Systems and Methods for Source Signal Separation

Abstract

A method of processing a signal, including taking a signal formed from a plurality of source signal emitters and expressed in an original domain, decomposing the signal into a mathematical representation of a plurality of constituent elements in an alternate domain, analyzing the plurality of constituent elements to associate at least a subset of the constituent elements with at least one of the plurality of source signal emitters, separating at least a subset of the constituent elements based on the association and reconstituting at least a subset of constituent elements to produce an output signal in at least one of the original domain, the alternate domain and another domain.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS[0001]This application claims the benefit of U.S. provisional patent application Ser. No. 61 / 749,606 filed Jan. 7, 2013, U.S. provisional patent application Ser. No. 61 / 785,029 filed Mar. 14, 2013, and U.S. provisional patent application Ser. No. 61 / 642,805 filed May 4, 2012 all of which are hereby incorporated by reference in their entirety.BACKGROUND[0002]1. Field of the Invention[0003]The present invention relates to methods and systems for signal processing and, more specifically, to methods and systems for separating a signal into different components.[0004]2. Description of the Related Art[0005]Signal separation (SS) is a separation of any digital signal originating from a source into its individual constituent elements, such that those elements may be deconstructed, isolated, extracted, enhanced, or reconstituted in isolation, in part, or in whole. SS may be performed on any form of data including auditory data and / or visual data or imag...

AI Technical Summary

Benefits of technology

[0018]In accordance with another exemplary and non-limiting embodiment, a method of processing a signal comprises receiving a time domain signal stream and creating a first window and a second window from the signal stream, wherein an initiation of the second window time lags an initiation of the first window, converting the first window and second window to a frequency domain and storing the resulting frequency domain data having a fundamental transform resolution, performing complex spectral phase evolution (CSPE) on the frequency domain data to estimate component frequencies of the windowed data at a resolution greater than the fundamental transform resolution, using the component frequencies, sampling a set of stored frequency-domain high resolution windows to select frequency-domain high resolution windows that may fit the modulation of a component of the underlying signal wherein the component comprises a plurality of oscillator peaks, using a tracking algorithm to identify at least one tracklet of oscillator peaks that emanate from a single oscillator source within the underlying signal and using information from a behavior of the tracklet with which an oscillator peak is associated to assist in providing an estimate of the modulation of the oscillator.

[0020]In accordance with another exemplary and non-limiting embodiment, a method of processing a signal to produce a mathematical decomposition of the signal in such a way that the decomposed elements of the signal can be recombined to produce a lossless representation of the original signal comprises creating a model of the signal in data windows using a sum of oscillator peaks created using short-time stable oscillators, frequency modulating oscillators, and amplitude modulating oscillators, removing each modeled signal element from the original signal until a desired degree of accuracy is achieved and so that all that remains is a sufficiently small residual signal, encoding the residual signal so that it can be reproduced exactly and storing the parameters of the oscillator peaks used in the modeling of the data along with the encoded residual signal so that they can be recombined into an exact lossless reconstruction of the original signal.

Problems solved by technology

Conventional SS techniques typically require prohibitive amounts of processing to achieve real or near real time performance and are thus far quite often incapable of effectively identifying and isolating signal sources within a given signal.

Images