Linear Predictive Analysis Apparatus Dynamic Coefficient Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional linear predictive analysis methods for audio and acoustic signals suffer from reduced precision due to the use of fixed coefficients in modifying autocorrelation, leading to degraded spectral envelope approximation when the spectral peak of the input signal is high.
Innovation Solution
A method that dynamically adjusts the coefficient used for modifying autocorrelation based on the intensity of periodicity or pitch gain of the input signal, allowing for a coefficient that monotonically decreases with increasing intensity, thereby improving the precision of linear predictive analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed coefficient is used for modifying autocorrelation, then the device complexity is reduced and ease of operation is improved, but the manufacturing precision and measurement precision of linear predictive analysis deteriorate when spectral peaks are high
Solution Approach 1:
The patent applies the dynamics principle by transitioning from a fixed coefficient to a dynamic coefficient that automatically adjusts based on the pitch gain of the input signal. The coefficient determining part selects different coefficients from a table based on the calculated pitch gain, making the system adaptive to varying signal characteristics. This resolves the contradiction by maintaining ease of operation through automated selection while improving measurement precision through context-appropriate coefficient selection.
Solution Approach 2:
The patent implements parameter changes by modifying the coefficient value based on the pitch gain parameter of the input signal. When pitch gain is high, a coefficient that suppresses spectral peaks is selected; when pitch gain is low, a coefficient that maintains spectral envelope accuracy is selected. This dynamic parameter adjustment resolves the contradiction by optimizing the coefficient parameter according to the actual signal conditions, thereby improving measurement precision without complicating operation.
2Device complexity
If a fixed coefficient is used for modifying autocorrelation, then the device complexity is reduced, but the manufacturing precision of spectral envelope approximation deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing multiple coefficients in a coefficient table before actual signal processing. The coefficient determining part simply selects the appropriate pre-computed coefficient based on pitch gain, avoiding complex real-time calculations. This resolves the contradiction by maintaining low device complexity through lookup table selection while achieving high manufacturing precision through optimized coefficients prepared in advance for different signal conditions.
Solution Approach 2:
The patent uses parameter changes by selecting different coefficient values from the table based on the pitch gain parameter. The system changes the coefficient parameter dynamically to match the signal's periodicity characteristics, improving spectral envelope approximation precision without adding significant device complexity since the changes are achieved through simple table lookup and selection.
3Ease of operation
If autocorrelation is multiplied by a fixed coefficient, then the ease of operation is improved, but the reliability of linear prediction deteriorates when pitch gain is high
Solution Approach 1:
The patent applies dynamics by making the coefficient adaptive rather than fixed. The coefficient determining part dynamically selects coefficients based on the pitch gain of the input signal, allowing the system to respond appropriately to different signal conditions. This resolves the contradiction by maintaining ease of operation through automated dynamic selection while improving reliability by using coefficients optimized for high pitch gain conditions when needed.
Solution Approach 2:
The patent implements feedback by using the calculated pitch gain as feedback to select the appropriate coefficient. The pitch gain calculation result feeds back into the coefficient selection process, creating a closed-loop system that automatically adjusts the coefficient based on signal characteristics. This resolves the contradiction by maintaining ease of operation through automated feedback-based selection while improving reliability by ensuring the coefficient matches the actual signal conditions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An autocorrelation calculating part 21 calculates autocorrelation Ro(i) from an input signal. A predictive coefficient calculating part 23 performs linear predictive analysis using modified autocorrelation R'o(i) obtained by multiplying the autocorrelation Ro(i) by a coefficient wo(i). Here, a case is comprised where, for at least part of each order i, the coefficient wo(i) corresponding to each order i monotonically decreases as a value having positive correlation with a pitch gain in an input signal of a current frame or a past frame increases.