Predictive Acoustic Echo Cancellation With Segmented Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Acoustic echo cancellation in communication devices with multiple microphones is complex and power-intensive, leading to performance constraints, especially with long room impulse responses and increased microphone configurations.
Innovation Solution
A predictive acoustic echo canceller (AEC-P) method that sections the room impulse response into three parts, using full complexity for the initial section, reduced complexity for the middle section, and shared tail section modeling across microphones, reducing overall computational complexity and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple microphones are used to improve voice interaction experience, then voice interface performance is improved, but echo cancellation complexity increases
Solution Approach 1:
The patent divides the echo cancellation system into multiple independent AEC instances, one for each microphone. Each AEC instance processes the reference signal separately to generate microphone-specific echo estimates, allowing parallel processing that maintains performance while managing complexity through modular architecture
Solution Approach 2:
The patent implements a universal AEC architecture that can handle any number of microphones using the same processing framework. The system processes multiple microphone signals through identical AEC pipelines, enabling the system to scale from one to many microphones without requiring fundamentally different approaches for each configuration
2Reliability
If full complexity AEC is applied to all microphones to maintain echo cancellation performance, then echo cancellation quality is preserved, but power consumption increases
Solution Approach 1:
The patent applies different processing complexities to different microphones based on their specific acoustic characteristics and requirements. The system can identify which microphones need full-complexity AEC processing and which can use reduced-complexity processing, optimizing the balance between echo cancellation quality and power consumption on a per-microphone basis
Solution Approach 2:
The patent implements variable complexity AEC processing where not all microphones receive the same level of processing. Some microphones may use full-complexity AEC when needed, while others use reduced-complexity processing, applying exactly the right amount of processing effort to each microphone rather than uniformly maximum processing to all
3Measurement precision
If AEC processing is applied to each microphone independently to handle different echo characteristics, then echo cancellation accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the echo cancellation processing into independent parallel AEC instances for each microphone, where each instance maintains its own adaptive filters and processing pipeline. This segmentation allows each microphone to be processed independently with appropriate complexity while maintaining overall system manageability through modular architecture
Solution Approach 2:
The patent implements dynamic complexity management where the system can adjust the processing level for each microphone based on real-time requirements. The architecture allows flexible allocation of computational resources, enabling full-complexity processing when echo conditions demand it and reduced-complexity processing when conditions permit, optimizing the accuracy-complexity tradeoff dynamically
Data Source
AI summary
A method for echo cancellation based on microphone signal correlation is disclosed. The method includes replaying a reference signal by a speaker; collecting a primary audio signal by a primary microphone and a secondary audio signal by a secondary microphone based on the reference signal being replayed; partitioning the reference signal into a plurality of sectioned reference signals, based on levels of correlation between the primary audio signal and the secondary audio signal; generating a plurality of sectioned primary echo signals for the primary microphone by processing the sectioned reference signals by an acoustic audio canceller (AEC); generating a plurality of sectioned secondary echo signals for the secondary microphone by processing the sectioned reference signals by a predictive acoustic audio canceller (AEC-P), wherein the AEC-P generates at least one of the sectioned secondary echo signals based on a corresponding sectioned primary echo signal; and performing echo cancellation for the secondary microphone by removing a combination of the sectioned secondary echo signals from the secondary audio signal.