Adaptive Noise Estimation via Variable Increments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction systems struggle to accurately estimate the power spectral density of background noise in real-time, especially in environments with rapid changes, leading to inadequate suppression of background noise and potential corruption of voice signals.
Innovation Solution
A system comprising a sensor unit, power spectral density calculation unit, time and frequency domain smoothing units, increment and decrement calculation units, and an estimate signal smoothing unit, which continuously adapts to changes in background noise levels by adjusting increment and decrement values based on previous estimates, allowing for faster response to changes in noise levels without significant voice signal interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voice detection methods are employed to prevent unwanted reduction in voice signal, then voice signal quality is improved, but implementation outlay becomes unattractively high
Solution Approach 1:
The patent extracts and removes voice signal components from the noise estimation process. By detecting and excluding voice signal portions from the power spectral density estimation, the system prevents voice signal corruption without requiring complex voice detection and control mechanisms. This extraction approach simplifies the overall system while maintaining voice signal quality.
Solution Approach 2:
The patent introduces an intermediary approach by using a simplified voice activity indicator that provides coarse voice signal detection. This intermediary mechanism enables basic voice protection without the complexity of full voice detection systems, allowing the noise reduction to proceed safely during non-voice periods while avoiding high implementation costs.
2Reliability
If smoothing filter with constant increments is used to estimate power spectral density, then voice signal corruption is avoided, but response speed to background noise changes becomes too slow
Solution Approach 1:
The patent applies dynamics by making the increment values adaptive rather than constant. The system dynamically adjusts the increment amounts based on the detected background noise characteristics and temporal patterns. This allows the estimator to respond quickly to genuine noise changes while maintaining stability during steady-state conditions, resolving the contradiction between response speed and estimate accuracy.
Solution Approach 2:
The patent changes the parameters of the smoothing filter by using variable increment values instead of fixed constants. The increment parameters are modified based on the observed noise level changes and temporal dynamics, enabling the system to accelerate its response to noise changes while preserving the smoothing benefits that prevent voice signal corruption.
3Object-affected harmful factors
If noise reduction is applied to improve signal quality, then background noise suppression is improved, but voice signal is also reduced which is not wanted
Solution Approach 1:
The patent segments the signal processing into distinct phases: voice activity detection, noise estimation during non-voice periods, and selective noise reduction application. By dividing the processing into these segments, the system applies noise reduction only when voice signals are absent, thereby improving background noise suppression while preserving voice signal integrity.
Solution Approach 2:
The patent employs preliminary anti-action by pre-identifying voice signal portions and marking them for protection before noise reduction is applied. This preliminary detection and protection mechanism ensures that when noise reduction processing occurs, voice signals are already identified and shielded from unwanted attenuation, thus maintaining voice signal integrity while achieving noise suppression.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A system and method for estimating the power spectral density of acoustical background noise is presented where for cases in which the level of a smoothed power spectral density signal increases, an increment value is increased, starting from a minimum increment value, by a predetermined amount until a maximum increment value is reached if at the same time the value of the power spectral density currently determined in a new calculation cycle is larger than the estimate value of the power spectral density of the background noise determined in the previous calculation cycle. For cases in which the level of the smoothed power spectral density falls, a decrement value is increased, starting from a minimum decrement value, by a predetermined amount until a maximum decrement value is reached if at the same time the value of the power spectral density currently determined in a new calculation cycle is smaller than the estimate value of the power spectral density of the background noise determined in the previous calculation cycle.