Voice Recognition Recommended Command Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional speech recognition systems require users to repeatedly utter a wake-up word for each command, leading to user inconvenience and increased processing costs when performing multiple actions within the same domain.
Innovation Solution
An electronic device that determines a recommended command by analyzing user voice inputs and performs speech recognition without a wake-up word using concurrent wake-up technology, matching the recommended command to a set wake-up word and storing it for subsequent use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional speech recognition requires wake-up word for each command, then speech recognition accuracy is maintained, but user convenience deteriorates and processing costs increase
Solution Approach 1:
The system performs preliminary analysis of user voice input patterns to predict upcoming commands within the same domain, preparing recommended commands in advance. This allows the system to recognize subsequent commands without requiring repeated wake-up words, as the prediction mechanism is already activated from the initial wake-up word recognition.
Solution Approach 2:
The system implements a feedback loop where each recognized command within the same domain reinforces the prediction model for subsequent commands. The system continuously learns from the sequence of commands and adjusts its prediction accuracy, creating a self-improving mechanism that reduces the need for repeated wake-up words while maintaining recognition accuracy.
2Reliability
If wake-up word is required for each command, then system reliability is maintained, but processing costs and resource usage increase
Solution Approach 1:
Instead of requiring full wake-up word processing for every command, the system applies partial action by using lightweight prediction mechanisms for commands within the same domain. The full speech recognition processing is only activated when the prediction confidence is low or when domain switching occurs, thus reducing overall processing resource consumption while maintaining reliability.
Solution Approach 2:
The wake-up word recognition mechanism serves multiple functions: it activates speech recognition mode, initiates domain identification, and starts the prediction mechanism. This multi-functionality reduces the need for separate processing steps, thereby lowering overall resource consumption while maintaining system reliability.
3Measurement precision
If wake-up word must be uttered for each command, then command accuracy is ensured, but operation complexity increases
Solution Approach 1:
The system segments command recognition into two stages: prediction stage for commands within the same domain (where wake-up word is not needed) and full recognition stage for domain-switching commands (where wake-up word is required). This segmentation maintains accuracy for predicted commands while reducing operational complexity by eliminating redundant wake-up word requirements.
Solution Approach 2:
The system dynamically adjusts its operation mode based on the predicted command domain. When the predicted domain matches the current domain, the system operates in a simplified mode without requiring wake-up words. When domain switching is detected, it transitions to full recognition mode. This dynamic adjustment reduces operational complexity while preserving accuracy.
Data Source
AI summary
Disclosed is an electronic device for performing voice recognition by using a recommended command. The electronic device according to various embodiments may include: a processor; an input module for receiving a voice input from a user; and a memory electrically connected to the processor and storing an instruction executable by the processor, a client module, and a recommended command, wherein the processor: determines whether domains of a plurality of plans, which are consecutively generated according to a command included in the voice input, are identical to each other by a configured number or more; when the domains are identical to each other, determines the recommended command on the basis of the domains and stores the recommended command in the memory; and executes the client module on the basis of the plans generated according to the recommended command.