The 2013 speaker recognition evaluation in mobile environment

• Main Authors: Khoury, E., Vesnicer, B., Franco-Pedroso, J., Violato, R., Boulkcnafet, Z., Mazaira Fernandez, L. M., Diez, M., Kosmala, J., Khemiri, H., Cipr, T., Saeidi, R., Gunther, M., Zganec-Gros, J., Candil, R. Zazo, Simoes, F., Bengherabi, M., Alvarez Marquina, A., Penagarikano, M., Abad, A., Boulayemen, M., Schwarz, P., Van Leeuwen, D., Gonzalez-Dominguez, J., Neto, M. Uliani, Boutellaa, E., Gomez Vilda, P., Varona, A., Petrovska-Delacretaz, D., Matejka, P., Gonzalez-Rodriguez, J., Pereira, T., Harizi, F., Rodriguez-Fuentes, L. J., El Shafey, L., Angeloni, M., Bordel, G., Chollet, G., Marcel, S.
• Format: Conference Proceeding
• Language: English
• Subjects: Calibration; Feature extraction; Logistics; Mel frequency cepstral coefficient; NIST; Training
• ISSN: 2376-4201
• DOI: 10.1109/ICB.2013.6613025

This paper evaluates the performance of the twelve primary systems submitted to the evaluation on speaker verification in the context of a mobile environment using the MOBIO database. The mobile environment provides a challenging and realistic test-bed for current state-of-the-art speaker verification techniques. Results in terms of equal error rate (EER), half total error rate (HTER) and detection error trade-off (DET) confirm that the best performing systems are based on total variability modeling, and are the fusion of several sub-systems. Nevertheless, the good old UBM-GMM based systems are still competitive. The results also show that the use of additional data for training as well as gender-dependent features can be helpful.