Zero-Crossing Rate Double Talk Detection in Acoustic Echo Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing acoustic echo cancellation techniques face challenges in detecting double talk, leading to filter divergence and echo leakage, especially on low-resource processors, due to high computational costs and inaccuracies in estimating the start and end of near-end speech.
Innovation Solution
The method employs zero-crossing rate (ZCR) to detect double talk by analyzing the number of zero crossings in audio signals, allowing for efficient double-talk detection and adaptive filter coefficient management to prevent echo cancellation divergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If classical DT detection methods using correlation measures are used, then DT detection accuracy is improved, but computational complexity increases significantly
Solution Approach 1:
The patent changes the detection parameter from correlation measures to zero-crossing rate (ZCR). ZCR is a simpler parameter that captures speech presence without requiring complex correlation computations, thus reducing computational complexity while maintaining detection accuracy.
Solution Approach 2:
The patent substitutes the complex mechanical computation system (correlation measures requiring large data amounts and multiple audio samples) with a simpler analytical approach (zero-crossing rate calculation based on signal sign changes), making the system more suitable for low-resource processors.
2Reliability
If DT detection measures are computed for each audio sample to ensure accurate detection, then detection reliability is improved, but processing time increases
Solution Approach 1:
By changing to zero-crossing rate as the detection parameter, the system achieves reliable DT detection without requiring computation for every single audio sample. ZCR can be computed efficiently over short windows, reducing processing time while maintaining detection reliability.
3Stability of the object's composition
If filter adaptation is frozen during DT to prevent divergence, then filter stability is improved, but echo cancellation performance deteriorates when echo path changes
Solution Approach 1:
The patent uses zero-crossing rate as a feedback mechanism to detect DT conditions and control filter adaptation accordingly. When ZCR indicates DT presence, the system adjusts adaptation behavior to prevent divergence while maintaining the ability to track echo path changes, balancing stability and performance.
Data Source
AI summary
A method for acoustic echo cancellation in a communication device is provided that includes receiving a first near-end audio signal in the communication device, wherein the first near-end audio signal comprises acoustic echo of a far-end audio signal reproduced by the communication device, and performing echo cancellation on the first near-end audio signal to generate a second near-end audio signal with at least some of the acoustic echo removed, wherein the echo cancellation is performed responsive to presence or absence of double-talk (DT), and wherein a zero-crossing rate (ZCR) is used to detect the presence or absence of DT.