Dynamic Speech Grammar Compilation for Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing interactive voice systems face high costs and inefficiencies due to the need for frequent updates of recognition grammars in databases that are constantly changing, such as employee names, which requires recompiling the entire grammar for each addition or change.
Innovation Solution
A method that compares received speech to a precompiled grammar and dynamically compiles a new grammar only with updated data, reducing the need for frequent full grammar recompilation by using a secondary grammar built on the fly from a table or database of updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire recognition grammar is recompiled for each database update, then the speech recognition accuracy is maintained, but the time and cost of updates increase significantly
Solution Approach 1:
The patent divides the grammar compilation process into two segments: a precompiled grammar containing the majority of database entries, and a dynamically compiled grammar containing only the updated entries. This segmentation allows the system to maintain high recognition accuracy while significantly reducing update time, as only the small dynamic portion needs to be regenerated for each update.
Solution Approach 2:
The patent performs preliminary compilation of the grammar in advance, creating a precompiled grammar that contains the bulk of the database content. This preliminary action allows the system to be ready for operation before all data is available, and only requires adding new entries later without full recompilation.
2Measurement precision
If the recognition grammar is precompiled based on a database, then speech recognition accuracy improves, but the system cannot adapt to frequently changing data
Solution Approach 1:
The patent introduces a dynamic grammar component that can be automatically updated whenever database changes occur. This dynamic portion works in conjunction with the static precompiled grammar, allowing the system to adapt to changing data in real-time while maintaining the benefits of precompiled recognition accuracy for the stable portion of the database.
Solution Approach 2:
The patent implements a feedback mechanism where the system monitors for database updates and automatically triggers grammar recompilation only for the affected entries. This feedback loop ensures the grammar remains synchronized with the database without requiring manual intervention or full system recompilation.
3Reliability
If highly trained technicians are used to build dialog systems, then system reliability improves, but the cost and complexity of maintenance increase
Solution Approach 1:
The patent implements automatic grammar compilation functionality that allows the system to update itself without requiring highly trained technicians for each database change. The automated system monitors database updates, identifies affected entries, and recompiles the grammar dynamically, reducing both the skill level required for maintenance and the overall complexity of system management.
Data Source
AI summary
Disclosed herein are methods and systems for recognizing speech. A method embodiment comprises comparing received speech with a precompiled grammar based on a database and if the received speech matches data in the precompiled grammar then returning a result based on the matched data. If the received speech does not match data in the precompiled grammar, then dynamically compiling a new grammar based only on new data added to the database after the compiling of the precompiled grammar. The database may comprise a directory of names.