Continuous passive assistance via audio-visual analysis
Publication Date: 2014-Jun-27
The IP.com Prior Art Database
Disclosed is a software application for smart phones or mobile devices that uses speech and physical behavioral recognition to learn the contexts in which a user may require certain assistance, and then provide assistance as needed.
Page 01 of 2
Continuous passive assistance via audio -visual analysis
Smart phones and mobile devices are equipped with sophisticated software that responds to voice commands and other audible cues. The devices are similarly equipped with advanced optics/camera systems that can detect and record videos of various sorts. However, the devices lack a system or method by which to utilize these advanced features in a manner that is helpful, non-intrusive, and does not need prompting by the user.
An intelligent and adaptive mechanism is needed whereby smart phones and mobile devices can render passive assistance to the user.
The novel contribution teaches the use of intelligent software that runs as a background service. This software uses speech and physical behavioral recognition to learn the contexts in which a user may require certain assistance, and then provide assistance as needed. The software application is installed and runs on a smart device and not only has access to audio and video inputs, but also interacts with those streams. When the software recognizes certain phrases, audio cues, video cues, etc. it renders the appropriate assistance.
The device is initially seeded with a list of phrases to recognize (i.e. for which to listen"). As a stream of text is transcribed, for example, the software searches to identify a number of text strings within it (e.g., "my number is"", let me give you my number"", let me send you"", what is your email address", etc.) Each such phrase is associated with an action. For example, "my number is" may be associated with the action of recording the phone number that follows. If the speaker is known, then a contact entry can be created or updated. If the speaker is not identified, then the number can be saved in a generic contact and applica...