Signal Classification Using Spectrum Fluctuation Variance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal classifying methods based on decision trees are complex, involving excessive parameter calculations and logical branches, making them inefficient for accurately classifying speech and music signals.
Innovation Solution
A signal classifying method that uses spectrum fluctuation parameters to classify signals by buffering and calculating ratios of frames with specific variance thresholds, employing a local statistical approach with adaptive thresholds based on Modified Sub-band Signal Noise Ratio (MSSNR) or Signal-to-Noise Ratio (SNR), thereby simplifying the logical relations and reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a decision tree method is used for signal classification, then signal types can be identified, but the classification process becomes complex with excessive parameter calculations and logical branches
Solution Approach 1:
The patent extracts only the essential feature (spectrum flux) needed for signal classification, discarding the complex decision tree structure with multiple parameters and logical branches. By focusing on a single key parameter and its statistical properties, the method achieves accurate classification while dramatically reducing computational complexity.
Solution Approach 2:
The patent transforms the classification approach by changing from multiple parameters in a decision tree to a single parameter (spectrum flux) analyzed through statistical methods. The use of variance and local statistical analysis converts a complex multi-parameter problem into a simpler single-parameter statistical evaluation, resolving the contradiction between accuracy and complexity.
2Reliability
If multiple parameters and logical branches are used in decision trees, then comprehensive signal analysis is achieved, but computational complexity increases significantly
Solution Approach 1:
The patent employs local statistical methods that allow the spectrum flux variance to self-evaluate the signal characteristics. The variance calculation and local statistics automatically adapt to different signal types without requiring complex external decision logic, achieving reliable classification with minimal computational overhead.
3Adaptability or versatility
If traditional decision tree methods are applied, then signal classification can be performed, but the logical relations become overly complex
Solution Approach 1:
The patent segments the signal analysis into distinct statistical evaluations (variance calculation, local statistics computation) based on a single spectrum flux parameter. This segmentation replaces the tangled logical branches of decision trees with clear, separate computational steps that are easier to implement and maintain while preserving adaptability to different signal types.
Data Source
AI summary
A signal classifying method and apparatus are disclosed. The signal classifying method includes: obtaining a spectrum fluctuation parameter of a current signal frame determined as a foreground frame, and buffering the spectrum fluctuation parameter; obtaining a spectrum fluctuation variance of the current signal frame according to spectrum fluctuation parameters of all buffered signal frames, and buffering the spectrum fluctuation variance; and calculating a ratio of signal frames whose spectrum fluctuation variance is above or equal to a first threshold to all the buffered signal frames, and determining the current signal frame as a speech frame if the ratio is above or equal to a second threshold or determining the current signal frame as a music frame if the ratio is below the second threshold. In the embodiments of the present invention, the spectrum fluctuation variance of the signal is used as a parameter for classifying the signals, and a local statistical method is applied to decide the type of the signal. Therefore, the signals are classified with few parameters, simple logical relations and low complexity.