Time-Variant Noise Spatial Covariance Matrix Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating noise spatial covariance matrices use a common matrix for all time blocks, which is inadequate for environments with varying noise levels, leading to reduced precision due to the shortening of the acoustic signal time interval.
Innovation Solution
The technique involves dividing acoustic signals into time-frequency signals and using mask information to calculate a time-independent noise spatial covariance matrix for a long time interval, then applying mixture weights for short time intervals to generate a time-variant noise spatial covariance matrix, effectively addressing temporal variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a noise spatial covariance matrix is estimated for each time block using only the acoustic signal of that time block, then the noise spatial covariance matrix can be updated frequently to track time-variant noise, but the time interval of the acoustic signal used for estimation becomes short, leading to reduced precision
Solution Approach 1:
The patent segments the noise spatial covariance matrix estimation into two distinct components: a long-time-interval matrix (first noise spatial covariance matrix) that ensures precision, and short-time-interval matrices (second noise spatial covariance matrices) that enable frequent updates. These segmented components are then combined through weighted summation to achieve both high precision and frequent updates.
Solution Approach 2:
The patent merges the long-time-interval noise spatial covariance matrix with multiple short-time-interval noise spatial covariance matrices through weighted summation. The first noise spatial covariance matrix serves as a baseline with high precision, while the second noise spatial covariance matrices provide time-variant adaptations. This merging allows the system to maintain precision while tracking rapid noise changes.
2Measurement precision
If a common noise spatial covariance matrix is used for all time blocks, then the estimation precision can be maintained by using a long time interval, but the system cannot respond to time-variant noise characteristics in different time blocks
Solution Approach 1:
The patent introduces dynamics into the noise spatial covariance matrix system by creating time-variant second noise spatial covariance matrices for different time blocks. These dynamic matrices are generated from short-time-interval acoustic signals and combined with the static first noise spatial covariance matrix through weighted summation, enabling the system to adapt to changing noise characteristics while maintaining overall precision.
Solution Approach 2:
The patent applies local quality by allowing different noise spatial covariance matrices to have different properties for different time blocks. The first noise spatial covariance matrix provides a stable baseline, while the second noise spatial covariance matrices provide localized adaptations for specific time blocks. This local differentiation enables the system to respond to time-variant noise while maintaining global precision.
3Adaptability or versatility
If the time interval for noise spatial covariance matrix estimation is shortened to track rapid noise changes, then the system can respond to time-variant noise, but the precision of the estimation decreases
Solution Approach 1:
The patent implements dynamics by creating a hierarchical structure where a static, high-precision long-time-interval matrix coexists with dynamic, responsive short-time-interval matrices. The dynamic matrices enable responsiveness to rapid noise changes, while the static matrix ensures precision. Their weighted combination allows the system to achieve both responsiveness and precision simultaneously.
Solution Approach 2:
The patent creates a composite noise spatial covariance matrix by combining two distinct types of matrices: the first noise spatial covariance matrix (long-time-interval, high-precision) and the second noise spatial covariance matrices (short-time-interval, responsive). This composite structure allows the system to inherit the precision properties of the long-time matrix while gaining the responsiveness of the short-time matrices.
Data Source
AI summary
A time-variant noise spatial covariance matrix is estimated effectively. Using time-frequency-divided observation signals based on observation signals acquired by collecting acoustic signals emitted from one or a plurality of sound sources and mask information expressing the occupancy probability of a component of each of the time-frequency-divided observation signals that corresponds to each noise source, a time-independent first noise spatial covariance matrix corresponding to the time-frequency-divided observation signals and the mask information belonging to a long time interval is acquired for each noise source. Further, using the mask information of each of a plurality of different short time intervals, a mixture weight corresponding to each noise source in each short time interval is acquired. Furthermore, a time-variant third noise spatial covariance matrix is acquired, the third noise spatial covariance matrix being based on a time-variant second noise spatial covariance matrix, which corresponds to the time-frequency-divided observation signals and the mask information belonging to each short time interval and relates to noise formed by adding together all of the noise sources, and a weighted sum of the first noise spatial covariance matrices with the mixture weights of the respective short time intervals.