Journal article

ROHDIN Johan A., SILNOVA Anna, DIEZ Sánchez Mireia, PLCHOT Oldřich, MATĚJKA Pavel, BURGET Lukáš and GLEMBEK Ondřej. End-to-end DNN based text-independent speaker recognition for long and short utterances. Computer Speech and Language. Amsterdam: Elsevier Science, 2019, vol. 2020, no. 59, pp. 22-35. ISSN 0885-2308. Available from: https://www.sciencedirect.com/science/article/pii/S0885230818303632
Publication language:english
Original title:End-to-end DNN based text-independent speaker recognition for long and short utterances
Title (cs):Rozpoznávání mluvčího závislé na textu založené na End-to-end DNN přístupu pro dlouhé a krátké promluvy
Pages:22-35
Place:NL
Year:2019
URL:https://www.sciencedirect.com/science/article/pii/S0885230818303632
Journal:Computer Speech and Language, Vol. 2020, No. 59, Amsterdam, NL
ISSN:0885-2308
DOI:10.1016/j.csl.2019.06.002
URL:http://www.fit.vutbr.cz/research/groups/speech/publi/2019/rohdin_elsevier_Journal_Paper_2020_18303632.pdf [PDF]
Keywords
Speaker verification, DNN, End-to-end, Text-independent, i-vector, PLDA
Annotation
Recently several end-to-end speaker verification systems based on deep neural networks (DNNs) have been proposed. These systems have been proven to be competitive for text-dependent tasks as well as for text-independent tasks with short utterances. However, for text-independent tasks with longer utterances, end-to-end systems are still outperformed by standard i-vector + PLDA systems. In this work, we present an end-to-end speaker verification system that is initialized to mimic an i-vector + PLDA baseline. The system is then further trained in an end-to-end manner but regularized so that it does not deviate too far from the initial system. In this way we mitigate overfitting which normally limits the performance of end-to-end systems. The proposed system outperforms the i-vector + PLDA baseline on both long and short duration utterances.
BibTeX:
@ARTICLE{
   author = {A. Johan Rohdin and Anna Silnova and Mireia
	S{\'{a}}nchez Diez and Old{\v{r}}ich Plchot and
	Pavel Mat{\v{e}}jka and Luk{\'{a}}{\v{s}} Burget
	and Ond{\v{r}}ej Glembek},
   title = {End-to-end DNN based text-independent speaker
	recognition for long and short utterances},
   pages = {22--35},
   journal = {Computer Speech and Language},
   volume = 2020,
 number = 59,
   year = 2019,
   ISSN = {0885-2308},
   doi = {10.1016/j.csl.2019.06.002},
   language = {english},
   url = {http://www.fit.vutbr.cz/research/view_pub.php?id=12038}
}

Your IPv4 address: 34.237.75.18
Switch to https