Speech Recognition Engine N-Best Cross-Referencing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated conversational UI speech recognition systems face challenges in accurately recognizing user inputs, leading to high false acceptance rates and user frustration, especially when multiple trials are required, and often degrade user experience due to repeated requests for redundant information.
Innovation Solution
The method involves prompting users with a primary query, repeating if necessary, and only requesting redundant information when a predefined limit of failures is reached, cross-referencing n-best results from both queries to improve recognition accuracy and transferring the user to a human agent if errors persist.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system requests redundant information multiple times to improve recognition accuracy, then recognition reliability improves, but user satisfaction deteriorates due to repeated requests
Solution Approach 1:
The system performs preliminary recognition attempts with the primary query before requesting redundant information. By establishing a failure threshold and only requesting redundant information after predefined limits are reached, the system prepares recognition strategies in advance rather than immediately requesting all available information, thereby improving user satisfaction while maintaining reliability
Solution Approach 2:
The recognition process is segmented into distinct phases: primary query attempts (with retry logic), fallback to redundant information requests, and final human agent escalation. This segmentation allows the system to request redundant information strategically only when necessary, rather than continuously, thus balancing reliability improvement with user satisfaction preservation
2Measurement precision
If the system uses n-best results from multiple queries to improve accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The system uses n-best results (a partial subset of all possible recognition outcomes) rather than processing every possible hypothesis. By limiting the cross-referencing to the top n results from each query and applying simple consensus logic, the system achieves improved precision without implementing complex exhaustive analysis algorithms
Solution Approach 2:
The n-best results from primary and secondary queries serve as intermediaries that bridge the gap between uncertain speech recognition and confident decision-making. By cross-referencing these intermediate results and requiring consensus, the system improves measurement precision through a relatively simple mediator mechanism rather than complex probabilistic modeling
Data Source
AI summary
A method for speech recognition includes: prompting a user with a first query to input speech into a speech recognition engine; determining if the inputted speech is correctly recognized; wherein in the event the inputted speech is correctly recognized proceeding to a new task; wherein in the event the inputted speech is not correctly recognized, prompting the user repeatedly with the first query to input speech into the speech recognition engine, and determining if the inputted speech is correctly recognized until a predefined limit on repetitions has been met; wherein in the event the predefined limit has been met without correctly recognizing the inputted user speech, prompting speech input from the user with a secondary query for redundant information; and cross-referencing the user's n-best result from the first query with the n-best result from the second query to obtain a top hypothesis.