457 search results - page 20 / 92 » Universal background model based speech recognition |
CSL
2007
Springer
14 years 9 months ago
2007
Springer
Humans are able to recognise a word before its acoustic realisation is complete. This in contrast to conventional automatic speech recognition (ASR) systems, which compute the lik...
ISCAS
2006
IEEE
15 years 3 months ago
2006
IEEE
Abstract—We present a silicon neuron with a dynamic, active leak that enables precise spike-timing with respect to a time-varying input signal. Our neuron models the mammalian bu...
FTSIG
2007
14 years 9 months ago
2007
Hidden Markov Models (HMMs) provide a simple and effective framework for modelling time-varying spectral vector sequences. As a consequence, almost all present day large vocabula...
ICPR
2008
IEEE
15 years 11 months ago
2008
IEEE
This work extends and improves a recently introduced (Dec. 2007) dynamic Bayesian network (DBN) based audio-visual automatic speech recognition (AVASR) system. That system models ...
ICDAR
2009
IEEE
14 years 7 months ago
2009
IEEE
In this paper we investigate the use of linguistic information given by language models to deal with word recognition errors on handwritten sentences. We focus especially on error...