IP Library Granted Patent US 11,581,005
Granted Patent B2
US 11,581,005 · App. 17/587,598 · Granted Feb 14, 2023

Methods and systems for improved signal decomposition

Inventors: Elias Kokkinis (Patras, GR); Alexandros Tsilfidis (Athens, GR)
Assignee: Meta Platforms Technologies, LLC
G10L21/0272G10L19/008G10L21/02G10L21/0208
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 11,581,005
App. No.
17/587,598
Granted
Feb 14, 2023
Kind
B2
Abstract

A method for improving decomposition of digital signals using training sequences is presented. A method for improving decomposition of digital signals using initialization is also provided. A method for sorting digital signals using frames based upon energy content in the frame is further presented. A method for utilizing user input for combining parts of a decomposed signal is also presented.

Claims (26)

1. An apparatus capable of creating an initial set of values in a row of a weight matrix in non-negative matrix factorization to decompose digital signals, the apparatus capable of:

generating the initial set of values of the row of the weight matrix from a ratio of a first function of a first signal of a plurality of digital audio signals divided by a second function of at least two other signals of the plurality of the digital audio signals, wherein the row in the weight matrix determines a decomposition of the plurality of digital audio signals into signal components; and

audibly outputting a portion of one or more of the components of the decomposed plurality of digital audio signals at least based on input from an interface element.

2. The apparatus of claim 1 , wherein the digital audio signals are one or more of binaural or multichannel audio signals.

3. The apparatus of claim 1 , wherein the first and second functions are calculated from one or more filtered versions of said digital audio signals.

4. The apparatus of claim 1 , wherein the first and second functions are calculated in one or more of the time domain, the frequency domain, theme-frequency domain.

5. The apparatus of claim 1 , wherein the first and second functions are calculated using one or more of energy, power, root mean square, geometric mean, arithmetic mean, euclidean norm, taxicab norm, or Lp norm.

6. The apparatus of claim 1 , wherein the digital audio signals are one or more of binaural or multichannel audio signals and the output portion of the one or more of the components are used for one or more of: source separation, signal restoration, signal enhancement, noise removal, un-mixing, up-mixing and re-mixing.

7. The apparatus of claim 1 , wherein the output portion of the one or more of the components are used for one or more of: source separation, signal restoration, signal enhancement, noise removal, un-mixing, up-mixing and re-mixing.

8. The apparatus of claim 1 , wherein the plurality of digital signals are a single source coming from an original mixture of multiple sources.

9. The apparatus of claim 1 , further comprising automatically sorting one or more of the components.

10. A non-transitory computer readable information storage media having stored

therein instructions, that when executed by one or more processors, cause to be performed a method of creating an initial set of values in a row of a weight matrix in non-negative matrix factorization to decompose digital signals, the method comprising:

generating the initial set of values of the row of the weight matrix from a ratio of a first function of a first signal of a plurality of digital audio signals divided by a second function of at least two other signals of the plurality of the digital audio signals, wherein the row in the weight matrix determines a decomposition of the plurality of digital audio signals into signal components; and

audibly outputting a portion of one or more of the components of the decomposed plurality of digital audio signals at least based on input from an interface element.

11. The media of claim 10 , wherein the digital audio signals are one or more of binaural or multichannel audio signals.

12. The media of claim 10 , wherein the first and second functions are calculated from one or more filtered versions of said digital audio signals.

13. The media of claim 10 , wherein the first and second functions are calculated in one or more of the time domain, the frequency domain, theme-frequency domain.

14. The media of claim 10 , wherein the first and second functions are calculated using one or more of energy, power, root mean square, geometric mean, arithmetic mean, euclidean norm, taxicab norm, or Lp norm.

15. The media of claim 10 , wherein the digital audio signals are one or more of binaural or multichannel audio signals and the output portion of the one or more of the components are used for one or more of: source separation, signal restoration, signal enhancement, noise removal, un-mixing, up-mixing and re-mixing.

16. The media of claim 10 , wherein the output portion of the one or more of the components are used for one or more of: source separation, signal restoration, signal enhancement, noise removal, un-mixing, up-mixing and re-mixing.

17. The media of claim 10 , wherein the plurality of digital signals are a single source coming from an original mixture of multiple sources.

18. The media of claim 10 , further comprising automatically sorting one or more of the components.

19. An apparatus capable of creating an initial set of values in a row of a weight matrix in non-negative matrix factorization to decompose digital signals, the apparatus comprising:

means for generating the initial set of values of the row of the weight matrix from a ratio of a first function of a first signal of a plurality of digital audio signals divided by a second function of at least two other signals of the plurality of the digital audio signals, wherein the row in the weight matrix determines a decomposition of the plurality of digital audio signals into signal components; and

means for audibly outputting a portion of one or more of the components of the decomposed plurality of digital audio signals at least based on input from an interface element.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Sep 19, 2022
From: ACCUSONUS, INC.
To: META PLATFORMS TECHNOLOGIES, LLC
Reel/Frame 061140/0027 →
CHANGE OF NAME Recorded Jun 8, 2022
From: FACEBOOK TECHNOLOGIES, LLC
To: META PLATFORMS TECHNOLOGIES, LLC
Reel/Frame 060314/0965 →
Continuity (4)
Continuation 16521844 · Jul 25, 2019
Continuation 15804675 · Nov 6, 2017
Continuation 14011981 · Aug 28, 2013
Related Publication 20220148612A1 · May 12, 2022