US20240161740
2024-05-16
Physics
G10L15/22
The patent application describes a method for managing multiple digital assistants on a single device using a system of "warm words." Each digital assistant has its own set of warm words, which are specific phrases used to trigger actions without full speech recognition. The method involves receiving these active sets of warm words for each assistant and executing a warm word arbitration routine to determine a final set of warm words that the device will listen for. When a warm word from this final set is detected, the associated digital assistant is instructed to perform the specified action.
In a speech-enabled environment, users can interact with digital assistants by speaking commands aloud, enabling hands-free operation across various settings like homes or workplaces. The presence of multiple digital assistants on a device can lead to overlapping warm words. The arbitration process helps resolve which assistant should respond when such overlap occurs, ensuring that the right action is taken by the appropriate assistant.
Key features of this method include the ability to dynamically adjust the set of active warm words based on changes in the environment or user interactions. This includes adding or removing digital assistants or modifying the set of warm words as needed. The system can also prioritize shared warm words that appear in multiple assistants' active sets, ensuring efficient operation and minimizing false detections.
The arbitration routine considers various constraints like memory and computing resources, as well as acceptable error rates, to optimize performance. It can also track the frequency and recency of warm word usage to refine the final set of detectable warm words. This adaptive approach allows for responsive and efficient management of multi-assistant interactions on a single device.
Additionally, the system supports simultaneous execution of multiple digital assistants through voice commands or configuration requests. This flexibility allows users to customize their interaction with the device, enabling seamless integration and cooperation among different digital assistants based on user preferences and requirements.