MinMax Noise Estimator for Nonstationary Signal Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise suppressors are ineffective in environments with non-stationary noise due to difficulties in noise level estimation, which impairs the intelligibility of voice signals in noisy communication channels.
Innovation Solution
An improved noise estimator that tracks both minimum and maximum signal statistics in each frequency band, using a nonstationarity measure to adjust noise estimation, allowing for better handling of non-stationary noise by applying a smoothing factor based on speech presence probability and noise variability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional noise estimators are used, then the system is simple to implement, but noise level estimation is inaccurate in non-stationary noise environments
Solution Approach 1:
The noise estimation process is segmented into multiple frequency bands, with separate minimum and maximum followers tracking different statistical characteristics in each band. This segmentation allows accurate tracking of non-stationary noise while maintaining computational efficiency through parallel independent processing of each frequency band.
Solution Approach 2:
The noise estimator dynamically adapts to changing noise conditions by using two followers with different tracking characteristics (minimum and maximum). The system dynamically selects or combines estimates from these followers based on the current non-stationarity measure, enabling accurate tracking of both stationary and non-stationary noise without manual intervention.
2Reliability
If simple noise estimation is used, then the device complexity is low, but the intelligibility of voice signals deteriorates in noisy environments
Solution Approach 1:
The system implements feedback through the non-stationarity measure that monitors the reliability of noise estimates in real-time. Based on this feedback, the system dynamically adjusts the weighting between minimum and maximum follower estimates, and controls the smoothing factor applied to gain adjustments, ensuring reliable voice signal processing under varying noise conditions.
Solution Approach 2:
The noise estimator changes parameters dynamically by adjusting the smoothing factor and gain adjustment weights based on the measured non-stationarity of the noise. When noise is highly non-stationary, the system modifies its estimation parameters to track rapid changes; when noise is stationary, it uses smoother estimates, optimizing voice signal intelligibility across different environmental conditions.
3Object-affected harmful factors
If aggressive noise suppression is applied, then noise reduction is improved, but artifacts and distortion increase
Solution Approach 1:
The system applies partial noise suppression by using a smoothing factor that is adjusted based on the non-stationarity measure. Instead of applying maximum suppression always, the system applies just enough suppression to reduce noise while preserving speech quality. The smoothing factor controls the degree of suppression, preventing excessive action that would create artifacts while still achieving effective noise reduction when appropriate.
Data Source
AI summary
A noise-level estimator for a noise suppressor includes a power smoother filter providing smoothed power estimates in timeslices, a minimum follower that represents the lowest smoothed input power, and a maximum follower that represents the highest smoothed input power, the followers subject to leakage factors. The estimator has a speech probability detector receiving outputs of the power smoother and minimum follower; a nonstationary noise detector receiving outputs of both followers; and an estimator receiving outputs of the nonstationary noise detector, power smoother, and speech probability detector and providing a noise estimate. The method includes smoothing intensity of the frequency band; tracking minima and maxima of the smoothed intensity; determining speech-absence probability from the minima and the intensity; determining a nonstationary noise measure from the tracked minima and maxima; determining presence of nonstationary noise; and estimating noise from speech-absence probability, the nonstationary noise measure, and the intensity.