Adaptive Filter Double-Talk Detection Using LMS Coefficient Peaks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional double-talk detection techniques in electronic devices suffer from latency and high computation costs due to the need for long analysis windows to account for variable echo latency in wireless communication systems, especially when using wireless loudspeakers with unpredictable delays.
Innovation Solution
The implementation of an adaptive filter with least mean squares (LMS) adaptive filter coefficients that updates based on speech detection, allowing for the identification of unique noise sources and differentiation between single-talk and double-talk conditions by analyzing the number and location of peaks in the filter coefficient values, thereby improving location estimation and reducing processing requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If long analysis windows are used to account for variable echo latency in wireless communication systems, then measurement precision of system conditions is improved, but loss of time (latency) and computation costs increase
Solution Approach 1:
The system performs preliminary double-talk detection using a short analysis window before applying aggressive audio processing. This preliminary detection allows the system to prepare and switch processing modes in advance, reducing the effective latency by having detection results ready before they are critically needed for processing decisions.
Solution Approach 2:
The system dynamically adjusts the analysis window length based on system conditions. During far-end single-talk conditions, a longer analysis window is used for accurate detection. During near-end single-talk or double-talk conditions, the system switches to shorter windows or alternative detection methods, adapting the detection strategy to current acoustic conditions to minimize latency when it matters most.
2Measurement precision
If long analysis windows are used to account for variable echo latency, then measurement precision of system conditions is improved, but device complexity and computation costs increase
Solution Approach 1:
The detection process is segmented into multiple stages: preliminary detection using short windows, conditional detection based on system state, and selective application of aggressive processing. This segmentation allows the system to use computationally intensive long-window analysis only when necessary (during far-end single-talk), while using simpler methods during other conditions, thereby reducing overall computational burden and device complexity requirements.
Solution Approach 2:
The system changes detection parameters (analysis window length, detection threshold, processing aggressiveness) based on the detected system conditions. By dynamically adjusting these parameters, the system achieves high detection accuracy when needed while avoiding unnecessary computational complexity during conditions where simpler detection suffices, thus resolving the contradiction between precision and complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach reduces latency and computation costs by accurately determining system conditions without the need for extensive analysis windows, enhancing the ability to isolate local speech and suppress unwanted noise in real-time communication scenarios.
Implementation Method 1
The implementation of an adaptive filter with least mean squares (LMS) adaptive filter coefficients that updates based on speech detection, allowing for the identification of unique noise sources and differentiation between single-talk and double-talk conditions by analyzing the number and location of peaks in the filter coefficient values
Data Source
AI summary
A system configured to improve double-talk detection. The system inputs microphone signals into an adaptive filter and determines whether double talk is present based on how the adaptive filter adapts to the microphone signals. For example, when an audible sound is detected, the adaptive filter updates the filter coefficients that correspond to a time difference of arrival of the audible sound. Thus, the device may detect single-talk conditions (e.g., a single peak in the filter coefficients) or double-talk conditions (e.g., two peaks in the filter coefficients). In addition, the device may track a location of the local speech or remote speech over time based on the time difference of arrival.