Weighted Feature Vector Generation for Speaker Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating feature vectors from long-time voice data equally weight all frames, which can lead to reduced speaker recognition capability due to varying certainty of speaker features across frames.
Innovation Solution
A signal processing system that generates a first feature vector, calculates weights for each frame, and computes a weighted average and high-order statistical vector to produce a second feature vector, enhancing class recognition capability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all frames are weighted equally in calculating statistical amounts, then the processing is simple, but the speaker recognition capability is reduced due to varying certainty of speaker features across frames
Solution Approach 1:
The patent applies different weights to different frames based on their local quality or reliability. The weight calculation unit computes specific weight values for each frame according to the certainty of speaker features in that frame, rather than using uniform weights. This allows frames with higher speaker feature certainty to contribute more to the statistical amounts, thereby improving speaker recognition capability while maintaining a relatively simple processing framework.
2Measurement precision
If weights are calculated for each frame to improve speaker recognition, then recognition accuracy improves, but processing complexity increases
Solution Approach 1:
The patent changes the parameter of weight values assigned to different frames. By calculating and applying varying weights based on frame-specific characteristics (such as speaker feature certainty), the system optimizes the contribution of each frame to the statistical amounts. This parameter change improves class recognition accuracy while the weight calculation methodology keeps the increased complexity manageable.
3Reliability
If weighted statistical amounts are calculated, then feature vector quality improves, but computational load increases
Solution Approach 1:
The patent performs preliminary weight calculation for each frame before computing the weighted statistical amounts. By pre-determining the weight values based on frame characteristics, the system prepares the necessary parameters in advance, which streamlines the subsequent calculation of weighted average vectors and high-order statistical vectors. This preliminary action improves feature vector reliability while optimizing the distribution of computational power usage.
Data Source
AI summary
A feature vector having high class identification capability is generated. A signal processing system provided with: a first generation unit for generating a first feature vector on the basis of one of time-series voice data, meteorological data, sensor data, and text data, or on the basis of a feature quantity of one of these; a weight calculation unit for calculating a weight for the first feature vector; a statistical amount calculation unit for calculating a weighted average vector and a weighted high-order statistical vector of second or higher order using the first feature vector and the weight; and a second generation unit for generating a second feature vector using the weighted high-order statistical vector.