Aligning a transcript to audio data
First Claim
1. A computer-implemented method, comprising:
- receiving audio data and a transcript of the audio data;
generating, from the transcript, a language model comprising a factor automaton that includes automaton states and arcs, each of the automaton arcs corresponding to a language element from the transcript;
receiving, from a speech recognizer, language elements recognized from the received audio data and times at which the recognized language elements occur in the audio data;
comparing the recognized language elements from the audio data to one or more of the language elements from the transcript and represented by the factor automaton to identify times at which the one or more of language elements from the transcript occur in the audio data;
aligning a portion of the transcript with a portion of the audio data using the identified times; and
outputting the aligned portion of the transcript.
2 Assignments
0 Petitions
Accused Products
Abstract
The subject matter of this specification can be implemented in, among other things, a computer-implemented method including receiving audio data and a transcript of the audio data. The method further includes generating a language model including a factor automaton that includes automaton states and arcs, each of the automaton arcs corresponding to a language element from the transcript. The method further includes receiving language elements recognized from the received audio data and times at which each of the recognized language elements occur in the audio data. The method further includes comparing the recognized language elements to one or more of the language elements from the factor automaton to identify times at which the one or more of the language elements from the transcript occur in the audio data. The method further includes aligning a portion of the transcript with a portion of the audio data using the identified times.
142 Citations
23 Claims
-
1. A computer-implemented method, comprising:
-
receiving audio data and a transcript of the audio data; generating, from the transcript, a language model comprising a factor automaton that includes automaton states and arcs, each of the automaton arcs corresponding to a language element from the transcript; receiving, from a speech recognizer, language elements recognized from the received audio data and times at which the recognized language elements occur in the audio data; comparing the recognized language elements from the audio data to one or more of the language elements from the transcript and represented by the factor automaton to identify times at which the one or more of language elements from the transcript occur in the audio data; aligning a portion of the transcript with a portion of the audio data using the identified times; and outputting the aligned portion of the transcript. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
-
-
21. A computer-implemented method comprising:
-
receiving audio data and a transcript of the audio data; generating a graph comprising states and arcs that represent language elements in the transcript, wherein the states are linked in a configuration to permit the representation of substantially all possible phrases that occur within the received transcript; generating a speech hypothesis of one or more language elements included in the audio using a speech recognizer, wherein the speech hypothesis includes time stamps associated with an occurrence of each of the one or more language elements; comparing the one or more language elements of the speech hypothesis with one or more arcs in the graph; and associating at least a portion of the language elements of the transcript with the time stamps based on the comparison. - View Dependent Claims (22)
-
-
23. A computer-implemented system, comprising:
-
one or more computers having an interface that receives audio data and a transcript of the audio data; a model builder module that generates, from the transcript, a language model comprising a factor automaton that includes automaton states and arcs, each of the automaton arcs corresponding to a language element from the transcript; a storage device that stores the factor automaton; a speech recognizer that recognizes language elements from the received audio data and determines times at which each of the recognized language elements occur in the audio data; and means for comparing the recognized language elements to one or more of the language elements from the stored factor automaton to identify times at which the one or more of the language elements from the transcript occur in the audio data and aligning a portion of the transcript with a portion of the audio data using the identified times, and outputting the aligned portion of the transcript.
-
Specification