Echo Suppression Filter Coefficient Computation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current echo suppression systems in conference systems fail to optimally handle different sound, tone, and noise components, leading to non-optimum echo suppression and the occurrence of tonal artifacts due to their inability to distinguish between stationary and non-stationary signal components in loudspeaker signals.

Innovation Solution

The system analyzes loudspeaker signals to separate stationary and non-stationary components, computing filter coefficients for an adaptive filter based on these components to effectively suppress echoes in microphone signals, using techniques such as floating averaging and gain filters adjusted by coherence functions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single adaptive filter is used for echo suppression, then the system structure remains simple, but the echo suppression performance deteriorates due to inability to handle different signal components optimally

Engineering Contradiction:
Improveecho suppression performanceVSAvoidfilter processing complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The loudspeaker signal is segmented into stationary and non-stationary components using a coherence function-based separation. Different adaptive filters are then applied to each component type, allowing optimal handling of each signal characteristic while maintaining manageable system complexity through modular processing.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system dynamically adapts the filter processing based on the signal characteristics. By continuously analyzing the coherence function and signal statistics, the system adjusts which filter type (stationary or non-stationary) is applied to each signal component, optimizing performance for varying acoustic conditions.

Inventive Principle:
Principle #15Dynamics

2Object-affected harmful factors

If uniform echo suppression is applied to all signal components, then the processing remains simple, but tonal artifacts arise due to non-optimal suppression of different sound, tone, and noise components

Engineering Contradiction:
Improvetonal artifactsVSAvoidsignal analysis complexity
Core Design Contradiction:
Object-affected harmful factorsVSDevice complexity

Solution Approach 1:

Different filtering approaches are applied to different signal components based on their local characteristics. Stationary components receive one type of adaptive filtering while non-stationary components receive another, ensuring each signal type is processed with the most appropriate method to minimize artifacts.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system changes processing parameters based on signal characteristics by using coherence function analysis to identify stationary versus non-stationary components. This parameter-based differentiation allows the system to adapt filter settings and processing methods to match the actual signal properties, reducing tonal artifacts.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2238590B1Apparatus and method for computing filter coefficients for echo suppression
Publication Date: 2016.06.01 FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV
  • EP2238590B1 patent drawingFigure 1
  • EP2238590B1 patent drawingFigure 2
  • EP2238590B1 patent drawingFigure 3A~3B

AI summary

An embodiment of an apparatus (200) for computing filter coefficients for an adaptive filter (210) for filtering a microphone signal so as to suppress an echo due to a loudspeaker signal includes extraction means (250) for extracting a stationary component signal or a non-stationary component signal from the loudspeaker signal or from a signal derived from the loudspeaker signal, and computing means (270) for computing the filter coefficients for the adaptive filter (210) on the basis of the extracted stationary component signal or the extracted non-stationary component signal.