Multi-Stage Acoustic Echo Cancellation for Long FIR Paths
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional acoustic echo cancellation systems using fixed-point processors face performance degradation due to limited bit precision and saturation issues when handling long acoustic echo paths, especially in applications like hand-free voice terminals, where the echo path span can be several hundred milliseconds, requiring thousands of taps for accurate representation.
Innovation Solution
The proposed system employs a two-stage echo cancellation approach with a first attenuator and filter to remove the primary echo components, followed by a second filter to address residual echoes, utilizing adaptive FIR filters and down-scaling factors to maintain precision and avoid saturation, and a third filter to further refine the echo cancellation, ensuring effective echo removal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a long FIR filter is used to represent the acoustic echo path impulse response, then echo cancellation coverage is improved, but the number of multiply-accumulate operations exceeds the accumulator length limit, causing saturation and loss of precision
Solution Approach 1:
The patent divides the single long FIR filter into multiple shorter FIR filters, each processing a portion of the echo path. This segmentation allows the system to represent long echo paths without exceeding the accumulator length limit in any single filter, thereby maintaining precision while reducing operational complexity.
Solution Approach 2:
The patent introduces an additional dimension by using multiple parallel filter paths instead of a single long filter. Each filter operates with reduced complexity while the combined output achieves the desired long-echo cancellation, effectively transforming a one-dimensional problem into a multi-dimensional solution.
2Measurement precision
If the accumulator length is increased to handle long FIR filters, then precision is improved, but the processor cost and complexity increase
Solution Approach 1:
By segmenting the filter into multiple shorter filters, the patent reduces the accumulator length requirement for each individual filter. This allows the use of standard, cost-effective fixed-point processors without requiring expensive high-precision accumulators, thereby maintaining affordability while achieving necessary precision.
Solution Approach 2:
The patent changes the parameter of filter length from a single long filter to multiple shorter filters. This parameter transformation reduces the accumulator length requirement and allows the system to operate with standard processor specifications, avoiding the need for costly high-precision hardware.
3Ease of manufacture
If fixed-point processors are used to reduce cost, then device cost is reduced, but bit precision is limited causing saturation in multiply-accumulate operations
Solution Approach 1:
The patent segments the signal processing into multiple parallel filter paths, each operating with reduced complexity. This segmentation allows fixed-point processors to handle each filter section without saturation, as the reduced filter length keeps the number of multiply-accumulate operations within the accumulator's capacity, thereby maintaining precision while using cost-effective hardware.
Solution Approach 2:
The patent applies partial action by using multiple shorter filters instead of one long filter. Each filter processes a partial portion of the echo path, and the combined output achieves the full cancellation effect. This partial approach prevents saturation in each individual filter while maintaining overall precision, avoiding the need for excessive precision hardware.
4Device complexity
If a single long FIR filter is used, then echo cancellation is simplified, but the system cannot represent the acoustic impulse response with sufficient precision due to saturation
Solution Approach 1:
The patent segments the single long FIR filter into multiple shorter filters arranged in parallel. Each filter represents a portion of the acoustic impulse response with sufficient precision, and the combined output provides complete echo cancellation. This segmentation prevents saturation in any single filter while maintaining the simplicity of the overall filter structure.
Solution Approach 2:
The patent transitions from a one-dimensional single-filter approach to a multi-dimensional parallel-filter approach. This dimensional change allows the system to represent long impulse responses with sufficient precision by distributing the representation across multiple filters, each operating within precision limits while collectively achieving the desired accuracy.
Data Source
AI summary
An embodiment of an acoustic echo cancellation system is disclosed. The system comprises an echo cancellation unit, a second filter and a subtraction unit. The echo cancellation unit comprises a first attenuator, a first filter and a first subtractor. The first attenuator has a first down-scaling factor for attenuating a first signal. The first filter generates a first echo signal estimate based on the attenuated first signal. The first subtractor generates a third signal by subtracting the first echo signal estimate from a second signal. The second filter generates a second echo signal estimate based on the first signal. The subtraction unit subtracts the second echo signal estimate from the third signal.