INVESTIGADORES
ALBORNOZ Enrique Marcelo
artículos
Título:
Classification of ASR Word Hypotheses using prosodic information and resampling of training data
Autor/es:
ENRIQUE M. ALBORNOZ; DIEGO H. MILONE; HUGO L. RUFINER; RAMÓN LÓPEZ-CÓZAR
Revista:
LATIN AMERICAN APPLIED RESEARCH
Editorial:
PLAPIQUI(UNS-CONICET)
Referencias:
Lugar: Bahia Blanca; Año: 2013 vol. 43 p. 213 - 217
ISSN:
0327-0793
Resumen:
In this work, we propose a novel resampling method based on word lattice information and we use prosodic cues with support vector machines for classification. The idea isto consider word recognition as a two-class classification problem, which considers the word hypotheses in the lattice of a standard recogniser either as True or False employing prosodic information. The technique developed in this paper was applied to set of words extracted from a continuous speech database. Our experimental results show that the method allows obtaining average word hypotheses recognition rate of 82%.
