I had this crazy idea…
a user calls and is put somewhere so that his audio stream can be used.
A second user, who is supposed to answer, rings the extension and is put somewhere else (in the same way?), but in different places, so that they cannot hear each other directly.
The audio of each user (which the other does not hear directly) is directed to a local ASR service, then translated and then synthesized (or recorded), after which it is sent to the other, who then hears what the first user said, but in his own language.
How could this be done?
- Custom C Module
- The same as many other things, ARI + External Media
Also, to work reliably you will need a translation service that can accept synchronised feeds from both parties. I don’t know if current, commercially available, machine based systems can do that, but without it, and especially if there are a lot of closed questions, you could get a lot of embarrassing mistranslations, because of homophones.