Estimating the Completeness of Discrete Speech Units