The US Navy is not only developing robotic warships, but it also wants to teach them how to speak. In a request for proposals, the Navy has outlined how it wants its autonomous surface vessels to be able to have back-and-forth verbal radio communications with the crew of other ships to avoid collisions.
If Unmanned Surface Vessels (USVs) are ever going to become practical, they will have to operate in the real world in a manner comparable to that of conventional ships. This means that they not only need to be able to go from point A to Point B without running aground or hitting anything, they also need to be able to deal with other sea traffic and the crews that control them.
This means that a USV must be able to follow the International Regulations for Preventing Collisions at Sea (COLREGS), which lay out how two vessels should respond if they meet one another. Unfortunately, COLREGS are only clear when it comes to two vessels. If there are three or more, things get very complicated, and the standard procedure is for the bridge crews involved to contact one another by VHF radio and work out how to avoid trouble.
While the current state of the art for US Navy autonomous warships provides them with the computers and software they need to follow COLREGS and to avoid collisions, there's a bottleneck in that the robotic ship can't talk, and mariners rely very heavily on VHF radio to communicate.
To clear this bottleneck, the Navy wants to develop a system that will (in near real-time) receive secure radio bridge-to-bridge communications, convert speech to text, convert that text into something that the autonomous vessel can understand, form a solution to the navigation problem, and then reply in natural speech.
In its request, the Navy says that it wants to follow a three-phase development process. In the first phase, the goal will be to demonstrate the feasibility of the concept – especially the capacity to understand simple, common call ups like "Sea Hunter, this is Sun Princess; propose a port-to-port passage" when spoken by native English speakers.
After the initial design is completed, the project will go on to Phase II. In this, a prototype of the technology will be incorporated into a VHF radio for testing and evaluation. In addition, the system will be expanded to include understanding and replying to non-native English speakers. This will be followed by three months of testing and offshore trials.
In Phase II, the project will concentrate on helping the Navy to adopt the new technology as it evolves into a final end-to-end system that can communicate like a human. Beyond its military applications, the Navy sees the talking ship technology to have civilian uses, including unmanned commercial vessels that need to operate when satellite links are unavailable, ships with small crews, and even pleasure craft.
Source: Department of Defense
They need to ditch the idea of experimenting with amateurs and half-baked ideas, and just get in touch with these guys: https://www.youtube.com/watch?v=yDI5oVn0RgM