Comment by momodomo

Sounds like a good way to solve a corner case in speech recognition: speech to text for names. In reverse, it could also improve speech synthesis for personal robots that would speak the names of their users.