Multichannel Sound Source Identification Using Collaborative PLCS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multichannel sound source identification techniques fail when sound sources collide in frequencies, as they rely on the orthogonality of source spectrums, which is not always valid.
Innovation Solution
The use of a collaborative technique for source separation in multichannel sound data, employing Probabilistic Latent Component Sharing (PLCS) to recognize spectral and temporal aspects and share them across channels, allowing for the identification and positioning of sound sources even when sources overlap.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional source separation techniques relying on spectral orthogonality are used, then the method is simple and computationally efficient, but the technique fails when sound sources collide in frequencies
Solution Approach 1:
The sound spectrum is segmented into multiple frequency bins, and the source separation process is divided into iterative steps where spectral and temporal aspects are separately recognized and then combined. This segmentation allows the system to handle frequency collisions by processing different frequency regions independently and combining results.
Solution Approach 2:
The patent transitions from relying solely on spectral dimension (frequency) to incorporating both spectral and temporal dimensions. By adding the temporal aspect as an additional dimension for analysis, the system can distinguish between sound sources that overlap in frequency but differ in time characteristics, thereby resolving the orthogonality limitation.
2Adaptability or versatility
If a collaborative technique with spectral and temporal aspect sharing is used, then sound source identification works even when sources overlap in frequency, but the processing complexity increases
Solution Approach 1:
The patent merges spectral aspect recognition and temporal aspect recognition into a unified collaborative processing framework. By combining these two recognition processes and sharing their results, the system achieves robust sound source separation that works for both orthogonal and overlapping sources, making the method adaptable to various acoustic scenarios.
Solution Approach 2:
The collaborative technique with aspect sharing creates a universal source separation method that can handle multiple types of sound source configurations - both orthogonal sources (traditional case) and overlapping sources (challenging case). This multi-functional approach eliminates the need for different methods for different scenarios.
3Measurement precision
If conventional techniques assuming spectral orthogonality are used, then the computational processing is straightforward, but the method fails to identify sound sources when they collide in one or more frequencies
Solution Approach 1:
The patent introduces spectral and temporal aspects as intermediary representations that mediate between the raw multichannel sound data and the final sound source identification. These aspects serve as intermediate processing stages that decompose the complex problem of separating overlapping sources into more manageable spectral and temporal components that can be analyzed separately and then integrated.
Data Source
AI summary
Multichannel sound source identification and location techniques are described. In one or more implementations, source separation is performed using a collaborative technique for a plurality of sound data that was captured by respective ones of a plurality of sound capture devices of an audio scene. The source separation is performed by recognizing spectral and temporal aspects from the plurality of sound data and sharing the recognized spectral and temporal aspects, one with another, to identify one or more sound sources in the audio scene. A relative position of the identified one or more sounds sources to the plurality of sound capture devices is determined based on the source separation.