DNN-based SRE Systems in Multi-Language Conditions

Systémy pro rozpoznávání mluvčího založené na DNN v multilingválních podmínkách

report

English

Novotný Ondřej, Ing., Ph.D.
Matějka Pavel, Ing., Ph.D. (DCGM)
Glembek Ondřej, Ing., Ph.D.
Plchot Oldřich, Ing., Ph.D. (DCGM)
Grézl František, Ing., Ph.D. (DCGM)
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Černocký Jan, prof. Dr. Ing. (DCGM)

• http://www.fit.vutbr.cz/research/groups/speech/publi/2016/dnn-based-sre_TECH_REP_v0.pdf

speaker recognition, multilinguality, DNN

This work studies the usage of the (currently state-of-the-art) Deep Neural Networks (DNN) i-vector/PLDA-based speaker recognition systems in multi-language (especially non-English) conditions. On the ``Language Pack'' of the PRISM set, we evaluate the systems' performance using NIST's standard metrics. We study the use of multi-lingual DNN in place of the original English DNN on these multi-language conditions. We show that not only the gain from using DNNs vanishes, but also the DNN-based systems tend to produce de-calibrated scores under the studied conditions. This work gives suggestions for directions of future research rather than any particular solutions.

This work studies the usage of the (currently state-of-the-art) Deep Neural Networks (DNN) i-vector/PLDA-based speaker recognition sys- tems in multi-language (especially non-English) conditions. On the "Lan- guage Pack" of the PRISM set, we evaluate the systems performance using NISTs standard metrics. We study the use of multi-lingual DNN in place of the original English DNN on these multi-language conditions. We show that not only the gain from using DNNs vanishes, but also the DNN-based systems tend to produce de-calibrated scores under the studied conditions. This work gives suggestions for directions of future research rather than any particular solutions.

2016

5

Faculty of Information Technology BUT

Brno