Science & Engineering Research Support soCiety (SERSC)
The performance of speaker verification system degrades when the test segments are utterances of short durations, therefore, the authors investigate the use of model representing their target speaker with their close speaker and the users own speech data. They propose to create a new speaker model who groups Close Speakers (CSs) achieved with two clustering algorithms in Automatic Speaker Verification (ASV). Intra and inter-speaker's variability are two clustering algorithm used in voice module. They compare the traditional approach which uses one specific customer model (maximum a posteriori adaptation) with the close speaker model (customer's families).