In addition to tracking the calls & durations, it should track the pitch and tone to provide more accurate results. E.g. did a user speak in a melodious voice o

In addition to tracking the calls & durations, it should track the pitch and tone to provide more accurate results. E.g. did a user speak in a melodious voice or had to raise the tone, while speaking with another? Coupled with a statistical model of the specific user's tendency about the tone, will give very precise information. A chilling scenario can be: serve ads to the user for improving pronunciation based on an analysis of their speech. Not sure about the privacy implications, though. Technically, it may or may not be possible to achieve at this stage, since powerful hardware is required for such analysis.