Methods and apparatus for flexible speech recognition
First Claim
Patent Images
1. A computer implemented method, comprising:
- receiving a speech signal;
extracting a sequence of acoustic feature vectors from the speech signal;
identifying a set of one or more words from the speech signal;
identifying a set of one or more phrases from the speech signal, the set of one or more phrases including a sequence of words, wherein the set of one or more phrases comprises key phrases and filler phrases;
assigning a first probability to a phrase of the one or more phrases, the first probability representing a likelihood that the sequence of acoustic feature vectors was produced by the phrase;
assigning a second probability to the phrase of the one or more phrases, the second probability representing a likelihood that the phrase would be received;
assigning a score to the phrase of the one or more phrases, the score calculated using the first probability and the second probability;
andevaluating the speech signal against the set of phrases using the score.
3 Assignments
0 Petitions
Accused Products
Abstract
The present invention is directed to a computer implemented method and apparatus for flexibly recognizing meaningful data items within an arbitrary user utterance. According to one example embodiment of the invention, a set of one or more key phrases and a set of one or more filler phrases are defined, probabilities are assigned to the key phrases and/or the filler phrases, and the user utterances is evaluated against the set of key phrases and the set of filler phrases using the probabilities.
-
Citations
34 Claims
-
1. A computer implemented method, comprising:
-
receiving a speech signal; extracting a sequence of acoustic feature vectors from the speech signal; identifying a set of one or more words from the speech signal; identifying a set of one or more phrases from the speech signal, the set of one or more phrases including a sequence of words, wherein the set of one or more phrases comprises key phrases and filler phrases; assigning a first probability to a phrase of the one or more phrases, the first probability representing a likelihood that the sequence of acoustic feature vectors was produced by the phrase; assigning a second probability to the phrase of the one or more phrases, the second probability representing a likelihood that the phrase would be received; assigning a score to the phrase of the one or more phrases, the score calculated using the first probability and the second probability; and evaluating the speech signal against the set of phrases using the score. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33)
-
-
34. A computer implemented method, comprising:
-
receiving a speech signal; extracting a sequence of acoustic feature vectors from the speech signal; identifying a set of one or more words from the speech signal; identifying a set of one or more phrases from the speech signal, the set of one or more phrases including a sequence of words, wherein the set of one or more phrases comprises key phrases and filler phrases; specifying a key phrase context-free grammar that produces the key phrases; specifying a filler context-free grammar that produces the filler phrases; assigning a first probability to a phrase of the one or more phrases, the first probability representing a likelihood that the sequence of acoustic feature vectors was produced by the phrase; assigning a second probability to the phrase of the one or more phrases, the second probability representing a likelihood of receiving the phrase; assigning a score to the phrase of the one or more phrases, the score calculated using the first probability and the second probability; defining a sentence context-free grammar using the key phrase context-free grammar, the filler context-free grammars, the scores; and evaluating the speech signal against the sentence context-free grammar to extract a key phrase from the speech signal.
-
Specification