Audio Signal Filtering Using Adaptive Gain Constraints

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing audio signal filtering methods, such as those using spectral subtraction, struggle to effectively separate target sound from noise when both contain voice signals, leading to spectral distortion and decreased audio quality.

Innovation Solution

A filtering method that involves receiving two audio signals from the same sound source at different volumes, generating a filter signal by convolving adaptive filter coefficients into one signal, removing components from the other signal, and limiting the gain of the adaptive filter coefficients to 1.0 or less.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If non-linear processing with different gains for different frequency components is performed, then target sound separation is achieved, but spectral distortion occurs and audio quality deteriorates

Engineering Contradiction:

Improvetarget sound separation accuracyVSAvoidspectral accuracy

Core Design Contradiction:

Measurement precisionVSManufacturing precision

Solution Approach 1:

The patent changes the parameter of gain application from frequency-selective (non-linear) to uniform across all frequency components (linear). By applying the same gain value to all frequency components, the method achieves target sound separation while preventing spectral distortion, thus resolving the contradiction between separation accuracy and spectral accuracy.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If non-linear processing is applied to separate target sound, then voice separation is improved, but accuracy of signal processing such as voice recognition deteriorates

Engineering Contradiction:

Improvevoice separation accuracyVSAvoidvoice recognition accuracy

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent applies uniform gain across all frequency components rather than frequency-selective gain. This parameter change maintains the envelope structure of the spectrum, ensuring that voice recognition processing based on envelope information remains accurate while still achieving effective voice separation.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If adaptive filter coefficients with gain greater than 1.0 are used, then filtering effectiveness is improved, but spectral distortion increases

Engineering Contradiction:

Improvefiltering effectivenessVSAvoidspectral fidelity

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The patent constrains the gain parameter of adaptive filter coefficients to be 1.0 or less across all frequency components. This parameter constraint prevents amplification that would cause spectral distortion while maintaining effective noise suppression through the adaptive filtering mechanism.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS12315528B2Filtering method, filtering device, and storage medium stored with filtering program

Publication Date: 2025.05.27 YAMAHA CORP

AI summary

A filtering method includes: receiving a first audio signal and a second audio signal that include sound emitted from a same sound source at different volumes; generating a filter signal by convoluting adaptive filter coefficients into the second audio signal; removing components of the filter signal from the first audio signal; and limiting a gain of the adaptive filter coefficients to 1.0 or less.