Dynamic Speaker Recognition Database Update Mechanism
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional speaker recognition systems face performance degradation due to changes in a speaker's voice caused by emotions, age, and environmental factors, as they typically rely solely on initially registered voice features, leading to inaccurate authentication of authorized users.
Innovation Solution
A processor-implemented method that determines whether an input feature vector meets specific similarity criteria, constructs candidate lists, and selectively updates a registration database based on these criteria, allowing for dynamic adaptation to changes in voice characteristics over time and environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speaker recognition is performed only based on initially registered voice features, then the system is simple and fast, but recognition performance degrades due to voice changes from emotions, age, and health state
Solution Approach 1:
The registration database is dynamically updated by replacing old voice features with new voice features when voice changes are detected. This allows the system to adapt to voice changes over time while maintaining a manageable database structure, resolving the contradiction between reliability and complexity.
Solution Approach 2:
The system changes the parameter of voice feature storage by transitioning from static initial registration to dynamic continuous updating. When voice characteristics change, the system updates the registered voice features accordingly, maintaining recognition accuracy without requiring complete system redesign.
2Reliability
If the registration database is continuously updated with new voice features, then speaker recognition accuracy improves under varying conditions, but the complexity of determining when to update increases
Solution Approach 1:
The system uses feedback from voice quality assessment to determine when updates should occur. By continuously monitoring voice characteristics and comparing them against registered features, the system automatically triggers updates only when significant changes are detected, simplifying the update determination process while maintaining accuracy.
Solution Approach 2:
The system performs self-assessment of voice quality and automatically determines when updates are needed without external intervention. The voice quality assessment module evaluates voice characteristics and triggers updates autonomously, reducing the complexity of manual update determination.
3Adaptability or versatility
If voice features are updated automatically without user intervention, then the system adapts to voice changes efficiently, but it may inadvertently update with unwanted voice characteristics
Solution Approach 1:
The system applies different quality thresholds and assessment criteria for different voice characteristics. By evaluating specific voice parameters (pitch, tone, rhythm) individually and applying localized quality judgments, the system adapts to legitimate voice changes while filtering out unwanted or erroneous updates, maintaining authentication accuracy.
Data Source
AI summary
Disclosed is a method and apparatus with recognition for speaker recognition. The method includes determining whether an input feature vector corresponding to a voice signal of a speaker meets a candidate similarity criterion with at least one registered data included in a registration database, selectively, based on a result of the determining of whether the input feature vector meets the candidate similarity criterion, constructing a candidate list based on the input feature vector, determining whether a candidate input feature vector, among one or more candidate input feature vectors constructed in the candidate list in the selective constructing of the candidate list, meets a registration update similarity criterion with the at least one registered data, and selectively, based on a result of the determination of whether the candidate input feature vector meets the registration update similarity criterion, updating the registration database based on the candidate input feature vector.