Přejít na obsah

Detail publikace


Zelinka Jan and Trmal Jan and Müller Luděk : Low-dimensional Space Transforms of Posteriors in Speech Recognition . Interspeech 2010, vol. 2010, p. 1193-1196, Curran Associates, Makuhari, Chiba, Japan, 2010.


In this paper we present three novel posterior transforms with the primary goal to achieve a high reduction of a feature vector size. The presented methods transform the posteriors to 1D or 2D space. For such a high reduction ratio the usually applied methods fail to keep the discriminative information. Contrary, the presented methods were specifically designed to retain most of the discriminative information. In our experiments, we used several different combinations of feature extraction methods nowadays commonly used, i.e. the PLP features (augmented with delta and acceleration coefficients) and two kinds of MLP-ANN features: the bottleneck (BN) and posterior estimates (PE). The experiments were designed with special attention to the assessment of possible improvements of the performance when the PLP features are combined either with the BN features or with the PE features whose dimensionality was reduced using the proposed feature transforms. The performance of the designed transforms was tested on two different speech corpora: a telephone speech SpeechDat-East corpus and multi-modal Czech Audio-Visual corpus.

Abstrakt v češtině

Článek popisuje transformaci posteriorů a jejich využití v rozpoznávání řeči.

Detail publikace

Název: Low-dimensional Space Transforms of Posteriors in Speech Recognition
Autor: Zelinka Jan ; Trmal Jan ; Müller Luděk
Název - česky: Tranformace posteriorů pro rozpoznávání řeči
Jazyk publikace: anglicky
Rok vydání: 2010
Typ publikace: Stať ve sborníku
Název časopisu / knihy: Interspeech 2010
Číslo vydání: 2010
Strana: 1193 - 1196
ISBN: 978-1-61782-123-3
Nakladatel: Curran Associates
Místo vydání: Makuhari, Chiba, Japan
Datum: 30.9.2010
2011-03-15 16:21:58 / 2011-03-15 16:21:58 / 1

Klíčová slova

speech recognition, posteriors, ANN, bottleneck

Klíčová slova v češtině



 author = {Zelinka Jan and Trmal Jan and M\"{u}ller Lud\v{e}k},
 title = {Low-dimensional Space Transforms of Posteriors in Speech Recognition},
 year = {2010},
 publisher = {Curran Associates},
 journal = {Interspeech 2010},
 address = {Makuhari, Chiba, Japan},
 volume = {2010},
 pages = {1193-1196},
 ISBN = {978-1-61782-123-3},
 url = {http://www.kky.zcu.cz/en/publications/ZelinkaJan_2010_Low-dimensionalSpace},