A Lightweight Speaker Recognition System Using Timbre Properties