Journal
Scientific and technical journal of information technologies, mechanics and optics
UDK621.391.037.372
Issue:6 (82)
The paper proposes to use the popular text independent speaker identification method based on i-vectors extraction for the solution of the speaker’s age group identification. Two realizations of speaker’s age group detector based on using i-vectors are researched: an approach based on a support vector machine classifier and another one based on a multiple linear regression. Both methods provide a high accuracy rate of speaker’s age group detection by spontaneous speech recordings. The average rate of correct speaker’s age group identification was 61% and 65% respectively for the NIST SRE 2008 speech database.