Wake Word Evaluation for Natural Language Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Natural language control devices face challenges in distinguishing intended natural language commands from background noise or conversations, leading to unintended device actions and user frustration.
Innovation Solution
The implementation of a 'wake word' system that evaluates candidate words based on metrics such as word frequency, false alarm score, ease of pronunciation, and similarity to taboo words to determine their suitability as a trigger for activating natural language command recognition, allowing users to interact with devices without additional input.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a computing device uses natural language input without requiring explicit activation indication, then user interaction convenience is improved, but the device may incorrectly interpret background noises or background speech as commands
Solution Approach 1:
The system performs preliminary wake word detection before processing natural language commands. By first detecting whether the user has activated the device with a wake word, the system can distinguish between intended commands and background speech, resolving the contradiction between ease of operation and reliability.
2Reliability
If a computing device requires explicit non-natural language input to activate natural language commands, then command accuracy is improved, but user interaction complexity increases
Solution Approach 1:
The system extracts and processes only the wake word portion of speech separately from the main natural language command. This allows the device to reliably detect activation intent through simple wake word recognition without requiring complex multi-step activation mechanisms, thus improving reliability while maintaining simplicity.
3Speed
If a computing device continuously monitors for natural language commands, then responsiveness is improved, but the device may respond to unintended background speech
Solution Approach 1:
The system continuously monitors for wake words as a preliminary activation step before processing full natural language commands. This preliminary detection enables the device to remain responsive to intended commands while avoiding responses to background speech by requiring successful wake word detection first.
Data Source
AI summary
Natural language controlled devices may be configured to activate command recognition in response to one or more wake words. Techniques are provided to receive a candidate word for evaluation as a wake word that activates a natural language control functionality of a computing device. The candidate word may include one or more words or sounds. Values for multiple wake word metrics are then determined. The candidate word is evaluated based on the various wake word metrics.