Spectral Refinement System for Audio Signal Processing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing audio signal processing methods, such as spectral subtraction, often distort speech when attempting to reduce background noise, and increasing discrete resolution to reduce sub-band overlap results in undesirable time delays in communication systems.

Innovation Solution

A process that refines the short-term spectrum of audio signals by transforming sub-bands into the frequency domain, applying a window function, and selectively filtering time-delayed sub-band spectra to minimize overlap, using a refinement matrix and finite impulse response filters to enhance noise reduction, echo compensation, and voice pitch estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Object-affected harmful factors

If spectral subtraction is used to reduce background noise, then noise damping is improved, but speech distortion occurs when spectral noise components are damped and not eliminated

Engineering Contradiction:

Improvebackground noiseVSAvoidspeech distortion

Core Design Contradiction:

Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The patent applies different processing strategies to different spectral components. Stationary noise components are damped using spectral subtraction, while transient speech components are preserved through adaptive thresholding and time-domain filtering. This local differentiation allows noise reduction in stationary regions without distorting transient speech elements.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system dynamically adapts its noise reduction strength based on the detected speech activity and noise characteristics. The spectral subtraction factor is adjusted in real-time based on the signal-to-noise ratio and speech presence detection, allowing aggressive noise damping when speech is absent and gentle processing when speech is present, thereby preventing distortion.

Inventive Principle:

Principle #15Dynamics

2Manufacturing precision

If discrete resolution is increased to reduce sub-band overlap, then sub-band overlap is reduced, but time resolution decreases causing undesirable time delays

Engineering Contradiction:

Improvesub-band separationVSAvoidtime delay

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The audio signal is divided into multiple sub-bands using filter banks, with each sub-band processed independently. This segmentation allows precise control over the resolution in each frequency band, achieving good sub-band separation without requiring excessive overall resolution that would increase time delays.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent transforms the problem from a time-domain resolution issue to a frequency-domain processing solution. By applying spectral subtraction and filtering in the frequency domain, the system achieves precise spectral control without being constrained by time-domain resolution limitations, thus avoiding excessive time delays.

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Data Source

PatentUS8190426B2Spectral refinement system

Publication Date: 2012.05.29 CERENCE OPERATING CO

AI summary

An audio enhancement refines a short-time spectrum. The refinement may reduce overlap between audio sub-bands. The sub-bands are transformed into sub-band short-time spectra. A portion of the spectra are time-delayed. The sub-band short-time spectrum and the time-delayed portion are filtered to obtain a refined sub-band short-time spectrum. The refined spectrum improves audio processing.