This HTML5 document contains 45 embedded RDF statements represented using HTML+Microdata notation.

The embedded RDF content will be recognized by any processor of HTML5 Microdata.

Namespace Prefixes

PrefixIRI
n17http://linked.opendata.cz/ontology/domain/vavai/riv/typAkce/
dctermshttp://purl.org/dc/terms/
n22http://localhost/temp/predkladatel/
n14http://purl.org/net/nknouf/ns/bibtex#
n13http://linked.opendata.cz/resource/domain/vavai/projekt/
n5http://linked.opendata.cz/resource/domain/vavai/riv/tvurce/
n18http://linked.opendata.cz/resource/domain/vavai/subjekt/
n8http://linked.opendata.cz/ontology/domain/vavai/
n11https://schema.org/
shttp://schema.org/
skoshttp://www.w3.org/2004/02/skos/core#
n3http://linked.opendata.cz/ontology/domain/vavai/riv/
n2http://linked.opendata.cz/resource/domain/vavai/vysledek/
rdfhttp://www.w3.org/1999/02/22-rdf-syntax-ns#
n20http://linked.opendata.cz/resource/domain/vavai/vysledek/RIV%2F49777513%3A23520%2F12%3A43916037%21RIV13-MK0-23520___/
n9http://linked.opendata.cz/ontology/domain/vavai/riv/klicoveSlovo/
n15http://linked.opendata.cz/ontology/domain/vavai/riv/duvernostUdaju/
xsdhhttp://www.w3.org/2001/XMLSchema#
n19http://linked.opendata.cz/ontology/domain/vavai/riv/jazykVysledku/
n4http://linked.opendata.cz/ontology/domain/vavai/riv/aktivita/
n21http://linked.opendata.cz/ontology/domain/vavai/riv/druhVysledku/
n12http://linked.opendata.cz/ontology/domain/vavai/riv/obor/
n10http://reference.data.gov.uk/id/gregorian-year/

Statements

Subject Item
n2:RIV%2F49777513%3A23520%2F12%3A43916037%21RIV13-MK0-23520___
rdf:type
n8:Vysledek skos:Concept
dcterms:description
This paper evaluates a recently published method for supervised and unsupervised adaptation of neural networks used in hybrid speech recognition systems. The neural networks used in the field of hybrid speech recognition have certain distinct characteristics that make the usual adaptation methods (such as retraining the neural network) unusable or uneffective. The recently published MELT (Minimum Error Linear Transform) method [1] has been developed to cope with this issue. By providing a way of establishing a link between the intermediate features and the long temporal features, the number of free variables can be reduced significantly and the resulting adaptation parameters can be estimated robustly. The experiments were performed on the WSJCAM0 speech corpus. Contrary to the original paper [1], the experiments were performed using a word recognizer instead of a phoneme recognizer. The experimental results suggest that the MELT method can be used both in an unsupervised as well as a semi-supervised manner and when applied, it leads to significant reduction of word error rate, even for a strong language model. This paper evaluates a recently published method for supervised and unsupervised adaptation of neural networks used in hybrid speech recognition systems. The neural networks used in the field of hybrid speech recognition have certain distinct characteristics that make the usual adaptation methods (such as retraining the neural network) unusable or uneffective. The recently published MELT (Minimum Error Linear Transform) method [1] has been developed to cope with this issue. By providing a way of establishing a link between the intermediate features and the long temporal features, the number of free variables can be reduced significantly and the resulting adaptation parameters can be estimated robustly. The experiments were performed on the WSJCAM0 speech corpus. Contrary to the original paper [1], the experiments were performed using a word recognizer instead of a phoneme recognizer. The experimental results suggest that the MELT method can be used both in an unsupervised as well as a semi-supervised manner and when applied, it leads to significant reduction of word error rate, even for a strong language model.
dcterms:title
Unsupervised and semi-supervised adaptation of a hybrid speech recognition system Unsupervised and semi-supervised adaptation of a hybrid speech recognition system
skos:prefLabel
Unsupervised and semi-supervised adaptation of a hybrid speech recognition system Unsupervised and semi-supervised adaptation of a hybrid speech recognition system
skos:notation
RIV/49777513:23520/12:43916037!RIV13-MK0-23520___
n8:predkladatel
n18:orjk%3A23520
n3:aktivita
n4:P
n3:aktivity
P(DF12P01OVV022)
n3:dodaniDat
n10:2013
n3:domaciTvurceVysledku
n5:6895972 n5:6899390 n5:7736967
n3:druhVysledku
n21:D
n3:duvernostUdaju
n15:S
n3:entitaPredkladatele
n20:predkladatel
n3:idSjednocenehoVysledku
176074
n3:idVysledku
RIV/49777513:23520/12:43916037
n3:jazykVysledku
n19:eng
n3:klicovaSlova
Neural Networks, Speech Recognition, Speaker Adaptation, MELT
n3:klicoveSlovo
n9:Speaker%20Adaptation n9:Neural%20Networks n9:Speech%20Recognition n9:MELT
n3:kontrolniKodProRIV
[E16F1E68D0A6]
n3:mistoKonaniAkce
Bejing (Peking)
n3:mistoVydani
Beijing (Peking)
n3:nazevZdroje
Proceedings 2012 IEEE 11th International Conference on Signal Processing
n3:obor
n12:JD
n3:pocetDomacichTvurcuVysledku
3
n3:pocetTvurcuVysledku
3
n3:projekt
n13:DF12P01OVV022
n3:rokUplatneniVysledku
n10:2012
n3:tvurceVysledku
Müller, Luděk Zelinka, Jan Trmal, Jan
n3:typAkce
n17:WRD
n3:zahajeniAkce
2012-10-21+02:00
s:numberOfPages
4
n14:hasPublisher
IEEE Press
n11:isbn
978-1-4673-2194-5
n22:organizacniJednotka
23520