In-Vehicle Speech Recognition Error Correction via Segmented Address Input
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In-vehicle communication systems often misrecognize speech commands, leading to user frustration and poor experience due to the inefficiency of re-inputting commands to correct mistakes, which exceeds the recommended glance time and task completion time guidelines.
Innovation Solution
The system allows users to re-enter only the portion of the destination address most likely to be incorrect, identified by the system's logic, reducing the need for re-inputting the entire address and minimizing user interface navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the user re-inputs the entire speech command to correct misrecognized input, then the system can potentially recognize the correct command, but the user experience deteriorates due to repeated failures and excessive time required
Solution Approach 1:
The patent segments the address input into multiple fields (state, city, street number, street name) and identifies which specific field is misrecognized. Instead of requiring the user to re-input the entire address, the system only requests correction of the problematic segment, thereby improving ease of operation while maintaining reliability.
Solution Approach 2:
The system applies local quality by treating different portions of the address differently based on recognition confidence. Fields with high confidence are preserved, while fields with low confidence are flagged for user correction. This selective approach improves user experience by minimizing re-input requirements while ensuring accurate recognition where possible.
2Reliability
If the user re-inputs the entire speech command to fix recognition errors, then the system may achieve correct recognition, but the time required exceeds the recommended glance time and task completion guidelines
Solution Approach 1:
By segmenting the address into discrete fields and identifying only the misrecognized portion, the system reduces the amount of information the user must re-input. This segmentation approach directly reduces task completion time while maintaining reliable recognition of the complete address.
Solution Approach 2:
The system applies partial action by requesting only the minimal necessary user input to correct the error. Instead of requiring full re-input, the system asks only for the specific fields that were misrecognized, thereby minimizing the time loss while ensuring reliable correction of the command.
3Reliability
If the system requires multiple re-inputs of speech commands to fix mistakes, then the system can potentially achieve correct recognition, but the complexity of the interaction increases
Solution Approach 1:
The patent simplifies interface complexity by segmenting the correction process into field-specific corrections rather than requiring navigation through multiple general re-input interfaces. The system presents a simplified interface that targets only the problematic address field, reducing the number of interface elements the user must navigate.
Solution Approach 2:
The system applies local quality by providing context-specific correction interfaces for each address field type (state, city, street). This localized approach reduces overall interface complexity by presenting only the relevant correction options for the specific field that needs correction, rather than displaying the entire address form.
Data Source
AI summary
A system, method, and computer program product for allowing a user to fix errors in a user input is described. An original user input providing a destination address or a point of interest is received by an in-vehicle computing system. Upon receiving a user input indicating an error in interpreting the original user input, the user is presented with a new query requesting the user to enter a portion of or a category associated with the original input. Once the updated user input providing a portion of the original input is received, the user input is identified and executed based on the original user input entry and the updated user input.