Comment by 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.