Pronunciation generation in speech recognition
DCFirst Claim
1. A method of adding a word to a speech recognition vocabulary, comprising:
- receiving a spelling of the word,receiving an utterance of the word,creating a collection of possible phonetic pronunciations of the word by;
comparing the spelling to a rules list of letter strings with associated phonemes, wherein the comparing includes searching the letter strings of the rules list for a letter string from the spelling of length greater than one letter, andlimiting the collection of possible phonetic pronunciations to phonetic pronunciations containing phonemes associated with the letter string of length greater than one,using speech recognition to find a best-matching pronunciation from the collection that best matches the utterance of the word, andadding the word to the speech recognition vocabulary using the spelling and the best-matching pronunciation.
8 Assignments
Litigations
0 Petitions
Reexamination
Accused Products
Abstract
A method of adding a word to a speech recognition vocabulary includes creating a collection of possible phonetic pronunciations from a spelling of the word and using speech recognition to find a pronunciation from the collection that best matches an utterance of the word. The collection is created by comparing the spelling to a rules list of letter strings with associated phonemes. The list is searched for a letter string from the spelling of length greater than one letter. The collection is limited to phonetic pronunciations containing phonemes associated with the letter string of length greater than one. In another method, a net of possible phonetic pronunciations of the word is created from the spelling and speech recognition is used to find the pronunciation from the net that best matches the utterance of the word. The invention also features methods of assigning a pre-filtering class to a word.
-
Citations
14 Claims
-
1. A method of adding a word to a speech recognition vocabulary, comprising:
-
receiving a spelling of the word, receiving an utterance of the word, creating a collection of possible phonetic pronunciations of the word by; comparing the spelling to a rules list of letter strings with associated phonemes, wherein the comparing includes searching the letter strings of the rules list for a letter string from the spelling of length greater than one letter, and limiting the collection of possible phonetic pronunciations to phonetic pronunciations containing phonemes associated with the letter string of length greater than one, using speech recognition to find a best-matching pronunciation from the collection that best matches the utterance of the word, and adding the word to the speech recognition vocabulary using the spelling and the best-matching pronunciation. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A method of adding a word to a speech recognition vocabulary, comprising:
-
receiving a spelling of the word, receiving an utterance of the word, creating a net of possible phonetic pronunciations of the word by comparing the spelling to a rules list of letter strings with associated phonemes, using speech recognition to find a best-matching pronunciation from the net that best matches the utterance of the word, and adding the word to the speech recognition vocabulary using the spelling and the best-matching pronunciation.
-
-
8. A method of assigning a pre-filtering class to a word when adding a word to a speech recognition dictionary, the method comprising:
-
matching first two letters of the word to classified words starting with the same two letters to form a sub-list of classified words, matching phonemes of the word to phonemes of the classified words in the sub-list to form a class list, and placing the word in the class list. - View Dependent Claims (9)
-
-
10. A method of assigning a pre-filtering class to a word when adding a word to a speech recognition dictionary, the comprising:
-
performing a direct look-up of a first phoneme of the word in a database of classified words organized alphabetically by their phonemes, matching the first phoneme of the word to a first word in the database having the same first phoneme, selecting the first word in the database having the same first phoneme and following words in the database to form a sub-list, matching the first phoneme of the word to the first phoneme of the classified words in the sub-list to form a class list, and placing the word in the class list. - View Dependent Claims (11, 12, 13, 14)
-
Specification