Intelligent Assistant Gaze Detection Wake Word Elimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Intelligent personal assistants require users to repeatedly utter wake words or phrases for each interaction, which becomes annoying and inefficient, especially for consecutive questions or commands.
Innovation Solution
Incorporating cameras and audio transducers into intelligent personal assistants to determine when a user is addressing the device, allowing for the elimination or reduction of the need to speak wake words by continuously monitoring visual and audio cues, such as gaze direction and voice input, to initiate interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If wake words are required for each interaction, then the assistant can reliably distinguish intended commands from background speech, but user interaction becomes tedious and inefficient
Solution Approach 1:
The system performs preliminary actions by continuously monitoring audio and visual cues before a command is given. The camera captures the user's gaze direction and the system pre-processes this visual information to detect when the user is looking at the device, preparing the system to recognize commands without requiring a wake word.
Solution Approach 2:
The patent introduces visual cues from the camera as an intermediary to bridge the gap between background speech monitoring and command execution. The gaze detection system acts as a mediator that distinguishes intended commands from background speech without requiring the user to speak a wake word, thereby maintaining reliability while improving ease of operation.
2Ease of operation
If the system continuously monitors for commands without wake words, then interaction becomes more natural and efficient, but the system may misinterpret background speech as commands
Solution Approach 1:
The patent merges audio monitoring with visual gaze detection to create a combined recognition system. By combining these two modalities, the system achieves more accurate command recognition than either modality alone, as the visual component provides additional context to distinguish intended commands from background speech.
Solution Approach 2:
The system uses feedback from the camera to continuously monitor and adjust its command recognition. When the camera detects the user is looking at the device, the system becomes more sensitive to audio input, and when gaze is not detected, the system reduces its sensitivity, thereby maintaining high accuracy while enabling natural interaction.
3Ease of operation
If cameras and visual monitoring are added to detect user gaze, then wake words can be eliminated, but device complexity increases
Solution Approach 1:
The camera serves multiple functions: it detects user gaze direction to enable wake-word-free interaction, and can also be used for other purposes such as identifying users, monitoring the environment, or providing visual feedback to the user. This multi-functionality justifies the added complexity by providing additional value beyond just eliminating wake words.
Data Source
AI summary
Aspects of the present invention are directed towards a more natural way to interact with an intelligent personal assistant. An intelligent personal assistant comprises a camera that provides images of an area surrounding the assistant. The assistant monitors images provided by the camera to determine when a user is addressing the assistant. When voice input is received after determining that a user is addressing the assistant, the assistant understands that the voice input is intended for the assistant, and acts on the voice input to respond to the user.