H92-1080 speaker normalization component or vocabulary adaptation component . The dictionary provided
W12-2705 paper we present an unsupervised vocabulary adaptation method for morph-based speech
H92-1080 our speaker normalization and vocabulary adaptation technology as well as experimenting
H92-1033 % error reduction . Our second vocabulary adaptation algorithm is to focus on the
P05-2021 recognition of inflective lan - guages . Vocabulary adaptation , however , brought considerable
W12-2705 of 18 % and 24 % . Unsupervised vocabulary adaptation was implemented through automatic
W12-2705 76.6 % and 80.7 % ) . Supervised vocabulary adaptation was implemented by manually retrieving
H92-1033 , it is desirable to implement vocabulary adaptation to make the VI system tailored
W12-2705 interpolation ( A = 0.1 ) supervised vocabulary adaptation reduces WER by 4 % ( general
H92-1033 vocabulary ( task ) . Our first vocabulary adaptation algorithm is to build vocabulary-adapted
H92-1033 . With the additional help of vocabulary adaptation , the vocabularyindependent system
H92-1033 speaker-independent system , two vocabulary adaptation algorithms \ -LSB- 5 \ -RSB-
H92-1033 Vocabulary.Adapted Decision Tree Our first vocabulary adaptation algorithm is to change the allophone
H92-1033 speaker adaptation techniques , our vocabulary adaptation algorithms only take advantage
H92-1033 paper , we first describe our two vocabulary adaptation algorithms , vocabulary-adapted
W00-0602 's multi-pass approach requires vocabulary adaptation and re-recognition of each complete
W12-2705 methods . <title> Unsupervised Vocabulary Adaptation for Morph-based Language Models
H92-1033 this work and future work . 3 Vocabulary Adaptation Unlike most speaker adaptation
H92-1033 paper , we have presented two vocabulary adaptation algonthms , including vocabulary-adapted
hide detail