Accuracy Analysis of Generalized Pronunciation Variant Selection in ASR Systems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F09%3A00157411" target="_blank" >RIV/68407700:21230/09:00157411 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Accuracy Analysis of Generalized Pronunciation Variant Selection in ASR Systems

Popis výsledku v původním jazyce

Automated speech recognition (ASR) systems work typically with pronunciation dictionary for generating expected phonetic content of particular words in recognized utterance. But the pronunciation can vary in many situations. Besides the cases with more possible pronunciation variants specified manually in the dictionary there are typically many other possible changes in the pronunciation depending on word context or speaking style, very typical for our case of Czech language. In this paper we have studied the accuracy of proper selection of automatically predicted pronunciation variants in Czech HMM ASR based systems. We have analyzed correctness of pronunciation variant selection in forced alignment of known utterances. Using the proper pronunciationvariant were created mainly for the more accurate training of acoustic HMM models. Finally, the accuracy of LVCSR results using different levels of automated pronunciation generation were tested.

Název v anglickém jazyce

Accuracy Analysis of Generalized Pronunciation Variant Selection in ASR Systems

Popis výsledku anglicky

Automated speech recognition (ASR) systems work typically with pronunciation dictionary for generating expected phonetic content of particular words in recognized utterance. But the pronunciation can vary in many situations. Besides the cases with more possible pronunciation variants specified manually in the dictionary there are typically many other possible changes in the pronunciation depending on word context or speaking style, very typical for our case of Czech language. In this paper we have studied the accuracy of proper selection of automatically predicted pronunciation variants in Czech HMM ASR based systems. We have analyzed correctness of pronunciation variant selection in forced alignment of known utterances. Using the proper pronunciationvariant were created mainly for the more accurate training of acoustic HMM models. Finally, the accuracy of LVCSR results using different levels of automated pronunciation generation were tested.

Druh

J_x - Nezařazeno - Článek v odborném periodiku (Jimp, Jsc a Jost)

CEP obor

JA - Elektronika a optoelektronika, elektrotechnika

OECD FORD obor

—

Projekt

<a href="/cs/project/GA102%2F08%2F0707" target="_blank" >GA102/08/0707: Rozpoznávání mluvené řeči v reálných podmínkách</a><br>

Návaznosti

Z - Vyzkumny zamer (s odkazem do CEZ)

Rok uplatnění

2009

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Lecture Notes in Artificial Intelligence

ISSN

0302-9743

e-ISSN

—

Svazek periodika

5641

Číslo periodika v rámci svazku

2009931057

Stát vydavatele periodika

DE - Spolková republika Německo

Počet stran výsledku

10

Strana od-do

—

Kód UT WoS článku

—

EID výsledku v databázi Scopus

—