Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Speaker Identification Using Discriminative Features Selection

Bogdan Sabac, Inge Gavat, Zica Valsan

Polytechnic University of Bucharest, Bucharest, Romania

A new method of text-dependent speaker identification using discriminative feature selection is proposed in this paper. The characteristics of the proposed method are as follows: feature parameters extraction, vector quantization with the growing neural gas (GNG) algorithm, model building using gaussian distributions and discriminative feature selection (DFS) according to the uniqueness of personal features. The speaker identification algorithm is evaluated on a database that includes 25 speakers each of them recorded in 12 different sessions. All speakers spoke the same phrase for 10 times in each recording session. The test results showed that both FRR (False Rejection Rate) and FAR (False Acceptance Rate) were about 1[%].


Full Paper

Bibliographic reference.  Sabac, Bogdan / Gavat, Inge / Valsan, Zica (2000): "Speaker identification using discriminative features selection", In ICSLP-2000, vol.3, 1025-1028.