Iris Voice AutomationAn voice automated, interactive assistant for information retrieval
|
Back to Independent Projects |
Outline |
|
Concept |
The Idea Behind Iris ...
Named after the the Greek messenger of the gods, Iris is a way to connect users to the information with as little work in between as possible. There are several quick questions that might come up in every day situations.
What is the weather like in Boston? What is the Vienna Philharmonic? Who is Landon Donovan? I often found myself jumping onto a computer or pulling out a mobile device to get the answer. Iris will listen to what you say, use Google's Voice Recognition servers to generate text from speech, then parse your text and find a short answer from freely available sources like Wikipedia. Iris currently supports 3 categories of questions. |
How it Works |
Iris works in three ways listening, searching and talking. The listening and talking parts of Iris are handled in the cloud by Google Voice Recognition and Speech Synthesis tools, Iris utilizes these to interface with the user.
Iris was design for users to ask questions natural, in grammatically correct English. Iris utilizes elements of natural language processing to understand what exactly what is being asked. Iris is engineered to get you your answer as fast as possible, and Iris responds in approximately 5 seconds. To achieve such speeds, Iris has been It is build from the ground up on a multithreaded framework, allowing it minimize the time each task above takes. In the approximately 500 milliseconds Iris takes to answer questions, it may poll up to three search engines with multiple variations of your query and return the most relevant one. |
Watch it in Action |
|
|
* More to Come: This project is still ongoing and not yet completed, but you can follow it on GitHub.
|
Back to Independent Projects |
