Method for learning linguistically valid word pronunciations from acoustic data
First Claim
1. In a computer system, a method of generating pronunciations for a word that is represented by a waveform and text, such that the pronunciations are spelled by phones in a phonetic alphabet for storage in a pronunciation dictionary, the method comprising:
- a. graphing sets of initial pronunciations;
thereafterb. in an ASR subsystem determining a highest-scoring set of initial pronunciations;
c. generating sets of alternate pronunciations, wherein each set of alternate pronunciations includes the highest-scoring set of initial pronunciations with a lowest-probability phone of the highest-scoring initial pronunciation substituted with a unique-substitute phone;
d. graphing the sets of alternate pronunciations;
e. determining in the ASR subsystem a highest-scoring set of alternate pronunciations; and
f. adding to a pronunciation dictionary the highest-scoring set of alternate pronunciations.
5 Assignments
0 Petitions
Accused Products
Abstract
A computerized method is provided for generating pronunciations for words and storing the pronunciations in a pronunciation dictionary. The method includes graphing sets of initial pronunciations; thereafter in an ASR subsystem determining a highest-scoring set of initial pronunciations; generating sets of alternate pronunciations, wherein each set of alternate pronunciations includes the highest-scoring set of initial pronunciations with a lowest-probability phone of the highest-scoring initial pronunciation substituted with a unique-substitute phone; graphing the sets of alternate pronunciations; determining in the ASR subsystem a highest-scoring set of alternate pronunciations; and adding to a pronunciation dictionary the highest-scoring set of alternate pronunciations.
45 Citations
28 Claims
-
1. In a computer system, a method of generating pronunciations for a word that is represented by a waveform and text, such that the pronunciations are spelled by phones in a phonetic alphabet for storage in a pronunciation dictionary, the method comprising:
-
a. graphing sets of initial pronunciations;
thereafterb. in an ASR subsystem determining a highest-scoring set of initial pronunciations; c. generating sets of alternate pronunciations, wherein each set of alternate pronunciations includes the highest-scoring set of initial pronunciations with a lowest-probability phone of the highest-scoring initial pronunciation substituted with a unique-substitute phone; d. graphing the sets of alternate pronunciations; e. determining in the ASR subsystem a highest-scoring set of alternate pronunciations; and f. adding to a pronunciation dictionary the highest-scoring set of alternate pronunciations. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18)
-
-
19. In a computer system, a method of generating pronunciations for a word that is represented by a waveform and text, such that the pronunciations are spelled by phones in a phonetic alphabet for storage in a pronunciation dictionary, the method comprising:
-
a. graphing sets of initial pronunciations;
thereafterb. in an ASR subsystem determining a highest-scoring set of initial pronunciations; c. generating a set of alternate pronunciations that includes the highest-scoring set of initial pronunciations with a lowest-probability phone of the highest-scoring initial pronunciation substituted with a unique-substitute phone; and d. adding to a pronunciation dictionary the set of alternate pronunciations and the highest-scoring set of initial pronunciations. - View Dependent Claims (20, 21, 22, 23, 24, 25, 26, 27, 28)
-
Specification