Acoustic Echo Canceling With Adaptive Filter Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing acoustic echo cancelers face challenges in achieving high echo cancelling performance, particularly in handling non-linearities and varying input signal conditions, which can lead to residual echoes and increased processor load.
Innovation Solution
A system and method that utilize multiple microphones and adaptive filters to approximate unknown transfer functions, along with a microphone calibration block to individually attenuate or amplify electrical microphone signals based on reference signals derived from estimated transfer functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If adaptive filters are used to model echo paths, then echo cancellation performance is improved, but residual echoes remain due to non-linearities and varying signal conditions
Solution Approach 1:
The system segments the echo cancellation task into multiple independent single-channel AEC systems, one for each microphone. Each adaptive filter independently models the transfer function for its specific microphone channel, allowing targeted processing that addresses channel-specific non-linearities and signal conditions without interfering with other channels.
Solution Approach 2:
The system dynamically adjusts the gain of each microphone signal based on real-time analysis of the estimated transfer functions and reference signals. This dynamic adaptation allows the system to respond to varying signal conditions and non-linearities, improving echo cancellation performance across different operating scenarios.
2Adaptability or versatility
If multiple microphones are used to capture sound from multiple directions, then acoustic coverage is improved, but system complexity and processing requirements increase
Solution Approach 1:
The system divides the multi-microphone processing into separate single-channel AEC systems, where each microphone is processed independently. This segmentation reduces overall system complexity by avoiding the need for complex multi-channel interactions while maintaining comprehensive acoustic coverage through parallel processing of each channel.
Solution Approach 2:
The system uses identical adaptive filter and calibration block structures for each microphone channel, creating a universal processing template that can be applied to any number of microphones. This multi-functional approach allows the same hardware and software architecture to handle varying numbers of microphones without requiring redesign, thus managing complexity while maintaining versatility.
3Measurement precision
If adaptive filters continuously model echo paths, then echo estimate accuracy is improved, but processor load and memory requirements increase
Solution Approach 1:
The system segments the processing load across multiple independent single-channel adaptive filters, allowing parallel processing that can be efficiently distributed across available computational resources. This segmentation enables accurate continuous modeling of echo paths for each microphone while managing overall processor load through independent channel processing.
Solution Approach 2:
The system adjusts processing parameters dynamically based on signal conditions, adjusting the complexity and computational intensity of adaptive filtering operations. By changing parameters such as filter update rates and processing depth based on current acoustic conditions, the system maintains high echo estimate accuracy while optimizing processor load and energy consumption.
Data Source
Figure 1~4
Figure 2
Figure 5~6
AI summary
A system and method for canceling acoustic echoes includes picking up with one or more microphones sound generated by a sound source and transferred from the sound source to the one or more microphones via one or more unknown transfer paths having one or more unknown transfer functions, and providing one or more electrical microphone signals therefrom. The system and method further include approximating the one or more unknown transfer functions with one or more estimated transfer functions of the one or more adaptive filters and filtering one or more electrical signals representative of the sound generated by the sound source with the one or more estimated transfer functions to provide one or more estimated signals therefrom. The system and method further include individually, frequency dependently or independently attenuate or amplify the one or more electrical microphone signals dependent on the one or more estimated transfer functions.