Environment-Aware Acoustic Noise Reduction for Non-Stationary Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing acoustic noise reduction techniques are inadequate for non-stationary and rapidly varying noise environments, as they are primarily designed for stationary or slowly varying noise, leading to ineffective noise reduction and potential speech distortion.
Innovation Solution
A user environment-aware acoustic noise reduction system that identifies the current user environment by comparing feature vectors of a noisy signal with classification data, estimates noise levels and speech presence probability, and adjusts noise reduction parameters accordingly to enhance the signal quality while minimizing distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If spectral subtraction techniques are used for noise reduction, then noise reduction capability is improved, but effectiveness deteriorates in non-stationary and rapidly varying noise environments
Solution Approach 1:
The patent applies dynamics by making the noise reduction system adaptive to changing environments. It uses environmental classification to dynamically adjust noise reduction parameters based on the detected user environment type (e.g., car, pub, café, office). This allows the system to transition from static spectral subtraction to dynamic environment-aware processing, resolving the contradiction between noise reduction capability and reliability in non-stationary conditions.
Solution Approach 2:
The patent changes parameters by introducing environment-specific processing parameters. Different user environments have different noise characteristics, and the system adjusts noise reduction parameters (such as spectral subtraction factors, filtering characteristics) according to the classified environment. This parameter adaptation enables effective noise reduction across varying noise conditions while maintaining speech quality.
2Adaptability or versatility
If environment classification is added to identify current user environment, then adaptability to varying noise conditions is improved, but device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the noise reduction process into distinct stages: environmental classification stage and noise reduction stage. The classification module separately identifies user environments, and the noise reduction module separately processes audio based on classification results. This segmentation allows the system to add adaptability through classification while keeping the noise reduction core relatively simple and modular.
Solution Approach 2:
The patent uses environmental classification as an intermediary between the audio input and noise reduction processing. The classifier acts as a mediator that analyzes environmental characteristics and provides guidance parameters to the noise reduction module. This intermediary approach enables the system to achieve high adaptability without directly complicating the core noise reduction algorithm, as the classification layer handles environment-specific logic.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Examples of the disclosure describe user environment aware single channel acoustic noise reduction. A noisy signal received by a computing device is transformed and feature vectors of the received noisy signal are determined. The computing device accesses classification data corresponding to a plurality of user environments. The classification data for each user environment has associated therewith a noise model. A comparison is performed between the determined feature vectors and the accessed classification data to identify a current user environment. A noise level, a speech level, and a speech presence probability from the transformed noisy signal are estimated based on the noise model associated with the identified current user environment and the noise signal is reduced based on the estimates. The resulting signal is outputted as an enhanced signal with a reduced or eliminated noise signal.