Input Processing with Profile Context

Raju Ansari, Alok Upadhyay, Zhengyang Wu, Bharath Kumar, John Howard

Abstract: A speech-processing system may be configured to provide certain endpoints (e.g., skills and/or applications) with data reflecting an explicitly selected user profile. When handling requests from a device configured to support a session context (e.g., a personal device), the skill may handle multiple requests in an interaction (e.g., a “session”) using the selected profile as the session context, even if the system recognizes a different user for one or more of the requests. Thus, if the system recognizes more than one user using a personal device during an interaction with the skill (e.g., playing a game or requesting songs), the skill may handle each request of the interaction according to the selected profile. For a device not configured to support a session person, however, (e.g., a communal device) the skill may handle each request according to the recognized profile regardless of an explicitly selected profile.