Acoustic Event Detection Using Prior Distribution Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing method for detecting acoustic events using Nonnegative Matrix Factorization (NMF) has low identification accuracy, particularly in unknown environments, due to the presence of local solutions and the inability to accurately estimate basis matrices, leading to false identifications and incomplete sound source separation.
Innovation Solution
A signal processing device and method that utilize a prior distribution storage to constrain the generation of acoustic event bases, using semi-supervised NMF to learn both known and unknown event bases, thereby improving the accuracy of acoustic event detection and sound source separation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If NMF is used for acoustic event detection and sound source separation, then the method can process acoustic signals and identify acoustic events, but the identification accuracy is insufficient and false identifications occur in unknown environments
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing prior distribution information representing typical acoustic event characteristics before actual detection. This prior distribution is computed from training data and stored in a prior distribution storage unit, enabling the system to have advance knowledge of expected acoustic patterns. During detection, this pre-computed prior distribution constrains the NMF process, guiding the basis matrix toward physically meaningful solutions and preventing false identifications in unknown environments.
Solution Approach 2:
The patent changes the parameter space of NMF by introducing a prior distribution constraint that modifies the basis matrix optimization process. Instead of relying solely on the standard NMF objective function, the system incorporates additional constraints based on prior distribution information, effectively changing how the basis matrix parameters are determined. This parameter modification enables more accurate separation of acoustic events from mixed signals.
2Measurement precision
If NMF is used to generate basis matrix from learning data, then acoustic events can be detected, but many local solutions exist and accurate basis matrix estimation fails
Solution Approach 1:
The patent introduces prior distribution information as an intermediary element that mediates between the learning data and the basis matrix generation process. This prior distribution acts as a guide or constraint that helps the NMF algorithm navigate through the complex solution space, preventing convergence to incorrect local solutions. The prior distribution storage unit and the basis generation unit work together, with the prior distribution serving as an intermediate representation that bridges training data and operational detection.
3Adaptability or versatility
If standard NMF is used for sound source separation, then spectral bases can be extracted, but the method cannot handle unknown sounds and produces incomplete separation
Solution Approach 1:
The system performs preliminary action by pre-computing prior distribution information from training data that represents known acoustic events. This prior knowledge is stored and then used to constrain the NMF process during actual separation tasks. The prior distribution acts as a reference framework that guides the separation of both known and unknown sounds, ensuring that the extracted spectral bases align with physically meaningful acoustic patterns while still allowing flexibility for unknown sources.
Data Source
AI summary
A signal processing device includes a prior distribution storage that stores a prior distribution group that is a set of prior distributions representing representative spectral shapes of spectral bases of an acoustic event specified as a detection target, and a basis generation unit that, using as input a spectrogram for basis generation, generates an acoustic event basis group that is a set of the spectral bases of the acoustic event specified as a detection target. The basis generation unit performs sound source separation on the spectrogram for basis generation and thereby generates an acoustic event basis group, using respective prior distributions included in a prior distribution group as a constraint for spectral bases of a corresponding acoustic event.