a neural network, a batch processing control unit that sets a batch size on the basis of information on the pieces X of input data, a memory control unit that reads out, from the input data storage unit, the pieces X of input data corresponding to the set batch size, and an inference operation unit that batch-processes operation in the neural network using, as input, the pieces X of input data corresponding to the batch size and the piece W of weight data and infers a feature of the pieces X of input data.

The present disclosure relates to an artificial intelligence (AI) system which simulates functions such as cognition, judgment, and the like of the human brain by utilizing machine learning algorithms such as deep learning and the like, and to an application thereof. According to various embodiments, an electronic device may comprise: a first impedance matching circuit configured to perform a first impedance matching on a power signal wirelessly received from a wireless power transmission device; a second impedance matching circuit configured to perform a second impedance matching on the first impedance-matched power signal using any one impedance value among a plurality of impedance values; a control circuit configured to perform control to change an impedance value of the second impedance matching circuit to an impedance value learned using an impedance matching network model, corresponding to a power and a frequency of the second impedance-matched power signal; and a power conversion circuit configured to convert a second impedance-matched power signal in an AC form into a power in a DC form for charging a battery according to the changed impedance value.