Methodology for implementing a vocabulary set for use in a speech recognition system
First Claim
1. A system for implementing a vocabulary set for a speech recognizer, comprising:
- a recognizer for analyzing utterances from said vocabulary set to generate N-best lists of recognition candidates;
an acoustical matrix configured to relate said utterances to top recognition candidates from said N-best lists;
a lexical matrix configured to relate said utterances to said top recognition candidates from said N-best lists only when second-highest recognition candidates from said N-best lists are correct recognition results; and
an utterance ranking created according to composite individual error/accuracy values for each of said utterances, said composite individual error/accuracy values being derived from both said acoustical matrix and said lexical matrix, a lowest-ranked utterance being eliminated from said vocabulary set when a total error/accuracy value for all of said utterances does not exceed a predetermined threshold.
1 Assignment
0 Petitions
Accused Products
Abstract
The present invention comprises a methodology for implementing a vocabulary set for use in a speech recognition system, and may preferably include a recognizer for analyzing utterances from the vocabulary set to generate N-best lists of recognition candidates. The N-best lists may then be utilized to create an acoustical matrix configured to relate said utterances to top recognition candidates from said N-best lists, as well as a lexical matrix configured to relate the utterances to the top recognition candidates from the N-best lists only when second-highest recognition candidates from the N-best lists are correct recognition results. An utterance ranking may then preferably be created according to composite individual error/accuracy values for each of the utterances. The composite individual error/accuracy values may preferably be derived from both the acoustical matrix and the lexical matrix. Lowest-ranked utterances from the foregoing utterance ranking may preferably be repeatedly eliminated from the vocabulary set when a total error/accuracy value for all of the utterances fails to exceed a predetermined threshold value.
11 Citations
43 Claims
-
1. A system for implementing a vocabulary set for a speech recognizer, comprising:
-
a recognizer for analyzing utterances from said vocabulary set to generate N-best lists of recognition candidates;
an acoustical matrix configured to relate said utterances to top recognition candidates from said N-best lists;
a lexical matrix configured to relate said utterances to said top recognition candidates from said N-best lists only when second-highest recognition candidates from said N-best lists are correct recognition results; and
an utterance ranking created according to composite individual error/accuracy values for each of said utterances, said composite individual error/accuracy values being derived from both said acoustical matrix and said lexical matrix, a lowest-ranked utterance being eliminated from said vocabulary set when a total error/accuracy value for all of said utterances does not exceed a predetermined threshold. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
-
-
21. A method for implementing a vocabulary set for a speech recognizer, comprising the steps of:
-
analyzing utterances from said vocabulary set with a recognizer to generate N-best lists of recognition candidates;
relating said utterances to top recognition candidates from said N-best lists with an acoustical matrix;
compiling a lexical matrix that relates said utterances to said top recognition candidates from said N-best lists only when second-highest recognition candidates from said N-best lists are correct recognition results; and
creating an utterance ranking according to composite individual error/accuracy values for each of said utterances, said composite individual error/accuracy values being derived from both said acoustical matrix and said lexical matrix, a lowest-ranked utterance being eliminated from said vocabulary set when a total error/accuracy value for all of said utterances does not exceed a predetermined threshold. - View Dependent Claims (22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40)
-
-
41. A computer-readable medium comprising program instructions for implementing a vocabulary set for a speech recognizer, by performing the steps of:
-
analyzing utterances from said vocabulary set with a recognizer to generate N-best lists of recognition candidates;
relating said utterances to top recognition candidates from said N-best lists with an acoustical matrix;
compiling a lexical matrix that relates said utterances to said top recognition candidates from said N-best lists only when second-highest recognition candidates from said N-best lists are correct recognition results; and
creating an utterance ranking according to composite individual error/accuracy values for each of said utterances, said composite individual error/accuracy values being derived from both said acoustical matrix and said lexical matrix, a lowest-ranked utterance being eliminated from said vocabulary set when a total error/accuracy value for all of said utterances does not exceed a predetermined threshold.
-
-
42. A system for implementing a vocabulary set for a speech recognizer, comprising the steps of:
-
means for analyzing utterances from said vocabulary set to generate N-best lists of recognition candidates;
means for relating said utterances to top recognition candidates from said N-best lists;
means for correlating said utterances to said top recognition candidates from said N-best lists only when second-highest recognition candidates from said N-best lists are correct recognition results; and
means for ranking said utterances according to composite individual error/accuracy values for each of said utterances, said composite individual error/accuracy values being derived from both said means for relating and said means for correlating, a lowest-ranked utterance being eliminated from said vocabulary set when a total error/accuracy value for all of said utterances does not exceed a predetermined threshold.
-
-
43. A system for implementing a vocabulary set for a speech recognizer, comprising:
-
a recognizer for analyzing utterances from said vocabulary set to generate recognition candidates;
an acoustical matrix configured to relate said utterances to top recognition candidates;
a lexical matrix configured to relate said utterances to said top recognition candidates only when second-highest recognition candidates are correct recognition results; and
an utterance ranking of said utterances based upon both said acoustical matrix and said lexical matrix, a lowest-ranked utterance being eliminated from said vocabulary set when a recognition accuracy for all of said utterances fails to exceed a predetermined threshold.
-
Specification