Abstract: A multi-channel transformer acoustic model that processes a plurality of audio signals output by microphones of a microphone array and outputs probabilities for acoustic units of an utterance represented in the audio signals. The audio signals represent the individual microphones' respective capturing of the utterance. The multi-channel model may perform self-attention on embeddings of the audio signals and then cross-channel attention across the attended audio signals. The cross-channel attention may involve processing of signals relative to each other to model the relationships across channels within and across time frames. The multi-channel model may include a transducer to perform processing frame-by-frame.

Abstract: A system for the detection and location of acoustic signals which can be used, for example, for the acquisition of voice messages or the like, in environments in which noises, echoes and reverberations are present. The system employs an array of microphones and is based on the Fourier antitransform calculus of only the information of phases of the normalised cross power spectrum of pairs of signals acquired from the microphones in the array. The system also enables an acoustic message cleared of the undesired components which are due to noises, echoes, etc to be reconstructed.