Amazon has patented a system that monitors user behavior, using a smartphone, to determine patterns of communication behavior in various situations. For example, patterns of communication include user behavior when replying to coworkers or user behavior when replying to spouse.
After detecting multiple occurrences of a type of communication in a particular situation, the system defines a pattern of behavior in that type of situation. Thereafter, when the smartphone is subsequently in that type of situation again, the smartphone prompts a user to perform an action corresponding to the pattern of behavior. For example, if a user makes a call each weekday when within a given distance of a geographical location, such as within a mile from a workplace, the device can prompt the user whether the device should initiate that call or text the next time the computing device is in that location.
While detecting patterns, the system also generates a set of speech models each type of recipient. Each speech model can include language that is consistent with patterns of speech typically used between a user and a type of recipient. For example, a user might use different words and phrasing when communicating with spouse than when communicating with a co-worker.
This ability to automatically generate communications according to various patterns of speech, for various recipients provides various advantages over conventional communication approaches. For example, when generating an email to a contact labeled as a business contact, the speech engine uses a business contact speech profile built partly by mining data from past business communications. The language contained in the business speech profile would likely be more conservative and professional as compared to the language contained in the friend speech profile.
In a short while, the smartphone will learn all communication patterns of a user. Using text to speech technology, the system can also “verbally” reply to many phone calls. Effectively, the system automates a large part of human conversation.
Publication number: US 9363104
Patent Title: Customized speech generation
Publication date: Jun 7, 2016
Filing date: Apr 21, 2014
Inventors: Isaac Scott Noble; Gabrielle M. Halberg; Kenneth Mark Karakotsios; Yuzo Watanabe;
Original Assignee: Amazon Technologies, Inc.