No. Speakers are labeled generically (e.g., Speaker 1, Speaker 2) unless manually renamed.
Overlapping speech is partially supported, but accuracy may be reduced depending on audio quality.
No. All speaker analysis is performed locally.