Impact of Phonetic Annotation Precision on Automatic Speech Recognition Systems

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F46747885%3A24220%2F16%3A00000466" target="_blank" >RIV/46747885:24220/16:00000466 - isvavai.cz</a>

Výsledek na webu

<a href="http://dx.doi.org/10.1109/TSP.2016.7760886" target="_blank" >http://dx.doi.org/10.1109/TSP.2016.7760886</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/TSP.2016.7760886" target="_blank" >10.1109/TSP.2016.7760886</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Impact of Phonetic Annotation Precision on Automatic Speech Recognition Systems

Popis výsledku v původním jazyce

In this paper we study the impact of phonetic annotation precision on the accuracy of a state-of-the art ASR (automatic speech recognition) system. This issue becomes important especially if we want to port the system to a new language without spending much time by collecting, checking and annotating a large amount of acoustic data in the target language. First, we describe a series of experiments that demonstrate how inaccurate annotation influences word error rate of a gaussian-mixture model and deep-neural net based ASR. Our results show that even significantly inaccurate annotations allow to achieve fairly good results. On the other side, if we want to get really good performance, we should assure that the annotations are as precise as possible. Therefore, we describe methods that can do most of the annotation work in an almost automatic (iterative) way with a high precision. We demonstrate them on an ASR system that is being adapted from Czech language to Polish. We provide results achieved on Polish GlobalPhone and Clarin speech databases.

Název v anglickém jazyce

Impact of Phonetic Annotation Precision on Automatic Speech Recognition Systems

Popis výsledku anglicky

In this paper we study the impact of phonetic annotation precision on the accuracy of a state-of-the art ASR (automatic speech recognition) system. This issue becomes important especially if we want to port the system to a new language without spending much time by collecting, checking and annotating a large amount of acoustic data in the target language. First, we describe a series of experiments that demonstrate how inaccurate annotation influences word error rate of a gaussian-mixture model and deep-neural net based ASR. Our results show that even significantly inaccurate annotations allow to achieve fairly good results. On the other side, if we want to get really good performance, we should assure that the annotations are as precise as possible. Therefore, we describe methods that can do most of the annotation work in an almost automatic (iterative) way with a high precision. We demonstrate them on an ASR system that is being adapted from Czech language to Polish. We provide results achieved on Polish GlobalPhone and Clarin speech databases.

Druh

D - Stať ve sborníku

CEP obor

JC - Počítačový hardware a software

OECD FORD obor

—

Projekt

<a href="/cs/project/TA04010199" target="_blank" >TA04010199: MULTILINMEDIA - Multilinguální platforma pro monitoring a analýzu multimédií</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2016

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proc. of the 39th International Conference on Telecommunications and Signal Processing (TSP 2016)

ISBN

978-1-5090-1287-9

ISSN

1805-5435

e-ISSN

—

Počet stran výsledku

4

Strana od-do

311-314

Název nakladatele

Institute of Electrical and Electronics Engineers Inc.

Místo vydání

Rakousko, Vídeň

Místo konání akce

Rakousko, Vídeň

Datum konání akce

1. 1. 2016

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

000390164000067