Sound Separation via Frequency Domain Subtraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound separation techniques fail to accurately extract sound components localized between different reproduction positions from two channel acoustic signals, leading to incomplete or inaccurate sound localization.
Innovation Solution
A sound separation device that generates a differential signal between two acoustic signals and uses frequency domain subtraction to isolate and separate sound components localized between specific positions, allowing for precise extraction and reproduction of targeted sounds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional sound separation techniques are used, then sound localization between reproduction positions can be achieved, but the accuracy of sound component extraction is insufficient
Solution Approach 1:
The patent segments the sound field into different spatial regions by dividing the frequency spectrum into multiple bands and processing each band separately. This allows precise extraction of sound components localized between reproduction positions by handling different frequency content independently, thereby improving both localization accuracy and extraction reliability
Solution Approach 2:
The patent applies different processing characteristics to different frequency bands and spatial regions. By analyzing the correlation between acoustic signals in the time domain and transforming to frequency domain, the system applies localized quality control to extract sound components with specific spatial characteristics, improving extraction accuracy for targeted sound regions
2Device complexity
If two channel acoustic signals are used for sound separation, then the complexity of the system is reduced, but the completeness of sound component extraction deteriorates
Solution Approach 1:
The patent transforms the acoustic signals from the time domain to the frequency domain, adding a dimensional aspect that enables better separation of sound components. By analyzing signals in both time and frequency domains, the system extracts complete sound components from just two channels, achieving high extraction completeness without increasing system complexity
Solution Approach 2:
The patent changes the parameter domain from time to frequency by applying Fourier transform. This parameter transformation reveals additional information about sound components that is not visible in the time domain alone, enabling complete extraction of sound components localized between reproduction positions using only two channel signals
Data Source
AI summary
A sound separation device includes: a signal obtainment unit which obtains a plurality of acoustic signals including a first acoustic signal and a second acoustic signal; a differential signal generation unit which generates a differential signal that is a signal representing a difference in a time domain between the first acoustic signal and the second acoustic signal; an acoustic signal generation unit which generates, using at least one acoustic signal among the acoustic signals, a third acoustic signal; and an extraction unit which generates a frequency signal by subtracting, from a signal obtained by transforming the third acoustic signal into a frequency domain, a signal obtained by transforming the differential signal into a frequency domain, and generates a separated acoustic signal by transforming the generated frequency signal into a time domain.