Multi-User Warm Word Arbitration for False Positive Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In shared environments where multiple users interact with assistant-enabled devices, existing systems struggle to manage competing voice commands efficiently due to limited computational resources and varying user preferences for warm words, leading to increased false positives and inefficiencies.
Innovation Solution
A method and system that detect the presence of multiple users and execute a warm word arbitration routine to select a final set of warm words for detection, considering user preferences, computational constraints, and ambient context, enabling efficient and accurate command recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all users' warm words are enabled for detection in a multi-user environment, then each user can access their preferred commands, but computational resources are overwhelmed and false positives increase
Solution Approach 1:
The patent segments the set of all warm words into user-specific subsets. Each user has their own collection of preferred warm words, and the system manages these segments separately. The arbitration routine selects which user segment to activate based on detected presence, rather than managing all warm words uniformly. This segmentation allows the system to accommodate multiple users' preferences while limiting resource usage at any given moment to only the active user's subset.
Solution Approach 2:
The patent implements dynamic warm word arbitration where the active set of warm words changes based on real-time user presence detection. The system continuously monitors which users are present and dynamically adjusts which warm words are enabled for detection. This dynamic adaptation allows the system to respond to changing environmental conditions (user presence) and optimize resource allocation accordingly, switching between different user preference sets as users enter or leave the environment.
2Reliability
If multiple warm words are enabled simultaneously for different users, then command recognition coverage is improved, but false positive detection rate increases
Solution Approach 1:
The patent applies local quality by making warm word detection sensitivity user-specific and context-dependent. Instead of uniformly enabling all warm words across all contexts, the system enables only the relevant subset of warm words corresponding to the currently detected user. This localized approach ensures high recognition accuracy for the active user's commands while preventing false positives from other users' warm words that are not currently enabled. Each user's warm words are treated as a distinct local set with its own activation conditions.
3Measurement precision
If the system continuously monitors user presence to manage warm words, then multi-user command accuracy is improved, but processing time and energy consumption increase
Solution Approach 1:
The patent implements preliminary action by pre-configuring user profiles with their respective preferred warm words before runtime. During operation, the system only needs to detect which user is present and then activate the corresponding pre-prepared set of warm words, rather than dynamically creating or analyzing warm word sets in real-time. This preliminary preparation significantly reduces processing time during actual command recognition, as the system simply switches between pre-defined configurations based on user presence detection.
Data Source
AI summary
A method includes detecting a presence of multiple users within an environment of an assistant-enabled device (AED) and obtaining, for each user, a respective active set of warm words that each specify a respective action for a digital assistant to perform. Based on each respective active set of warm words, the method also includes executing a warm word arbitration routine to enable a final set of warm words for detection by the AED. Here, the final set of warm words includes warm words selected from the respective active set of warm words. While the final set of warm words are enabled, the method also includes receiving audio data corresponding to an utterance captured by the AED, detecting a warm word from the final set of warm words in the audio data, and instructing the digital assistant to perform the respective action specified by the detected warm word.