Noname manuscript No. (will be inserted by the editor) Combined I-vector and Extreme Learning Machine Approach for Robust Speaker Identification and Evaluation with SITW 2016, NIST 2008, TIMIT Databases Musab T. S. Al-Kaltakchi · Mohammed A. M. Abdullah · Wai L. Woo · Satnam S. Dlay Received: date / Accepted: date Abstract In this article, a novel combined I-vector and classification approach using an Extreme Learning Machine (ELM) is proposed for speaker identifi- cation. The ELM is chosen because it is fast to train and has the universal approximator property. The system is evaluated with three different databases: the 2016 SITW database, the NIST 2008, and the TIMIT.The proposed system is compared with the Gaussian Mixture Model-Universal Background Model (GMM-UBM) and other states of the art approaches. The outcomes show that the I-vector method outperforms the GMM-UBM approach and other states of the art methods under specific conditions, and that fusion techniques can be used to improve robustness to noise and handset effects. Keywords Text independent speaker identification · Non-stationary Noise · Additive Gaussian Noise · TIMIT database 1 Introduction Several biometrics traits have been proposed utilizing different traits Al-Nima et al. (2017b),Abdullah et al. (2017),Ma et al. (2017), Scheidat et al. (2017) Morales et al. (2017), Al-Nima et al. (2017a) including the speaker biometric Bhukya et al. (2019). An important application in biometrics and forensics M. Kaltakchi Department of Electrical Engineering, College of Engineering, Al-Mustansiriya University, Baghdad, Iraq{Email: musab.tahseen@gmail.com } M. Abdullah Computer and Information Engineering Department, College of Electronics Engineering, Ninevah University, Mosul, Iraq. W. Woo Department of Computer and Information Sciences, Northumberia University, UK S. Dlay School of Electrical and Electronic Engineering Newcastle University, NE1 7RU, UK.