User Adaptive Speech Recognition Threshold Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional speech recognition techniques require frequent user confirmation, leading to inconvenience and a lower success rate due to the use of a uniform threshold value that does not adapt to individual speech features.
Innovation Solution
A user-adaptive speech recognition method that calculates a confidence score for recognition candidates and sets a new threshold value based on user confirmation and confidence scores, allowing for automatic output of recognition results when the score exceeds the adapted threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a uniform threshold value is applied to every user, then the speech recognition system is simple to operate, but the frequency of user confirmations increases and the success rate decreases
Solution Approach 1:
The patent implements dynamic threshold adjustment by continuously adapting the recognition threshold based on each user's speech characteristics and confirmation feedback. The system transitions from a static uniform threshold to a dynamic user-specific threshold that evolves over time, resolving the contradiction between maintaining simple operation and improving recognition success rate.
Solution Approach 2:
The system changes the threshold parameter based on user-specific data. By collecting confirmation results and speech features, the system adjusts the threshold value to optimize recognition accuracy for each individual user, thereby improving success rate without significantly increasing operational complexity.
2Reliability
If a uniform threshold value is applied to every user, then the system complexity is low, but the speech recognition success rate decreases
Solution Approach 1:
The system performs self-adjustment by automatically adapting thresholds based on user feedback and speech characteristics. The threshold management becomes self-service rather than requiring manual configuration, which improves recognition success rate while keeping the system complexity manageable through automated adaptation.
Solution Approach 2:
The system uses user confirmation feedback to continuously refine and adapt the threshold value. This feedback mechanism enables the system to improve recognition success rate dynamically without requiring complex manual threshold management, as the adaptation occurs automatically based on observed user responses.
3Reliability
If user confirmation is required for every recognition result, then the speech recognition success rate is verified, but the user convenience decreases and time is lost
Solution Approach 1:
The system applies user confirmation only partially - not for every recognition result. By adapting the threshold to user-specific characteristics, the system can confidently output results without confirmation in many cases, thereby maintaining accuracy verification while significantly reducing the time lost to confirmations.
Solution Approach 2:
The system performs preliminary adaptation during an initial phase to learn user speech characteristics and set optimal thresholds. This preliminary action enables subsequent recognition operations to proceed with higher accuracy and reduced confirmation requirements, minimizing time loss while maintaining reliability.
Data Source
AI summary
A user adaptive speech recognition method and apparatus is disclosed that controls user confirmation of a recognition candidate using a new threshold value adapted to a user. The user adaptive speech recognition method includes calculating a confidence score of a recognition candidate according to the result of speech recognition, setting a new threshold value adapted to the user based on a result of user confirmation of the recognition candidate and the confidence score of the recognition candidate, and outputting a corresponding recognition candidate as a result of the speech recognition if the calculated confidence score is higher than the new threshold value. Thus, the need for user confirmation of the result of speech recognition is reduced and the probability of speech recognition success is increased.