Filtering transcriptions of utterances

• US 8,498,872 B2
• Filed: 09/15/2012
• Issued: 07/30/2013
• Est. Priority Date: 04/05/2006
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Canyon IP Holdings LLC (Intellectual Ventures LLC)
• Inventors: White, Marc, Strohofer, Cliff
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US13/621,194
• Publication Number:
  US 20130018656A1
• Time in Patent Office: 318 Days
• Field of Search: 704/270.1, 455/412.1
• US Class Current: 704/270.1
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

Audio data that includes speech may be transcribed by a speech recognition engine to generate speech recognition results, such as a transcription. One or more filters may be selected and applied to the speech recognition results to generate filtered speech recognition results. The one or more filters may be selected based at least in part on a characteristic of the speech recognition results, a characteristic of the audio data, or any other characteristic.

Claims

1. A computer-implemented method, the computer-implemented method comprising:

under control of one or more computing devices configured with specific computer-executable instructions,

obtaining audio data comprising speech from a first device;

transcribing the speech using an automatic speech recognition engine having a statistical language model to generate a transcription comprising one or more words;

determining a characteristic related to the audio data, wherein the characteristic is not determined from the one or more words of the transcription;

selecting a filter for the transcription, wherein the filter is selected based at least in part on the characteristic; and

applying the selected filter to the transcription to generate a filtered transcription.

2. The computer-implemented method of claim 1 further comprising determining a location of the first device, and wherein the characteristic related to the audio data comprises the location of the first device.

3. The computer-implemented method of claim 1, wherein the audio data comprises a message from the first device to a second device.

4. The computer-implemented method of claim 3 further comprising determining a location of the second device, and wherein the characteristic related to the audio data comprises the location of the second device.

5. The computer-implemented method of claim 4, wherein the selected filter comprises a locale filter associated with the location of the second device.

6. The computer-implemented method of claim 3 further comprising providing the filtered transcription to the second device.

7. The computer-implemented method of claim 1, wherein determining the characteristic related to the audio data comprises receiving a selection of the characteristic from the first device.

8. A system comprising:

an electronic data store configured to store;

a statistical language model; and

one or more filters; and

a computing device in communication with the electronic data store, the computing device being configured to;

obtain audio data from a first device;

perform automatic speech recognition on the audio data with the statistical language model to generate speech recognition results;

determine a first characteristic associated with at least one of the audio data or the speech recognition results;

select a first filter based at least in part on the first characteristic; and

apply the first filter to the speech recognition results to produce first filtered speech recognition results.

9. The system of claim 8, wherein:

the computing device is configured to determine a name of a user of the first device;

the first characteristic comprises the name of the user of the first device; and

the first filter comprises a name filter corresponding to the name of the user of the first device.

10. The system of claim 8, wherein at least one of the audio data or the speech recognition results comprises a message from the first device to a second device.

11. The system of claim 10, wherein:

the computing device is configured to determine a type of a user of the second device;

the first characteristic comprises the type of the user of the second device; and

the first filter comprises a type filter corresponding to the type of the user of the first device.

12. The system of claim 10, wherein:

the computing device is further configured to determine a type of the message; and

the first characteristic comprises the type of the message.

13. The system of claim 12, wherein the type of the message comprises at least one of a text message, an email message, an instant message, or a voicemail.

14. The system of claim 12, wherein the computing device is configured to determine the type of the message by receiving an indication of the type of the message from the first device.

15. The system of claim 10, wherein the computing device is further configured to transmit the message to the second device.

16. The system of claim 15, wherein:

the first filter comprises a screening filter associated with a screening word;

the computing device is further configured to determine that the message includes the screening word; and

the computing device only transmits the message to the second device if the message includes the screening word.

17. The system of claim 15, wherein:

the first filter comprises a screening filter associated with a screening word;

the computing device is further configured to determine that the message does not include the screening word; and

the computing device only transmits the message to the second device if the message does not include the screening word.

18. The system of claim 15, wherein:

the first filter comprises a priority filter associated with a priority word;

the computing device is further configured to determine that the message includes the priority word; and

the computing device is further configured to transmit an indicator of high priority to the second device with the message.

19. A non-transitory computer-readable medium having stored thereon a computer-executable module configured to execute in one or more processors, the computer-executable module being further configured to:

receive, from a first device, audio data comprising speech of a user;

perform automatic speech recognition on the audio data with a statistical language model to generate speech recognition results;

select a first filter based at least in part on a first characteristic, wherein the first characteristic is not determined from the speech recognition results; and

apply the first filter to the speech recognition results to generate first filtered speech recognition results.

20. The non-transitory computer-readable medium of claim 19, wherein the computer-executable component is further configured to determine a designation of the first device, and wherein:

the first characteristic comprises the designation of the first device; and

the first filter comprises a designation filter corresponding to the designation of the first device.

21. The non-transitory computer-readable medium of claim 20, wherein the designation of the first device comprises a phone number of the first device.

22. The non-transitory computer-readable medium of claim 19, wherein the computer-executable component is further configured to:

select a second filter based at least in part on a second characteristic; and

apply the second filter to the first filtered speech recognition results to generate second filtered speech recognition results.

23. The non-transitory computer-readable medium of claim 19, wherein:

the audio data comprises a message from the first device to a second device; and

the first characteristic corresponds to at least one of a type of the message, a user of the second device, or a type of the user of the second device.

24. The non-transitory computer-readable medium of claim 19, wherein the first filter comprises a finite grammar.