“…Hence, a speech signal from the given speaker x k (t) with a duration T k = (2, ..., 3)R × J k, r × τ = (1.2, ..., 1.8)R can be regarded as suffi cient for the formation of the PDB X k . In particular, in the case of application of six vowel phonemes of the Russian language, we get T k = 8-10 s, which is in good agreement with the possibilities of contemporary methods aimed at the automatic processing of speech [3,15].…”