System and method for tuning and testing in a speech recognition system
First Claim
1. A method of testing a speech recognizer, the method comprising:
- receiving a plurality of digital audio data files, each audio file comprising audio recorded in response to a first prompt by a speech recognition application;
receiving a grammar associated with the first prompt, the grammar comprising a plurality of concepts, each concept having a set of phrases organized under a single idea, the idea representing an expected response to the first prompt;
producing a first recognition result for each audio data file based at least in part on the grammar using the speech recognizer;
receiving a user-defined transcript of each audio file, and scoring the first recognition results for each audio data file based at least in part on the transcript of each audio file;
modifying the grammar based on the scoring of the first recognition result of each audio data file;
producing a second recognition result for each audio data file based on the modified grammar using the speech recognizer;
using the user-defined transcript of each audio data file to score the second recognition result for each audio data file;
comparing the scoring of the first recognition result with the second result for each audio data file; and
outputting the first or second recognition result for each audio data file based on the comparison.
4 Assignments
0 Petitions
Accused Products
Abstract
Systems and methods for improving the performance of a speech recognition system. In some embodiments a tuner module and/or a tester module are configured to cooperate with a speech recognition system. The tester and tuner modules can be configured to cooperate with each other. In one embodiment, the tuner module may include a module for playing back a selected portion of a digital data audio file, a module for creating and/or editing a transcript of the selected portion, and/or a module for displaying information associated with a decoding of the selected portion, the decoding generated by a speech recognition engine. In other embodiments, the tester module can include an editor for creating and/or modifying a grammar, a module for receiving a selected portion of a digital audio file and its corresponding transcript, and a scoring module for producing scoring statistics of the decoding based at least in part on the transcript.
78 Citations
44 Claims
-
1. A method of testing a speech recognizer, the method comprising:
-
receiving a plurality of digital audio data files, each audio file comprising audio recorded in response to a first prompt by a speech recognition application; receiving a grammar associated with the first prompt, the grammar comprising a plurality of concepts, each concept having a set of phrases organized under a single idea, the idea representing an expected response to the first prompt; producing a first recognition result for each audio data file based at least in part on the grammar using the speech recognizer; receiving a user-defined transcript of each audio file, and scoring the first recognition results for each audio data file based at least in part on the transcript of each audio file; modifying the grammar based on the scoring of the first recognition result of each audio data file; producing a second recognition result for each audio data file based on the modified grammar using the speech recognizer; using the user-defined transcript of each audio data file to score the second recognition result for each audio data file; comparing the scoring of the first recognition result with the second result for each audio data file; and outputting the first or second recognition result for each audio data file based on the comparison. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. A system for testing a speech recognizer, the system comprising:
-
an audio recorder module for receiving a plurality of digital audio data files, each data file comprising audio recorded in response to a first prompt of a speech recognition application; a grammar editor module configured to access and modify a grammar based on scoring of a recognition result, the grammar comprising a plurality of concepts, each concept having a set of phrases organized under a single idea, the idea representing an expected response to the first prompt; a speech recognition engine configured to output a first recognition result for each audio data file of the plurality of digital audio data files and the accessed grammar using the speech recognizer; and a scoring module configured to score the first recognition results based at least in part on a user-defined transcript of each audio data file of the plurality of audio data files, wherein said speech recognition engine is configured to output a second recognition result for each audio data file of the plurality of digital audio data files based on said modified grammar and wherein said scoring module is further configured to compare the scoring of the first recognition result with the second recognition result for each audio data file and output the first or second recognition result for each audio data file based on said comparison. - View Dependent Claims (16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30)
-
-
31. A system for testing a speech recognizer, the system comprising:
-
an audio data input module for receiving a plurality of digital audio data files, each audio data file comprising audio recorded in response to a first prompt from a speech recognition application; a grammar editor module configured to access and modify a grammar, the grammar comprising a plurality of concepts, each concept having a set of phrases organized under a single idea, the idea representing an expected response to the first prompt; a test module configured to initiate a first testing cycle, the testing cycle comprising transmitting the plurality of digital audio data files and the grammar to a speech recognition engine; and a scoring module configured to receive a first recognition result for each of the plurality of audio data files from the speech recognition engine, and further configured to score the first recognition results based at least in part on a user-defined transcript of the audio input, wherein said test module is configured to initiate a second testing cycle comprising transmitting the plurality of digital audio data files and the modified grammar to said speech recognition engine, and wherein said scoring module is configured to receive a recognition result for said second testing cycle for each of the plurality of audio data files based on the modified grammar, from the speech recognition engine and is further configured to compare the scoring of the first recognition result with the second recognition result for each audio data file and output the first or second recognition result for each audio data file based on said comparison. - View Dependent Claims (32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44)
-
Specification