Efficient parameterization for Automatic speaker recognition using Support Vector Machines Rania Chakroun 1,4 , Mondher Frikha 1,3 , Leila Beltaïfa Zouari 1,2 1 Advanced Technologies for Medicine and Signals (ATMS) Research Unit 2 National School of Engineering of Sousse, Sousse, Tunisia 3 National School of Electronics and Telecommunications of Sfax, Sfax, Tunisia 4 National School of Engineering of Sfax, Sfax, Tunisia Abstract. Recent advances in the field of speaker recognition have proved to highly outperform algorithms. However this performance degrades when limited data are presented. This paper presents examples on how SVM can improve speaker recognition. The main contribution in this approach is the use of new low-dimensional vectors when training data are limited. We show how different kernels function of Support Vector M achines (SVM ) can be used to deal a new approach for speaker recognition system. We achieved remarkable results using TIMIT database Keywords: Support Vector M achines, limited data, speaker recognition, speaker identification. 1 Introduction Speaker recognition is a biometric modality [1] that depends on speech information to determine a person identity [2]. It is widely employed in various kind of applications such as banking over a telephone network, voice mail, voice dialing, database access services, security control for confidential information, remote access to computers, and information and reservation services. The field of speaker recognition can be divided into two main applications: speaker identification and speaker verification. The task of Speaker identification consists on determining the identity of an unknown speaker based on his/her speech utterances. However, the task of speaker verification concerns the use of the voice of an unknown speaker to verify the identity claimed by the speaker. The process of speaker recognition operates on two different modes. In fact, recognition of speakers can be text-dependent or text-independent. For text-dependent applications, the system supposes that the presented speaker say exactly an utterance which is determined by the system. However, in independent mode of the text, the speaker is free to pronounce any sentence and there is no constraint on the speech content of the speaker.