Adaptive Voice Recognition Threshold for Vehicle Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing diversification of vehicle functions leads to higher user interaction loads, distracting drivers and compromising safe driving, necessitating a reduction in user-manipulated loads through an effective voice recognition technology.
Innovation Solution
A vehicle system equipped with a communication device, voice input device, storage, and controller that recognizes user voices, searches for name candidates based on input voices, determines confidence scores, and selects corresponding names, allowing for voice-activated functions like phone calls by integrating phone book and call history information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If voice recognition technology is implemented to reduce user manipulation loads, then ease of operation is improved, but measurement precision of voice recognition may deteriorate due to background noise and ambiguous commands
Solution Approach 1:
The system dynamically adjusts the confidence threshold parameter based on the number of name candidates and their confidence scores. When multiple candidates exist with similar scores, the threshold is lowered to prevent false rejections. When a clear winner exists, the threshold is raised to ensure accuracy. This adaptive parameter adjustment resolves the contradiction by making the recognition system both sensitive and selective as needed.
Solution Approach 2:
The system implements feedback mechanisms where recognition results are evaluated against confidence scores and thresholds. The outcome feeds back into adjusting future recognition decisions, creating a learning loop that improves accuracy over time while maintaining ease of operation through automated adjustments.
2Measurement precision
If multiple name candidates are considered to improve recognition accuracy, then measurement precision is improved, but device complexity increases due to additional processing steps
Solution Approach 1:
The system considers multiple name candidates beyond the single best match, evaluating top candidates with confidence scores above a dynamic threshold. This partial expansion of candidate consideration improves accuracy without exhaustively processing all possible candidates, thus avoiding excessive complexity while maintaining enhanced recognition precision.
3Reliability
If confidence thresholds are adjusted dynamically to improve recognition reliability, then reliability is improved, but device complexity increases due to adaptive threshold management
Solution Approach 1:
The system performs self-adjustment of confidence thresholds based on the distribution of candidate confidence scores. The controller automatically determines appropriate thresholds by analyzing the number of candidates and their score distributions, eliminating the need for external configuration or complex manual threshold management while maintaining high reliability.
Data Source
AI summary
A vehicle includes a communication device configured to communicate with a terminal capable of providing a communication function; a sensor configured to receive voice of a user; a storage configured to store a user pattern related to a call pattern of the user; and a controller configured to search for at least one name candidate corresponding to input voice when receiving the input voice, determine a threshold for a confidence score of the at least one name candidate based on the user pattern, and select a name corresponding to the input voice from among the at least one name candidate based on the determined threshold.