Pronunciation discovery for spoken words
AssigneeNuance Communications, Inc.
InventorsDaniel L. Roth, Laurence S. Gillick, Mike Shire
Life of this patent
Estimating the term…
Abstract
A method of generating an alternative pronunciation for a word or phrase, given an initial pronunciation and a spoken example of the word or phrase, includes providing the initial pronunciation of the word or phrase, and generating the alternative pronunciation by searching a neighborhood of pronunciations about the initial pronunciation via a constrained hypothesis, wherein the neighborhood includes pronunciations that differ from the initial pronunciation by at most one phoneme. The method further includes selecting a highest scoring pronunciation within the neighborhood of pronunciations.
Claims30
A method comprising:
- providing a lexicon;
- receiving a spoken utterance of a word or phrase;
- after receiving the spoken utterance, obtaining an initial pronunciation for the spoken utterance;
- after obtaining the initial pronunciation, modifying the initial pronunciation to generate a plurality of alternative pronunciations;
- using the received spoken utterance to score each of the pronunciations among the plurality of alternative pronunciations;
- identifying a highest scoring pronunciation among the plurality of alternative pronunciations;
- and
- updating the lexicon with the highest scoring pronunciation.
A method comprising:
- providing a lexicon;
- receiving a spoken utterance of a word or phrase;
- after receiving the spoken utterance, obtaining an initial pronunciation for the spoken utterance;
- after obtaining the initial pronunciation, modifying the initial pronunciation to generate a plurality of alternative pronunciations by changing the initial pronunciation by one phoneme;
- identifying a highest scoring pronunciation among the plurality of alternative pronunciations;
- and
- updating the lexicon with the highest scoring pronunciation.
A non-transitory computer readable medium storing executable instructions which when executed on a computer system cause the computer system to:
- receive a spoken utterance of a word or phrase;
- after receiving the spoken utterance, obtain an initial pronunciation for the spoken utterance;
- after obtaining the initial pronunciation, modify the initial pronunciation to generate a plurality of alternative pronunciations by changing the initial pronunciation by one phoneme;
- and
- use the received spoken utterance to score each of the pronunciations among the plurality of alternative pronunciations;
- identify a highest scoring pronunciation among the plurality of alternative pronunciations:
- and
- update a lexicon with the highest scoring pronunciation.
Dependencies are inferred from the claim text.