Publication Details
Analysis Of DNN Approaches To Speaker Identification
Glembek Ondřej, Ing., Ph.D.
Novotný 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)
automatic speaker identification, deep neural networks, bottleneck features, i-vector
This work studies the usage of the Deep Neural Network (DNN) Bottleneck (BN) features together with the traditional MFCC features in the task of i-vector-based speaker recognition. We decouple the sufficient statistics extraction by using separate GMM models for frame alignment, and for statistics normalization and we analyze the usage of BN and MFCC features (and their concatenation) in the two stages. We also show the effect of using full-covariance GMM models, and, as a contrast, we compare the result to the recent DNN-alignment approach. On the NIST SRE2010, telephone condition, we show 60% relative gain over the traditional MFCC baseline for EER (and similar for the NIST DCF metrics), resulting in 0.94% EER.
We have analyzed the i-vector based systems with Deep Neural Network (DNN) Bottleneck (BN) features together with the traditional MFCC features, and we have demonstrated substantial gain for NIST SRE 2010, telephone condition.
@inproceedings{BUT130927,
author="Pavel {Matějka} and Ondřej {Glembek} and Ondřej {Novotný} and Oldřich {Plchot} and František {Grézl} and Lukáš {Burget} and Jan {Černocký}",
title="Analysis Of DNN Approaches To Speaker Identification",
booktitle="Proceedings of the 41th IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2016), 2016",
year="2016",
pages="5100--5104",
publisher="IEEE Signal Processing Society",
address="Shanghai",
doi="10.1109/ICASSP.2016.7472649",
isbn="978-1-4799-9988-0",
url="https://www.fit.vut.cz/research/publication/11140/"
}