Source Separation Using Extended Observation Signal Vector
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods such as FCA and FCAd struggle to achieve sufficient separation performance, especially when the reverberation time is long, due to the increase in the number of algorithm parameter updates.
Innovation Solution
A source separation apparatus that creates a second observation signal vector by extending the first observation signal vector to consider time delays, optimizes parameters including a correlation matrix and signal source power using a predetermined algorithm, and separates target signals using these optimized parameters and the extended observation signal vector.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional FCA or FCAd methods are used for source separation, then the separation can be performed with basic algorithms, but the separation performance deteriorates as reverberation time increases and algorithm repetitions increase
Solution Approach 1:
The patent extends the observation signal vector from a single time frame to multiple time frames, adding a temporal dimension to the separation process. This allows the algorithm to capture time-delayed reverberation components that are invisible in single-frame analysis, thereby maintaining separation performance even in highly reverberant environments.
Solution Approach 2:
The patent pre-constructs a time delay set based on the room impulse response before performing source separation. This preliminary action enables the algorithm to anticipate and account for reverberation paths in advance, improving separation accuracy without requiring excessive algorithm repetitions during the actual separation process.
2Measurement precision
If the number of algorithm repetitions is increased to improve separation accuracy, then parameter optimization may improve, but separation performance deteriorates when reverberation time is long
Solution Approach 1:
By incorporating multiple time frames into the observation signal vector, the patent enables the algorithm to capture reverberation information more efficiently. This dimensional extension allows for accurate separation with fewer algorithm repetitions, as the temporal information is already embedded in the extended vector rather than requiring iterative refinement.
Solution Approach 2:
The patent optimizes parameters including a correlation matrix representing transmission characteristics and signal source power at each time, considering time delays. This parameter optimization approach, combined with the extended observation vector, achieves high separation performance without requiring excessive algorithm repetitions, thus balancing accuracy and efficiency.
3Measurement precision
If short-time Fourier transform is used to convert time waveform to frequency waveform, then frequency analysis is enabled, but reverberation components outside the time frame length are not captured
Solution Approach 1:
The patent compensates for the information loss in STFT by extending the observation signal vector across multiple time frames. This temporal extension ensures that reverberation components that fall outside individual time frame boundaries are captured in subsequent frames, preserving complete reverberation information while maintaining frequency analysis capability.
Solution Approach 2:
The patent processes a sequence of continuous time frames rather than isolated frames, ensuring that reverberation energy is continuously tracked across frame boundaries. This continuous processing approach prevents loss of reverberation information that would occur if analysis were restricted to single, isolated time frames.
Data Source
AI summary
A source separation apparatus includes circuitry configured to create a second observation signal vector by using (i) a first observation signal vector representing an observation signal that is obtained by mixing target signals from a plurality of signal sources and (ii) a time delay set in which a time delay until a part of the target signals is observed is used as an element, the second observation signal vector being obtained by extending the first observation signal vector in consideration of the time delay; optimize, by using a predetermined algorithm, a parameter for a correlation matrix representing a transmission characteristic in consideration of the time delay for the target signals, and power of a given signal source at each time; and separate the target signals by using an optimized parameter and the second observation signal vector.