Adaptive Step-Size Reverberation Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing reverberation suppression technologies face challenges in accuracy and computational efficiency, particularly when using fixed step sizes or high nonlinearity in multi-channel semi-blind independent component analysis, leading to increased calculation costs and reduced performance.
Innovation Solution
A reverberation suppressing apparatus and method that calculates an optimal separation matrix using a first evaluation function, updates the separation matrix by segmenting the step-size function into linear segments, and minimizes a second evaluation function to improve the degree of separation, while whitening input signals to enhance accuracy and reduce computational processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a separation matrix is updated with a fixed step size at every frame, then the updating process is simple, but the reverberation suppression accuracy deteriorates when the step size is not adequate and the process takes a long time when the step size is small
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed step size to an adaptive step size that changes over time. The step size is dynamically adjusted based on the convergence state of the separation matrix, allowing the system to achieve both fast initial convergence and high final accuracy. This resolves the contradiction by making the updating process neither too simple nor too complex, but optimally adapted at each stage.
Solution Approach 2:
The patent changes the parameter (step size) from a constant value to a time-varying value. By modifying the step size parameter dynamically during the updating process, the system can achieve rapid initial convergence with larger step sizes and then refine accuracy with smaller step sizes, thereby resolving the contradiction between simplicity and accuracy.
2Measurement precision
If the order of separation filter is increased to handle high nonlinearity in multi-channel semi-blind independent component analysis, then the accuracy of sound source separation is improved, but the calculation time increases and calculation cost increases
Solution Approach 1:
The patent applies segmentation by dividing the updating process into multiple frames, where the separation matrix is updated incrementally. This allows the system to achieve high separation accuracy through cumulative refinement rather than requiring a high-order filter that would demand excessive computational resources at once, thus resolving the contradiction between accuracy and calculation efficiency.
Solution Approach 2:
The patent uses dynamic adaptation of the separation matrix through iterative updating across multiple frames. Instead of relying on a static high-order filter structure that increases computational complexity, the system dynamically refines the separation matrix using adaptive algorithms, achieving high accuracy with manageable computational cost at each step.
Data Source
AI summary
A reverberation suppressing apparatus separating sound source signals based on input signals output from microphones collecting the plurality of sound source signals, includes a sound signal output unit generating sound signals and outputting the generated sound signals, a sound acquiring unit acquiring the input signals from microphones, a first evaluation function calculation unit calculating a separation matrix, the input signals, and the sound source signals, and calculating a first evaluation function, a reverberation component suppressing unit calculating an optimal separation matrix, and suppressing a reverberation component by separating the sound source signals other than the generated sound signals, and a separation matrix updating unit dividing a step-size function, approximating each segment to a linear function, calculating step sizes based on the approximated linear functions, and repeatedly updating the separation matrix so that the degree of separation of the sound source signals exceeds the predetermined value.