Vehicle ASR Context Classification via Speech Duration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatic speech recognition (ASR) systems in vehicles face inefficiencies in recognizing point-of-interest (POI) and address requests without an initiating word, leading to increased processing time or failure in recognition.
Innovation Solution
The system measures the duration of received speech to classify it as either a POI or address request by establishing time windows for each, and performs a keyword search when ambiguity occurs, allowing it to load specific grammars for processing without relying on a gateway command.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the ASR system uses a wide variety of grammars to recognize different topics, then the recognition coverage is improved, but the processing time increases significantly
Solution Approach 1:
The patent segments the large grammar into multiple context-specific smaller grammars (POI grammar, address grammar, general grammar). Instead of loading one large comprehensive grammar, the system divides it into specialized modules that can be selectively activated based on the speech context, thereby reducing the overall processing burden while maintaining comprehensive recognition capability.
Solution Approach 2:
The system performs preliminary context classification by analyzing speech duration and characteristics before selecting which grammar to apply. This preliminary action of identifying speech context (POI vs. address vs. general) allows the system to pre-select the appropriate grammar, avoiding the need to process through a large comprehensive grammar structure.
2Measurement precision
If the ASR system requires an initiating word (gateway command) to identify context, then the context identification accuracy is improved, but the ease of operation deteriorates
Solution Approach 1:
The patent extracts and removes the requirement for explicit gateway commands (POI, address, etc.) from the interaction flow. Instead of requiring users to say initiating words, the system infers context automatically from inherent speech characteristics like duration, allowing users to speak naturally without memorizing or recalling specific command structures.
Solution Approach 2:
The system performs self-service context identification by automatically analyzing speech duration and characteristics to determine whether the user is providing a POI, address, or general information. This eliminates the need for users to manually indicate context through gateway commands, making the system more intuitive and easier to use.
3Productivity
If the ASR system uses context-based grammars, then the processing efficiency is improved, but the device complexity increases
Solution Approach 1:
The patent implements a dynamic grammar selection mechanism that adapts the grammar choice based on real-time speech characteristics. The system dynamically switches between different grammar types (POI, address, general) based on speech duration analysis, rather than using a static single grammar or requiring complex manual context configuration. This dynamic adaptation achieves high processing efficiency with manageable system complexity.
Data Source
AI summary
A system and method of performing automatic speech recognition (ASR) includes: receiving speech at a vehicle microphone; communicating the received speech to an ASR system; measuring an amount of time that elapses while speech is received; selecting a point-of-interest (POI) context or an address context based on the measured amount of received time; and processing the received speech using a POI context-based grammar when a POI context is selected or an address-based grammar when an address context is selected.