Method, apparatus and system

Patent metadata

Field	Value
Publication number	US-10657973-B2
Application number	US-201515516368-A
Country	US
Kind code	B2
Filing date	Sep 29, 2015
Priority date	Oct 2, 2014
Publication date	May 19, 2020
Grant date	May 19, 2020

Abstract

A method including decomposing a magnitude part of a signal spectrum of a mixture signal into spectral components, each spectral component including a frequency part and a time activation part; and clustering the spectral components to obtain one or more clusters of spectral components, wherein the clustering of the spectral components is computed in the time domain.

First claim

The invention claimed is:

1. A method, comprising:
decomposing a magnitude part of a signal spectrum of a mixture signal into a plurality of spectral components, each spectral component of the plurality of spectral components comprising a frequency component and a time component;
applying an inverse short-time Fourier transform (ISTFT) to the plurality of spectral components to generate a plurality of time components; and
clustering the plurality of time components, in a time domain and based on an iterative algorithm, to obtain one or more clusters of time components, the iterative algorithm minimizing an energy of a compression error of estimated source signals.

2. The method of claim 1 , wherein the clustering comprises generating the one or more clusters of time components based on the spectral components.

3. The method of claim 1 , wherein the clustering the time components is based on a compressibility of estimated source signals.

4. The method of claim 1 , wherein the clustering the time components comprises minimizing the compression error of the estimated source signals.

5. The method of claim 1 , wherein the clustering the time components is based on minimizing the energy of an overall compression error according to


e


2

=


x


2

-

2
⁢

∑

i
=
1

M

⁢

x
T

⁢

S
^

i

⁡

(

S
^

i
T

⁢

S
^

i

)

-
1

⁢

S
^

i
T

⁢

s
^

i

+

∑

i
=
1

M

⁢

∑

j
=
1

M

⁢

s
^

i
T

⁢

S
^

i

⁡

(

S
^

i
T

⁢

S
^

i

)

-
1

⁢

S
^

i
T

⁢

S
^

j

⁡

(

S
^

j
T

⁢

S
^

j

)

-
1

⁢

S
^

j
T

⁢

s
^

j

.

Assignees

• Sony Corp

Inventors

• Guo Xin
• Uhlich Stefan
• Mitsufuji Yuhki

Classifications

• G10H2210/056

  for extraction or identification of individual instrumental parts, e.g. melody, chords, bass; Identification or separation of instrumental parts by their characteristic voices or timbres · CPC title

• G10L21/0308

  characterised by the type of parameter measurement, e.g. correlation techniques, zero crossing techniques or predictive techniques · CPC title

• G10L21/0272

  Voice signal separating · CPC title

• G10L19/008 Primary

  Multichannel audio signal coding or decoding using interchannel correlation to reduce redundancy, e.g. joint-stereo, intensity-coding or matrixing · CPC title

• G10H2250/235

  Fourier transform; Discrete Fourier Transform [DFT]; Fast Fourier Transform [FFT] · CPC title