Sirius is an open-source virtual assistant, a bit like Apple's Siri (pictured above), Google's Google Now, or Microsoft's Cortana. But unlike those well-known helpers--and like Linux itself--Sirius is an open platform anyone can use and contribute to, from universities to startups. It's currently being tested on Ubuntu, and you can download and install it on your own Linux PC today... if you're particularly adventurous.
How it works
Sirius includes speech recognition, image recognition, and text recognition components. Ask it a question, and Sirius will analyze the meaning of the words, then extract the relevant knowledge from Wikipedia.
One big feature Sirius offers is the image recognition integration. For example, you could take a photo of the Eiffel Tower and ask "when was this built?" Sirius would analyze the image, determine it was the Eiffel Tower, and then go find out when the Eiffel Tower was built. The big commercial assistant programs don't yet offer a similar feature. This University of Michigan video provides a good introduction.
Organizations like Google, the Defense Advanced Research Projects Agency (DARPA), and the National Science Foundation (NSF) are funding this project, but don't expect Google to replace Google Now any time soon. Instead, it's being developed by researchers at the University of Michigan's Clarity Lab, and technologies from it may one day make it into the commercial virtual assistants.
But Sirius isn't just a dry research project--it's actual software you can download and run today.
Get the code
This project is licensed under the BSD license and hosted on GitHub, so anyone can contribute and play with the code. You'll also find a "Sirius Suite" download package, which you should use if you want to install this stuff on your own Linux PC.
This isn't a piece of ready-made consumer software, so getting it on your own PC involves compiling the code from source. Sirius is built from many other open-source projects, including Carnegie Mellon University's Sphinx for recognizing speech and UC Berkeley's Caffe deep learning framework software. For image recognition, it uses OpenCV's SURF. To answer all your questions, Sirius uses Carnegie Mellon's OpenEphyra software. All of these bits of software are included in the Sirius Suite package.
Linux desktop environments and distributions could one day integrate this software with a pretty graphical interface, providing their own alternative to Siri, Google Now, and Cortana.
Yes, you could use this stuff on your own Linux PC today and impress your friends. Cortana integration is still in development in Windows 10, and Apple hasn't made any moves to add Siri to Mac OS X. Currently, the best you can do on Windows and Mac OS X--and yes, Linux too--is use Google Now in Google Chrome.
But Sirius isn't just a cool little open-source program to install. It's a long-term vision for "the Linux of virtual assistants," an open project anyone can improve and use for their own needs. That's something worth supporting--and something worth looking forward to.