Transcription generation technique selection

Patent metadata

Field	Value
Publication number	US-12266366-B2
Application number	US-202318457184-A
Country	US
Kind code	B2
Filing date	Aug 28, 2023
Priority date	May 27, 2020
Publication date	Apr 1, 2025
Grant date	Apr 1, 2025

Abstract

A method to transcribe communications may include selecting a first transcription generation technique from among multiple transcription generation techniques for generating transcriptions of audio of one or more communication sessions that involve a user device and obtaining performances of the multiple transcription generation techniques with respect to generating the transcriptions of the audio. The method may also include monitoring comparisons between the performances of the multiple transcription generation techniques and obtaining input from the user with respect to the comparisons. The method may further include selecting a second transcription generation technique from among the multiple transcription generation techniques based on the input from the user.

First claim

We claim:

1. A method to transcribe communications, the method comprising:
obtaining a performance of at least one of a plurality of transcription generation techniques with respect to generating transcriptions of audio;
determining a report based on the performance of the at least one of the plurality of transcription generation techniques, wherein the performance of the at least one of the plurality of transcription generation techniques is based on one or more of the following: transcription accuracy and transcription latency;
directing the report to a device that obtains transcriptions of a first audio session involving the device using one of the plurality of transcription generation techniques, the performance of the at least one of the plurality of transcription generation techniques is based on the transcriptions of the first audio session;
after directing the report, obtaining an indication from the device; and
selecting, based on the indication from the device, another one of the plurality of transcription generation techniques to generate transcriptions of a future audio session involving the device that occurs after the first audio session.

2. The method of claim 1 , wherein the report includes a recommendation for the other one of the plurality of transcription generation techniques and the indication includes a selection of the other one of the plurality of transcription generation techniques.

3. The method of claim 2 , further comprising determining the recommendation based on a cost associated with the one of the plurality of transcription generation techniques and a cost associated with the other one of the plurality of transcription generation techniques.

4. The method of claim 1 , wherein the steps of directing the report and of obtaining the indication occur during the first audio session.

5. The method of claim 1 , further comprising:
obtaining a performance of the other one of the plurality of transcription generation techniques with respect to generating transcriptions of audio; and
comparing the performance of the other one of the plurality of transcription generation techniques and the performance of the at least one of the plurality of transcription generation techniques,
wherein the report is based on the comparison.

6. The method of claim 5 , wherein comparing the performance of the other one of the plurality of transcription generation techniques and the performance of the at least one of the plurality of transcription generation techniques includes comparing two or more aspects of the performance of the other one of the plurality of transcription generation techniques with two or more aspects of the performance of the at least one of the plurality of transcription generation techniques,
wherein the report includes aspects of the performance of the other one of the plurality of transcription generation techniques that are better than aspects of the performance of the at least one of the plurality of transcription generation techniques.

7. The method of claim 1 , wherein one of the plurality of transcription generation techniques includes a revoicing of audio before transcription generation.

8. At least one non-transitory computer-readable media configured to store one or more instructions that when executed by at least one processor cause or direct a system to perform the method of claim 1 .

9. A method to transcribe communications, the method comprising:
selecting a first transcription generation technique from among a plurality of transcription generation techniques for generating transcriptions of audio obtained by a device during a first audio session;
obtaining a performance of at least one of the plurality of transcription generation techniques with respect to generating transcriptions;
after presentation of the transcriptions of the audio by the device, obtaining input from a user of the device regarding the performance of at least one of the plurality of transcription generation techniques with respect to generating transcriptions of the first audio session; and
selecting a second transcription generation technique from among the plurality of transcription generation techniques in response to the input from the device for a second audio session that occurs after the first audio session, the second transcription generation technique being used for generating transcriptions of second audio obtained by the device during the second audio session.

10. The method of claim 9 , wherein the performance of the at least one of the plurality of transcription generation techniques is based on one or more of: transcription accuracy and transcription latency.

11. The method of claim 9 , wherein the performance of the at least one of the plurality of transcription generation techniques is measured without consideration of the transcriptions generated during the first audio session.

12. The method of claim 9 , wherein the selection of the first transcription generation technique occurs before the second audio session begins.

13. The method of claim 9 , wherein the performance of the at least one of the plurality of transcription generation techniques is the performance of the first transcription generation technique, the method further comprising:
obtaining performance of the second transcription generation technique with respect to generating transcriptions; and
before obtaining the input from the device, providing a report to the device based on a comparison between the performance of the first transcription generation technique and the performance of the second transcription generation technique.

14. The method of claim 13 , further comprising:
determining an aspect of the performance of the second transcription generation technique that is better than a same aspect of the performance of the first transcription generation technique; and
generating the report to include the aspect of the performance of the second transcription generation technique.

15. The method of claim 13 , wherein the second transcription generation technique does not generate a transcription of audio such that the performance of the second transcription generation technique is an estimated performance.

16. The method of claim 9 , wherein the performance of at least one of the plurality of transcription generation techniques is the performance of the second transcription generation technique.

17. At least one non-transitory computer-readable media configured to store one or more instructions that when executed by at least one processor cause or direct a system to perform the method of claim 9 .

18. A system comprising:
one or more processors; and
one or more non-transitory computer-readable mediums configured to store instructions that when executed by the processors cause or direct the system to perform operations, the operations comprising:
obtaining a performance of at least one of a plurality of transcription generation techniques with respect to generating transcriptions of audio;
determining a report based on the performance of the at least one of the plurality of transcription generation techniques, wherein the performance of the at least one of the plurality of transcription generation techniques is based on one or more of the following: transcription accuracy and transcription latency;
directing the report to a device that obtains transcriptions of a first audio session involving the device using one of the plurality of transcription generation techniques, the performance of the at least one of the plurality of transcription generation te

Assignees

• Sorenson Ip Holdings Llc

Inventors

• Thomson David

Classifications

• G06F18/2178

  based on feedback of a supervisor · CPC title

• G10L21/10

  Transforming into visible information · CPC title

• G10L15/30

  Distributed recognition, e.g. in client-server systems, for mobile phones or network applications · CPC title

• G10L15/01

  Assessment or evaluation of speech recognition systems · CPC title

• G10L15/32

  Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems · CPC title