What are currently the state-of-the art algorithms for this purpose? This is an area that I know absolutely nothing about. I'm basically interested in the following two things (which I assume go hand-in-hand):
Source separation
Locating sources
I'm basically trying to build some simple Raspberry Pi based robot that would turn towards a voice and needs to be able to work when background noise is present from multiple sources. Basically I would like to separate different signal sources on the Raspberry Pi, compute their direction, and then do further fingerprinting of the signals to find the one that I want to turn towards (this can be offloaded to a more powerful system over wifi).
Even better would be if someone knows any open source research software for this.
