Title:
CROSS-LINGUAL INITIALIZATION OF LANGUAGE MODELS
Document Type and Number:
WIPO Patent Application WO/2012/148957
Kind Code:
A4
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for initializing language models for automatic speech recognition. In one aspect, a method includes receiving logged speech recognition results from an existing corpus that is specific to a given language and a target context, generating a target corpus by machine-translating the logged speech recognition results from the given language to a different, target language, and estimating a language model that is specific to the different, target language and the same, target context, using the target corpus.
Inventors:
NAKAJIMA KAISUKE (US)
STROPE BRIAN (US)
STROPE BRIAN (US)
Application Number:
PCT/US2012/034889
Publication Date:
March 28, 2013
Filing Date:
April 25, 2012
Export Citation:
Assignee:
GOOGLE INC (US)
NAKAJIMA KAISUKE (US)
STROPE BRIAN (US)
NAKAJIMA KAISUKE (US)
STROPE BRIAN (US)
International Classes:
G10L15/00; G06F17/28; G10L15/18
Attorney, Agent or Firm:
CARTER, Michael, D. et al. (P.O. Box 1022Minneapolis, Minnesota, US)
Download PDF:
Previous Patent: FLANGE OVERSHOT RETRIEVAL TOOL
Next Patent: BIOLOGICALLY INSPIRED BEAM FORMING SMALL ANTENNA ARRAYS
Next Patent: BIOLOGICALLY INSPIRED BEAM FORMING SMALL ANTENNA ARRAYS