Noise Estimation Parameter Learning for Target Sound Enhancement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large spaces like sports fields and manufacturing factories, existing noise suppression techniques fail to effectively enhance target sounds due to reverberation and time frame differences between microphones, making it difficult to execute spectral subtraction methods.
Innovation Solution
A noise estimation parameter learning device that models and estimates time frame differences and transfer function gains using a statistical model, allowing multiple microphones to cooperate and apply spectral subtraction for target sound enhancement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple microphones are disposed at distant positions in a large space, then the coverage area increases, but the time frame difference between microphones worsens
Solution Approach 1:
The system performs preliminary estimation of time frame differences between multiple microphones based on their positional information before actual noise estimation. This preliminary action allows the system to compensate for time delays in advance, enabling effective cooperation of microphones disposed at distant positions without suffering from time frame difference issues during noise suppression.
2Area of stationary object
If multiple microphones are disposed at distant positions in a large space, then the coverage area increases, but the reverberation effect worsens
Solution Approach 1:
The system performs preliminary estimation of reverberation characteristics in the large space before noise estimation. By understanding the reverberation properties in advance, the system can design appropriate filtering and noise suppression strategies that account for the specific acoustic environment, thereby reducing the harmful effects of reverberation on target sound enhancement.
3Speed
If spectral subtraction method is applied without considering time frame difference, then the noise suppression speed increases, but the noise estimation accuracy worsens
Solution Approach 1:
The system performs preliminary estimation of time frame differences based on microphone positions before applying spectral subtraction. This preliminary step enables the system to align signals from multiple microphones in the time domain, ensuring that noise estimation is performed on properly synchronized signals, thereby maintaining both speed and accuracy.
Solution Approach 2:
The system introduces time frame difference estimation as an intermediary step between signal acquisition and spectral subtraction. This intermediary process transforms the raw signals into time-aligned signals that are suitable for accurate noise estimation, acting as a bridge that preserves both processing efficiency and estimation accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A noise estimation parameter learning device is provided according to which even in a large space causing a problem of the reverberation and the time frame difference, multiple microphones disposed at distant positions cooperate with each other, and a spectral subtraction method is executed, thereby allowing the target sound to be enhanced. A noise estimation parameter learning device for learning noise estimation parameters used to estimate noise included in observed signals through a plurality of microphones, the noise estimation parameter learning device comprising: a modeling part that models a probability distribution of observed signals of the predetermined microphone, models a probability distribution of time frame differences, and models a probability distribution of transfer function gains; a likelihood function setting part that sets a likelihood function pertaining to the time frame difference, and a likelihood function pertaining to the transfer function gain, based on the modeled probability distributions; and a parameter update part that alternately and repetitively updates two variables of two likelihood functions, and outputs the time frame difference and the transfer function gain that have converged, as the noise estimation parameters.