[davitb]

The hard part is to separate background voices (e.g. TV, chatter, etc) from the primary speaker's voice. Basically do voice isolation. Voice fingerprinting would help only in this context.