Automated learning for speech-based applications
View Patent ↗Systems and methods for modifying a computer-based speech recognition system. A speech utterance is processed with the computer-based speech recognition system using a set of internal representations, which may comprise parameters for recognizing speech in a speech utterance, such as parameters of an acoustic model and/or a language model. The computer-based speech recognition system may perform a first task in response to the processed speech utterance. The utterance may also be provided to a human who performs a second task based on the utterance. Data indicative of the first task, performed by the computer system, is compared to data indicative of a second task, performed by the human in response to the speech utterance. Based on the comparison, the set of internal representations may be updated or modified to improve the speech recognition performance and capabilities of the speech recognition system.
1. A method for modifying a computer-based speech recognition system, comprising:
receiving, by the computer-based speech recognition system, a speech utterance, the computer-based speech recognition system comprising at least one processor and at least one memory device;
processing the speech utterance to identify a first task for the speech utterance, the processing being performed without human involvement and with the computer-based speech recognition system using a set of internal representations for the computer-based speech recognition system, the set of internal representations comprising one or more parameters for recognizing speech in the speech utterance;
performing, by the computer-based speech recognition system, the first task in response to the computer-based speech recognition system identifying the first task;
providing the speech utterance to a human and refraining from providing information to the human that identifies the first task;
receiving input from the human that identifies a second task for the speech utterance;
comparing the first task performed by the computer-based speech recognition system to the second task identified by the human; and
based at least in part on the comparison, modifying the set of internal representations used by the computer-based speech recognition system in an event that the first task differs from the second task.
2. The method of claim 1 , wherein the speech utterance is received as part of a telephone call.
3. The method of claim 1 , wherein
the internal representations include a parameter of an acoustic model that is used by the computer-based speech recognition system.
4. The method of claim 1 , wherein
the internal representations include a parameter of a language model that is used by the computer-based speech recognition system.
5. The method of claim 1 , wherein the providing the speech utterance to the human comprises providing a recording of the speech utterance to the human.
6. The method of claim 5 , wherein:
the first task is a machine transcription of the speech utterance; and
the second task is a human transcription of the speech utterance.
7. The method of claim 1 , wherein at least one of the first task or the second task comprises at least one of accessing account information, accessing flight information, accessing an employee directory, opening a file, or inputting GPS information.
8. A computer-based speech recognition system, comprising:
at least one processor; and
at least one memory device storing instructions that when executed by the at least one processor cause the at least one processor to perform the acts comprising:
processing a speech utterance to identify a first task for the speech utterance, the processing using a set of internal representations for the computer-based speech recognition system, the set of internal representations comprising one or more parameters for recognizing speech;
performing the first task that is identified for the speech utterance;
providing the speech utterance to a human and refraining from providing information to the human that identifies the first task;
receiving input from the human that identifies a second task for the speech utterance;
comparing the first task identified by the computer-based speech recognition system to the second task identified by the human; and
modifying the set of internal representations of the computer-based speech recognition system in an event that the comparison indicates that the first task differs from the second task.
9. The computer-based speech recognition system of claim 8 , wherein the acts further comprise receiving the speech utterance as part of a telephone call.
10. The computer-based speech recognition system of claim 8 , wherein:
the computer-based speech recognition system comprises a speech recognition module; and
the internal representations include a parameter of an acoustic model of the speech recognition module.
11. The computer-based speech recognition system of claim 8 , wherein:
the computer-based speech recognition system comprises a speech recognition module; and
the internal representations include a parameter of a language model of the speech recognition module.
12. The computer-based speech recognition system of claim 8 , wherein:
the first task is a machine transcription of the speech utterance; and
the second task is a human transcription of the speech utterance.
13. The computer-based speech recognition system of claim 8 , wherein at least one of the first task or the second task comprises at least one of accessing account information, accessing flight information, accessing an employee directory, opening a file, or inputting GPS information.
14. A non-transitory computer readable storage medium having stored thereon instructions that when executed by a processor cause the processor to perform the acts comprising:
processing a speech utterance to identify a first task for the speech utterance, the processing using a set of internal representations, the set of internal representations comprising one or more parameters for recognizing speech;
performing the first task that is identified for the speech utterance;
outputting the speech utterance to a human;
after outputting the speech utterance to the human, receiving input from the human that identifies a second task for the speech utterance;
comparing the first task to the second task identified by the human; and
modifying the set of internal representations in an event that the comparison indicates that the first task differs from the second task.
15. The non-transitory computer readable storage medium of claim 14 , wherein the acts further comprise receiving the speech utterance as part of a telephone call.
16. The non-transitory computer readable storage medium of claim 14 , wherein the internal representations include a parameter of an acoustic model for recognizing speech.
17. The non-transitory computer readable storage medium of claim 14 , wherein the internal representations include a parameter of a language model for recognizing speech.
18. The non-transitory computer readable storage medium of claim 14 , wherein at least one of the first task or the second task comprises at least one of accessing account information, accessing flight information, accessing an employee directory, opening a file, or inputting GPS information.