Efficient text-independent speaker recognition with short utterances in both clean and uncontrolled environments

• Published in: Multimedia tools and applications Vol. 79; no. 29-30; pp. 21279 - 21298
• Main Authors: Chakroun, Rania, Frikha, Mondher
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2020; Springer Nature B.V
• Subjects: Background noise; Biometrics; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Identification systems; Multimedia; Multimedia Information Systems; Noise; Signal processing; Speaking; Special Purpose and Application-Based Systems; Speech; Speech recognition; Voice recognition; Speaker identification; Speaker recognition; PLDA; Short utterances; Noise; I-vector
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-020-08824-7

Automatic speaker recognition has emerged as an important technology for voice-based biometric systems. However, text-independent speaker recognition against short utterances remains a challenging task despite of recent advances in the domain of speaker recognition. The presence of background noise presents another critical issue in this field. In this paper, we propose effective features for speaker identification with short utterances, which perform well in both clean and noisy conditions. Speaker identification performance for utterances having very short training and testing durations are presented which provide a clearer description of the proposed system performance. Te proposed features have shown strong robustness in these challenging situations and they consistently perform better than the well known MFCC and GFCC features. The efficiency of the proposed approach was thoroughly tested by comparisons with the most recently successful SVM and i-vector PLDA baseline speaker recognition systems.