Cross-Modal Multimedia Capture for Adaptive Noise Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multimedia terminals face challenges in managing background noise during multimedia capturing, as noise suppression methods cannot optimally distinguish between voice and background noise, leading to suboptimal audio quality and context-dependent noise handling.
Innovation Solution
A method that captures multimedia content by obtaining control information from either the audio or video signal to dynamically adjust pre-processing, such as noise suppression, voice level adjustment, and microphone beamforming, based on priority values determined by the image area covered by audio sources or objects in the video scene, allowing for context-dependent noise management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If fixed trade-off noise suppression is applied, then background noise attenuation is improved, but voice signal degradation increases
Solution Approach 1:
The patent applies dynamics by making the noise suppression characteristics adjustable and adaptive rather than fixed. The system dynamically modifies noise suppression parameters based on detected audio characteristics (voice presence, noise level) and user preferences, allowing the trade-off between noise attenuation and voice quality to be optimized in real-time for different scenarios
Solution Approach 2:
The patent implements parameter changes by modifying noise suppression parameters (such as attenuation levels, frequency ranges, time constants) based on detected conditions. The system analyzes audio signals to determine voice presence and noise characteristics, then adjusts suppression parameters accordingly to maintain voice quality while reducing background noise
2Object-affected harmful factors
If aggressive noise suppression is applied, then background noise reduction is improved, but ambience preservation deteriorates
Solution Approach 1:
The patent applies local quality by treating different frequency ranges and time periods differently in the noise suppression process. Instead of uniform suppression, the system identifies and preserves frequency bands and temporal segments that contain ambience information while aggressively suppressing harmful noise in other regions of the audio spectrum
Solution Approach 2:
The system dynamically adjusts suppression aggressiveness based on detected audio content. When ambience or non-speech sounds are detected, the system reduces suppression intensity for those specific time periods or frequency ranges, allowing ambience information to be preserved while still providing noise reduction for speech segments
Data Source
AI summary
A method for controlling capturing of multimedia content, the method comprising: capturing multimedia content by an apparatus, said multimedia content comprising at least an audio signal and a video signal; obtaining control information based on at least either of said audio signal or video signal; controlling pre-processing of the audio signal based on the control information obtained from the video signal; and/or controlling pre-processing of the video signal based on the control information obtained from the audio signal.