Covariance Matrix Smoothing to Reduce Multichannel Audio Artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio signal processing technologies face challenges in smoothing covariance matrices over time, particularly when there are a small number of samples, leading to noticeable artifacts in the output, such as in immersive audio experiences.
Innovation Solution
A method involving the use of a smoothing algorithm that adjusts covariance matrix values based on an effective bin count, incorporates a forgetting factor, and includes transient detection to prevent over-smoothing, ensuring smooth transitions and reducing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If covariance matrix values are computed from a small number of samples, then processing speed is improved, but noticeable artifacts appear in the output
Solution Approach 1:
The patent applies preliminary smoothing action to the covariance matrix values before they are used in audio signal processing. By pre-smoothing the covariance values across frequency bands and time frames, the system prepares the data to avoid artifacts while maintaining processing speed. The smoothing operation is performed in advance using weighted combinations of current and previous covariance estimates.
Solution Approach 2:
The patent introduces an intermediate smoothing operation between covariance computation and audio processing. This intermediary step computes smoothed covariance values by combining current covariance estimates with previously smoothed values, acting as a buffer that prevents direct transmission of noisy artifacts to the output while preserving the essential signal characteristics.
2Object-affected harmful factors
If smoothing is applied to covariance values, then artifacts are reduced, but transitions may become overly smoothed
Solution Approach 1:
The patent implements dynamic smoothing where the smoothing factor is adjusted based on the detected signal characteristics. When transients or rapid changes are detected in the audio signal, the smoothing effect is reduced or temporarily suspended to preserve the natural transitions. This dynamic adaptation allows the system to maintain smooth covariance values during steady-state conditions while preserving sharp transitions during transient events.
Solution Approach 2:
The patent changes the smoothing parameter (smoothing factor) based on the signal conditions. The smoothing factor is modified dynamically to control the degree of smoothing applied to covariance values. This parameter adaptation enables the system to reduce artifacts during normal operation while maintaining faithful representation of transient events when they occur.
Data Source
AI summary
Methods and systems for improving signal processing by smoothing the covariance matrix of a multi-channel signal by setting a forgetting factor based on the bins of a band. A method and system for resetting the smoothing based on transient detection is also disclosed. A method and system for resampling for the smoothing during a banding transition is also disclosed.


