The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
La reconnaissance statistique de la parole utilisant des modèles de Markov cachés à densité continue (CDHMM) a donné naissance à de nombreuses applications pratiques. Cependant, en général, les discordances entre les données d’apprentissage et les données d’entrée dégradent considérablement la précision de la reconnaissance. Diverses techniques d'adaptation de modèles acoustiques utilisant quelques énoncés d'entrée ont été utilisées pour surmonter ce problème. Dans cet article, nous examinons ces techniques d'adaptation, notamment l'estimation du maximum a posteriori (MAP), la régression linéaire du maximum de vraisemblance (MLLR) et l'eigenvoice. Nous présentons également une vue schématique appelée pyramide d'adaptation pour illustrer les relations entre ces méthodes.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Koichi SHINODA, "Acoustic Model Adaptation for Speech Recognition" in IEICE TRANSACTIONS on Information,
vol. E93-D, no. 9, pp. 2348-2362, September 2010, doi: 10.1587/transinf.E93.D.2348.
Abstract: Statistical speech recognition using continuous-density hidden Markov models (CDHMMs) has yielded many practical applications. However, in general, mismatches between the training data and input data significantly degrade recognition accuracy. Various acoustic model adaptation techniques using a few input utterances have been employed to overcome this problem. In this article, we survey these adaptation techniques, including maximum a posteriori (MAP) estimation, maximum likelihood linear regression (MLLR), and eigenvoice. We also present a schematic view called the adaptation pyramid to illustrate how these methods relate to each other.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.E93.D.2348/_p
Copier
@ARTICLE{e93-d_9_2348,
author={Koichi SHINODA, },
journal={IEICE TRANSACTIONS on Information},
title={Acoustic Model Adaptation for Speech Recognition},
year={2010},
volume={E93-D},
number={9},
pages={2348-2362},
abstract={Statistical speech recognition using continuous-density hidden Markov models (CDHMMs) has yielded many practical applications. However, in general, mismatches between the training data and input data significantly degrade recognition accuracy. Various acoustic model adaptation techniques using a few input utterances have been employed to overcome this problem. In this article, we survey these adaptation techniques, including maximum a posteriori (MAP) estimation, maximum likelihood linear regression (MLLR), and eigenvoice. We also present a schematic view called the adaptation pyramid to illustrate how these methods relate to each other.},
keywords={},
doi={10.1587/transinf.E93.D.2348},
ISSN={1745-1361},
month={September},}
Copier
TY - JOUR
TI - Acoustic Model Adaptation for Speech Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 2348
EP - 2362
AU - Koichi SHINODA
PY - 2010
DO - 10.1587/transinf.E93.D.2348
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E93-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2010
AB - Statistical speech recognition using continuous-density hidden Markov models (CDHMMs) has yielded many practical applications. However, in general, mismatches between the training data and input data significantly degrade recognition accuracy. Various acoustic model adaptation techniques using a few input utterances have been employed to overcome this problem. In this article, we survey these adaptation techniques, including maximum a posteriori (MAP) estimation, maximum likelihood linear regression (MLLR), and eigenvoice. We also present a schematic view called the adaptation pyramid to illustrate how these methods relate to each other.
ER -