#20943We describe the methods and hardware that we are using to produce a real-time demonstration of an integratedSpoken Language System.
other,12-2-H90-1016,ak
computation needed to compute the
<term>
N-Best sentence hypotheses
</term>
. To avoid
<term>
grammar coverage
#20959We describe algorithms that greatly reduce the computation needed to compute theN-Best sentence hypotheses.
other,2-3-H90-1016,ak
sentence hypotheses
</term>
. To avoid
<term>
grammar coverage problems
</term>
we use a
<term>
fully-connected first-order
#20965To avoidgrammar coverage problems we use a fully-connected first-order statistical class grammar.
lr,8-3-H90-1016,ak
grammar coverage problems
</term>
we use a
<term>
fully-connected first-order statistical class grammar
</term>
. The
<term>
speech-search algorithm
#20971To avoid grammar coverage problems we use afully-connected first-order statistical class grammar.
tech,1-4-H90-1016,ak
statistical class grammar
</term>
. The
<term>
speech-search algorithm
</term>
is implemented on a board with a
#20978Thespeech-search algorithm is implemented on a board with a single Intel i860 chip, which provides a factor of 5 speed-up over a SUN 4 for straight C code.
tech,19-5-H90-1016,ak
controls the system and contains the
<term>
natural language system
</term>
and application back end . This paper
#21027The board plugs directly into the VME bus of the SUN4, which controls the system and contains thenatural language system and application back end.