Navigation Device Offline Speech Recognition via Pre-loaded Data
Solution Overview
Problem
Existing navigation devices cannot perform speech recognition when outside a communication area, as they rely on connectivity with a server for processing.
Innovation Solution
A navigation device that pre-loads terminal-use speech recognition data before entering a communication impossible area, allowing it to perform speech recognition locally when disconnected from the server, by using a speech input unit, transmitting unit, receiving unit, and obtaining unit to handle speech input and recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the navigation device relies on server connection for speech recognition, then speech recognition accuracy is improved, but speech recognition becomes impossible in communication impossible areas
Solution Approach 1:
The navigation device performs preliminary actions by obtaining and storing speech recognition data from the server before entering communication impossible areas. The determination unit checks whether the guide route includes communication impossible areas, and if so, the obtaining unit retrieves necessary speech recognition data in advance. This preliminary preparation enables the device to perform speech recognition locally without server connection during the communication impossible area, resolving the contradiction between accuracy (server-based recognition) and availability (offline recognition).
2Reliability
If the navigation device obtains speech recognition data before entering communication impossible area, then speech recognition availability is improved, but device complexity increases
Solution Approach 1:
The navigation device segments the speech recognition system into two distinct parts: server-based speech recognition for communication areas and local terminal-use speech recognition data for communication impossible areas. The determination unit segments the route into communication possible and impossible areas, and the obtaining unit selectively obtains data only for the necessary segments (communication impossible areas). This segmentation reduces the overall data burden and system complexity while maintaining availability.
Solution Approach 2:
Instead of obtaining complete speech recognition data for all areas, the system performs partial action by obtaining only the necessary terminal-use speech recognition data for communication impossible areas. The determination unit identifies specific segments where offline recognition is needed, and the obtaining unit retrieves only that portion of data, avoiding the complexity of full offline capability while ensuring availability where truly needed.
Data Source
AI summary
When a navigation device receives an input of speech from a user, it transmits data of inputted speech to a server and receives a speech recognition result. If a guide route includes a communication impossible area, the navigation device obtains terminal-use speech recognition data necessary for the navigation device to perform speech recognition from the server before reaching the communication impossible area. Thus, even if the speech is inputted by the user while traveling the communication impossible area, the navigation device can perform the speech recognition by itself by using the terminal-use speech recognition data.