Result Details
BUT Neural Network Features for Spontaneous Vietnamese in BABEL
Grézl František, Ing., Ph.D., DCGM (FIT)
Hannemann Mirko, Ph.D., DCGM (FIT)
Černocký Jan, prof. Dr. Ing., DCGM (FIT)
The paper deals with multiple facets of NN feature extraction training.Not surprisingly, we found that data preparation is crucial for thesuccess of NN training. In case we dispose of data from other (wellrepresented) languages, we should go for it as we have shown thatmultilingual fine-tuning outperforms unsupervised training.
speech recognition, discriminative training, bottleneckneural networks, adaptation of neural networks, regiondependent transforms
This paper presents our work on speech recognition of Vietnamese spontaneous telephone conversations. It focuses on feature extraction by Stacked Bottle-Neck neural networks: several improvements such as semi-supervised training on untranscribed data, increasing of precision of state targets, and CMLLR adaptations were investigated. We have also tested speaker adaptive training of this architecture and significant gain was found. The results are reported on BABEL Vietnamese data.
@inproceedings{BUT111541,
author="Martin {Karafiát} and František {Grézl} and Mirko {Hannemann} and Jan {Černocký}",
title="BUT Neural Network Features for Spontaneous Vietnamese in BABEL",
booktitle="Proceedings of ICASSP 2014",
year="2014",
pages="5659--5663",
publisher="IEEE Signal Processing Society",
address="Florencie",
doi="10.1109/ICASSP.2014.6854679",
isbn="978-1-4799-2892-7",
url="http://www.fit.vutbr.cz/research/groups/speech/publi/2014/karafiat_icassp2014_p5659.pdf"
}
IARPA Building Speech Recognition for Keyword Search in a New Language in a Week with Limited Training Data (BABEL) - Babelon, BBN, start: 2012-03-05, end: 2016-11-04, completed
Speech recognition for low-resource languages, GACR, Postdoktorandské granty, GPP202/12/P604, start: 2012-01-01, end: 2014-12-31, completed
Technologies of speech processing for efficient human-machine communication, TAČR, Program aplikovaného výzkumu a experimentálního vývoje ALFA, TA01011328, start: 2011-01-01, end: 2014-12-31, completed