Sound Pick-up Apparatus Adaptive Spectral Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional area sound pick-up technologies face issues with sound quality deterioration and musical noise in high noise environments, where they struggle to accurately extract target area sounds without introducing distortion or discontinuity, and often fail to effectively reduce noise in sections without target area sounds.
Innovation Solution
A sound pick-up apparatus and method that utilize a beamformer to form directionalities in a target area, perform spectral subtraction to extract non-target area sounds, determine the presence of target area sounds based on amplitude spectra, and adjust mixing levels to minimize noise and improve sound quality by mixing target area sounds with level-adjusted signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spectral subtraction is performed to extract target area sounds, then directionality and sound separation are improved, but musical noise and sound quality deterioration occur in high noise environments
Solution Approach 1:
The patent dynamically adjusts the subtraction coefficient based on noise level detection. When noise is detected, the coefficient is reduced to prevent excessive subtraction that causes musical noise, while still maintaining some noise reduction effect. This parameter adaptation resolves the contradiction by optimizing the subtraction intensity according to actual acoustic conditions.
Solution Approach 2:
The system transitions from static spectral subtraction to dynamic adaptive subtraction. The beamformer directionality patterns are adjusted in real-time based on detected noise levels, allowing the system to maintain effective sound separation while avoiding the musical noise artifact that occurs with fixed aggressive subtraction coefficients.
2Object-generated harmful factors
If aggressive spectral subtraction is applied to reduce noise, then noise reduction is improved, but distortion and discontinuity are introduced in target area sounds
Solution Approach 1:
The system uses feedback from noise level detection to control the subtraction process. The detected noise level feeds into the coefficient adjustment mechanism, which then modulates the subtraction intensity. This closed-loop control prevents over-subtraction that would distort target sounds while maintaining effective noise reduction when appropriate.
Solution Approach 2:
The subtraction coefficient is changed as a function of detected noise characteristics. Rather than using a fixed aggressive coefficient, the system adapts the parameter based on actual noise conditions, reducing the coefficient when noise levels suggest that aggressive subtraction would cause distortion or discontinuity in target area sounds.
3Measurement precision
If beamformer directionality is strengthened to emphasize target area, then sound separation is improved, but noise in non-target areas is not effectively reduced
Solution Approach 1:
The patent combines beamformer directionality with spectral subtraction techniques. The beamformer provides spatial filtering and directionality, while the spectral subtraction component addresses frequency-domain noise characteristics. This merging of spatial and spectral processing methods allows effective noise reduction in non-target areas without compromising the directionality of target area sound extraction.
Data Source
AI summary
The present invention relates to a sound pick-up apparatus. The sound pick-up apparatus according to the present invention includes: a unit configured to acquire target direction signals based on beamformer outputs of a plurality of microphone arrays; a unit configured to extract non-target area sound by performing spectral subtraction processing on the acquired target direction signals, and extract target area sound by performing spectral subtraction in a manner that a spectrum of the non-target area sound is subtracted from spectra of the target direction signals; a unit configured to perform target area sound determination processing for determining whether input signals include the target area sound; a unit configured to decide a level adjustment coefficient for adjusting a level of a mixing signal on the basis of an element including a result of the target area sound determination processing; and a unit configured to mix the extracted target area sound with a level-adjusted mixing signal obtained by adjusting the level of the mixing signal with the decided level adjustment coefficient, and output a mixed signal as an area sound pick-up result.