Audio Noise Reduction via Spectral Slice Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction methods for audio signals fail to effectively suppress background noise, especially in environments with low signal-to-noise ratios and high noise variability, leading to poor noise reduction results.
Innovation Solution
A method that processes audio signals by dividing them into spectral time slices, selecting slices based on their spectra, and using these selected slices to compose an output signal for noise periods, thereby reducing targeted noise while preserving desirable audio components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If spectral attenuation or mean square error minimization methods are used for noise reduction, then noise suppression works well for high signal-to-noise ratio signals, but noise reduction performance deteriorates for low signal-to-noise ratio signals with high noise variability
Solution Approach 1:
The patent implements dynamic noise reduction by continuously tracking noise characteristics over time and adapting the filtering parameters accordingly. The system transitions from static filtering to dynamic adaptation, where the noise profile is updated based on recent signal statistics, allowing effective noise suppression across varying noise conditions including low signal-to-noise ratio environments with high noise variability.
Solution Approach 2:
The patent changes the parameters used for noise estimation from fixed or slowly adapting values to dynamically updated parameters that reflect current noise conditions. By modifying how noise statistics are calculated and applied in real-time, the system achieves better adaptability to different noise scenarios while maintaining reliability across diverse signal-to-noise ratio conditions.
2Object-affected harmful factors
If aggressive noise filtering is applied to low signal-to-noise ratio signals, then noise levels are reduced, but audio quality and naturalness deteriorate
Solution Approach 1:
The patent applies local quality by treating different frequency bands and time segments of the audio signal with different filtering strengths. Instead of uniform noise reduction across the entire signal, the system adapts the filtering intensity to local characteristics, preserving important audio information while suppressing noise in specific regions where it is most harmful.
Solution Approach 2:
The patent incorporates feedback mechanisms where the output of the noise reduction process is monitored and used to adjust subsequent filtering operations. This closed-loop approach prevents over-filtering by detecting when noise reduction begins to degrade audio quality, allowing the system to maintain the optimal balance between noise suppression and audio preservation.
Data Source
AI summary
Methods, machines, systems and machine-readable instructions for processing input audio signals are described. In one aspect, an input audio signal has a noise period that includes a targeted noise signal and a noise-free period free of the targeted noise signal. The input audio signal in the noise-free period is divided into spectral time slices each having a respective spectrum. Ones of the spectral time slices of the input audio signal are selected based on the respective spectra of the spectral time slices. An output audio signal is composed for the noise period based at least in part on the selected ones of the spectral time slices of the input audio signal in the noise-free period.