Abstract: A divisive normalization method, device, audio feature extractor, and a chip are disclosed. For improving the robustness against variations of the background noise, the proposed per-channel divisive normalization method comprises: obtaining an average number of input spikes over an averaging window by a low-pass filter to produce a threshold parameter, then deciding whether to enable an integrate-and fire (IAF) counter counting over a clock period of a divisive normalization module, when the count value of IAF counter reaches the threshold, resetting IAF counter and producing a single spike at the output. Compared with the prior art, the proposed method does not require any Linear Feedback Shift Register (LFSR) and is easier to implement. It also has a simpler structure, higher accuracy and better statistical performance, and a lower cost and a lower power consumption.

Abstract: A computer-implemented method for determining the Walsh-Hadamard transform of N samples of a signal, comprises electing a plurality of hashing C matrices ?1, . . . ?C, computing C hashes of a particular length based at least on a hashing front end and the plurality of matrices ?1, . . . ?C, forming a bipartite graph with a plurality of variable nodes and a plurality of check nodes, the variable nodes being non-zero coefficients to recover, and the check nodes being hashed samples, finding an isolated check node and recovering non-zero coefficients connected to the isolated check node by employing collision detection and support estimation of the signal, peeling from the bipartite graph the recovered non-zero coefficients, and repeating the computing step, forming step, finding step and peeling step until all the nodes in the plurality of check nodes are zero.

Abstract: A computer-implemented method for determining the Walsh-Hadamard transform of N samples of a signal, comprises electing a plurality of hashing C matrices ?1, . . . ?C, computing C hashes of a particular length based at least on a hashing front end and the plurality of matrices ?1, . . . ?C, forming a bipartite graph with a plurality of variable nodes and a plurality of check nodes, the variable nodes being non-zero coefficients to recover, and the check nodes being hashed samples, finding an isolated check node and recovering non-zero coefficients connected to the isolated check node by employing collision detection and support estimation of the signal, peeling from the bipartite graph the recovered non-zero coefficients, and repeating the computing step, forming step, finding step and peeling step until all the nodes in the plurality of check nodes are zero.