Abstract: This drone includes an image sensor configured to take an image of a scene including a plurality of objects, and an electronic determination device including an electronic detection module configured to detect, via a neural network, in the image taken by the image sensor, a representation of a potential target from among the plurality of objects represented, an input variable of the neural network being an image depending on the image taken, at least one output variable of the neural network being an indication relative to the representation of the potential target. A first output variable of the neural network is a set of coordinates defining a contour of a zone surrounding the representation of the potential target.

Abstract: The method comprises the steps of: digitizing sound signals picked up simultaneously by two microphones (N, M); executing a short-term Fourier transform on the signals (xn(t), xm(t)) picked up on the two channels so as to produce a succession of frames in a series of frequency bands; applying an algorithm for calculating a speech-presence confidence index on each channel, in particular a probability a speech that is present; selecting one of the two microphones by applying a decision rule to the successive frames of each of the channels, which rule is a function both of a channel selection criterion and of a speech-presence confidence index; and implementing speech processing on the sound signal picked up by the one microphone that is selected.

Abstract: The method comprises, in the frequency domain: estimating (18), for each frequency band of the spectrum (Y(k,l)) of each current time frame (y(k)), a speech presence probability in the signal (p(k,l)); calculating (16) a spectral gain (GOMLSA(k,l)), proper to each frequency band of each current time frame, as a function i) of an estimation of the noise energy in each frequency band, ii) of the speech presence probability estimated at step c1), and iii) of a scalar minimal gain value; and selectively reducing the noise (14) by applying the calculated gain at each frequency band. The scalar minimal gain value, representative of a parameter of soundproofing hardness, is a value (Gmin(k)) that can be dynamically modulated at each successive time frame, calculated for the current time frame as a function of a global variable linked to this current time frame with application of an increment/decrement to a parameterized nominal value (Gmin) of the minimal gain.

Abstract: The method comprises the steps of: digitizing sound signals picked up simultaneously by two microphones (N, M); executing a short-term Fourier transform on the signals (xn(t), xm(t)) picked up on the two channels so as to produce a succession of frames in a series of frequency bands; applying an algorithm for calculating a speech-presence confidence index on each channel, in particular a probability a speech that is present; selecting one of the two microphones by applying a decision rule to the successive frames of each of the channels, which rule is a function both of a channel selection criterion and of a speech-presence confidence index; and implementing speech processing on the sound signal picked up by the one microphone that is selected.

Abstract: The device comprises a microphone for detecting a speech signal from a near speaker, and a loudspeaker for reproducing a speech signal from a remote speaker. The processing for canceling the interfering acoustic echo implements an adaptive linear filtering algorithm. Double talk situations are detected by: evaluating an index representative of the convergence or divergence of the algorithm; assessing a predetermined condition for detecting a double talk situation; and if the condition is satisfied, modifying at least one parameter of the algorithm in response to the detection. The representative index may be the norm of the gradient vector describing the adaptation of the filter from one iteration of the algorithm to the next, the conditions being a comparison between the gradient and a threshold. The parameter that is modified double talk situation may be the adaptation stepsize of the algorithm, and also the gain control of an echo suppression stage.

Abstract: The device comprises a microphone for picking up a speech signal from a near speaker, and a loudspeaker for reproducing a speech signal from a remote speaker. The processing for canceling the interfering acoustic echo implements an adaptive linear filtering algorithm. Double talk situations are detected by: evaluating an index representative of the degree of convergence or divergence of the algorithm; assessing a predetermined condition for detecting a double talk situation; and if said condition is satisfied, modifying at least one parameter of the algorithm in response to said detection. The representative index may be the norm of the gradient vector describing the adaptation of the filter from one iteration of the algorithm to the next, the conditions being a comparison between the gradient and a threshold. The parameter that is modified double talk situation may be the adaptation stepsize of the algorithm, and also the gain control of an echo suppression stage.