Voiceprint Classification
L. G. Kersta · The Journal of the Acoustical Society of America · 1965
A binary-code automated classification system has been devised for voiceprints. The 10-digit code that enables 210 possible different codes per word resulted in unique speaker selection of better than 96% for a population of 12 speakers. When the classification is based serially on 5 cue words, the selection accuracy was better than 99.5%. The classification system is based upon the uniqueness of a selected amplitude-quantized frequency vs time matrix.