Sound Source Separation Device Using Log-Likelihood Demixing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing sound source separation technologies face challenges in accurately isolating voice signals from multiple sound sources amidst noise interference, particularly in complex acoustic environments.

Innovation Solution

A sound source separation device that utilizes a matrix unit to generate an objective function based on estimated source and noise vectors, employing independent component analysis and log-likelihood functions to estimate a demixing matrix, thereby minimizing mutual information and enhancing signal separation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional sound source separation methods are used, then the separation process can be implemented, but the accuracy of voice signal separation is insufficient due to noise interference

Engineering Contradiction:
Improvevoice signal separation accuracyVSAvoidnoise interference
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent extracts and removes noise components from the mixed sound signal by estimating noise vectors and applying suppression techniques. The noise estimation unit specifically extracts noise characteristics from the input signal, and the noise suppression unit removes these extracted noise components, thereby improving voice signal separation accuracy in noisy environments.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces a demixing matrix as an intermediary computational tool that transforms the mixed signal into separated source signals. This demixing matrix acts as a mediator between the noisy input signal and the cleaned output signals, enabling effective separation by optimizing the transformation based on statistical independence assumptions.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If simple separation methods are used, then the computational process is fast, but the separation accuracy is insufficient in complex acoustic environments

Engineering Contradiction:
Improveseparation accuracyVSAvoidalgorithm complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the sound source separation task into distinct functional modules: noise estimation unit, noise suppression unit, and demixing matrix computation unit. Each module handles a specific aspect of the separation process, making the complex algorithm more manageable and implementable while maintaining high separation accuracy through coordinated operation of these segmented components.

Inventive Principle:
Principle #1Segmentation

3Reliability

If noise suppression is applied, then voice signal quality improves, but the processing time and computational load increase

Engineering Contradiction:
Improvevoice recognition performanceVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent performs noise estimation and demixing matrix computation as preliminary actions before the actual noise suppression is applied. By pre-estimating noise characteristics and pre-computing the optimal demixing matrix, the system reduces the computational burden during real-time processing, thereby improving voice recognition performance without excessive processing delays.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4456066B1Sound source seperation device
Publication Date: 2025.06.18 MPWAV INC
  • EP4456066B1 patent drawingFigure 1~3
  • EP4456066B1 patent drawingFigure 4~5
  • EP4456066B1 patent drawing

AI summary

A sound source separation device according to an embodiment of the present invention may include a plurality of microphones, a matrix unit, and an output unit. The plurality of microphones may receive a plurality of microphone input signals transmitted from a plurality of sound sources. The matrix unit may generate an objective function according to an estimated source vector and an estimated noise vector estimated based on the plurality of microphone input signals, and replace a first term and a second term included in the objective function using a log-likelihood function to estimate a demixing matrix. The output unit may provide output vectors calculated based on the microphone input signals and the demixing matrix. According to a sound source separation device of the present invention, it is possible to more accurately separate voice signals transmitted from each of the plurality of sound sources by generating an objective function according to an estimated source vector and an estimated noise vector estimated based on a plurality of microphone input signals and replacing a first term and a second term included in the objective function using a log-likelihood function to estimate a demixing matrix.