| | |
Summary: Bayesian Language Model Interpolation for Mobile Speech Input
Cyril Allauzen, Michael Riley
Google Research, 76 Ninth Avenue, New York, NY, USA
allauzen@google.com, riley@google.com
Abstract
This paper explores various static interpolation methods for ap-
proximating a single dynamically-interpolated language model
used for a variety of recognition tasks on the Google Android
platform. The goal is to find the statically-interpolated first-
pass LM that best reduces search errors in a two-pass system
or that even allows eliminating the more complex dynamic sec-
ond pass entirely. Static interpolation weights that are uniform,
prior-weighted, and the maximum likelihood, maximum a pos-
teriori, and Bayesian solutions are considered. Analysis argues
and recognition experiments on Android test data show that a
Bayesian interpolation approach performs best.
Index Terms: speech recognition, language modeling, lan-
guage model interpolation
1. Introduction
Various speech-enabled features are available on the Android
|
