Originally Posted by GrendelQ
You can also take a look at one very practical means of assessing what's heard: how she sounds. Output is going to reflect what she hears. So if she's robotic, you can assume that's what she's hearing. If she sounds melodic ... that's how she hears. She's extremely melodic.
One can be melodic with access to a limited range of frequencies. For that matter, even someone profoundly deaf can develop a melodic speaking voice with enough therapy and training. Your anecdote doesn't contradict anything I have said.