Acoustic Source Separation Using Spherical Harmonics in Reverberation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for acoustic source direction of arrival estimation and separation using microphone arrays face challenges in real-time operation, especially in reverberant environments, and struggle with adaptation to different microphone arrays, particularly when multiple sound sources overlap in the time-frequency domain.
Innovation Solution
A method utilizing spherical harmonic decomposition coefficients and a dictionary-based representation of the steered response function, employing spatial weighting and orthogonal matching pursuit, to separate acoustic sources by calculating directional maps and applying directional weighting for accurate source separation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing methods are used for acoustic source direction of arrival estimation and separation, then basic functionality is achieved, but performance deteriorates in reverberant environments and real-time operation
Solution Approach 1:
The patent segments the acoustic signal processing into spherical harmonic decomposition and spatially weighted spectral representation. By decomposing the sound field into spherical harmonic coefficients and then applying spatial weighting in the spectral domain, the method separates directional information from reverberant components, improving source separation performance in reverberant environments.
Solution Approach 2:
The patent transforms the problem from traditional time-frequency domain processing to spherical harmonic domain processing. By representing the sound field using spherical harmonic decomposition and applying spatial weighting in this transformed domain, the method extracts directional information more effectively, achieving 8-10dB improvement in SIR, SDR, and SAR metrics.
2Adaptability or versatility
If traditional beamforming methods are used, then source separation is achieved, but adaptation to different microphone arrays is difficult
Solution Approach 1:
The patent employs spherical harmonic decomposition which provides a universal representation of the sound field that is independent of microphone array geometry. The spherical harmonic coefficients can represent any microphone array configuration, and the spatial weighting operation adapts to different arrays through the spectral representation, enabling the method to work with various array geometries without fundamental changes.
Solution Approach 2:
The patent changes the representation parameters from traditional time-domain or frequency-domain signals to spherical harmonic domain parameters. By expressing the sound field in terms of spherical harmonic coefficients and applying spatial weighting through spectral representation, the method adapts to different microphone arrays by changing the mathematical parameters of the representation rather than the physical array configuration.
3Measurement precision
If complex separation algorithms are used, then source separation accuracy improves, but real-time operation becomes difficult
Solution Approach 1:
The patent replaces complex iterative optimization algorithms with a direct spectral weighting approach. Instead of using computationally intensive iterative methods for source separation, the method applies spatial weighting directly in the spherical harmonic spectral domain, which provides accurate direction of arrival estimation while maintaining computational efficiency for real-time operation.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention is related to a method that enables acoustic source direction of arrival estimation and acoustic source separation, via spatial weighting of the dictionary based display of the steered response function calculated for a certain number of directions from spherical harmonic decomposition coefficients obtained from microphone array recordings of the sound field. The usage of spatial band limited functions of plane waves to represent more complex directional maps of the sound field constitutes the algorithm of the invention. These functions are calculated for pre-defined directions on an analysis surface (such as a sphere). The directions of arrival of sound sources are calculated with the same method in order to group source estimates to localize sound sources. Thereby, directions of arrival can be obtained from the recordings of the sound sources captured by means of a microphone array and following this, sound sources can be separated by using this direction information or predetermined source arrival directions.