Electronic apparatus for real-time conversation interpretation and method for controlling the same

Patent metadata

Field	Value
Publication number	US-12561535-B2
Application number	US-202318221190-A
Country	US
Kind code	B2
Filing date	Jul 12, 2023
Priority date	Oct 15, 2021
Publication date	Feb 24, 2026
Grant date	Feb 24, 2026

Abstract

Provided is an electronic apparatus that includes a microphone; a communication interface; a memory storing information on a first encoder, a first decoder, a first interpretation model comprising a first version conversion module, and at least one first version which the first conversion module may convert; and a processor. The processor is configured to communicate with an external device and receive information about at least one second version corresponding to a second interpretation model in the external device; obtain information on a compatible version based the at least one first version and the least one second version; based on a user voice, obtain a first feature vector corresponding to the user voice using the first encoder, convert the first feature vector to a second feature vector corresponding to the compatible version, and transmit the second feature vector to the external device.

First claim

What is claimed is:

1 . An electronic apparatus comprising:
a microphone; a communication interface including communication circuitry; a memory storing first information associated with each of: a first encoder, a first decoder, a first interpretation model including a first version conversion module, and at least one first version which the first version conversion module is able to convert, wherein a version of the first encoder and a version of the first decoder is a highest version among the at least one first version; and a processor configured to:
communicate, via the communication interface, with an external device and receive, from the external device, second information associated with at least one second version which a second version conversion module included in a second interpretation model stored in the external device is able to convert,
obtain compatibility information associated with a compatible version based on the at least one first version and the least one second version and store the compatibility information in the memory,
based on a user voice being obtained through the microphone, obtain a first feature vector corresponding to the user voice using the first encoder,
convert the first feature vector to a second feature vector corresponding to the compatible version using the first version conversion module, and
transmit, via the communication interface, the second feature vector to the external device.

2 . The electronic apparatus of claim 1 , wherein the processor is further configured to identify a highest version included in both the at least one first version and the at least one second version as the compatible version.

3 . The electronic apparatus of claim 1 , wherein, based on the highest version among the at least one first version being lower than a highest version among the at least one second version, a version of the first feature vector and a version of the second feature vector are identical.

4 . The electronic apparatus of claim 1 , wherein the processor is further configured to:
receive, from the external device, a third feature vector corresponding to a voice of a second user obtained by the external device and is based on the compatible version, determine whether the version of the first decoder is identical with the compatible version, and obtain a text corresponding to the voice of the second user based on the determination result.

5 . The electronic apparatus of claim 4 , wherein the processor is further configured to:
based on the version of the first decoder being identical with the compatible version, obtain a text corresponding to the third feature vector using the first decoder, and based on the version of the first decoder being different from the compatible version, convert the third feature vector to a fourth feature vector corresponding to the version of the first decoder using the first version conversion module, and obtain a text corresponding to the fourth feature vector using the first decoder.

6 . The electronic apparatus of claim 1 , wherein the processor is further configured to:
determine compatibility with the external device based on a comparison of the at least one first version and the at least one second version, and based on determining that the electronic apparatus is not compatible with the external device, output a message requesting update of the first interpretation model.

7 . The electronic apparatus of claim 1 , wherein the processor is further configured to:
based on communication with the external device, receive identification information of the external device, and match the identification information of the external device with the compatibility information associated with the compatible version and store the information in the memory.

8 . A method for controlling an electronic apparatus, the method comprising:
storing first information associated with each of: a first encoder, a first decoder, a first interpretation model including a first version conversion module, and at least one first version which the first version conversion module is able to convert, wherein a version of the first encoder and a version of the first decoder is a highest version among the at least one first version; communicating with an external device and receiving, from the external device, second information about at least one second version which a second version conversion module included in a second interpretation model stored in the external device is able to convert; obtaining compatibility information associated with a compatible version based on the at least one first version and the least one second version and storing the compatibility information; based on a user voice being obtained, obtaining a first feature vector corresponding to the user voice using the first encoder; converting the first feature vector to a second feature vector corresponding to the compatible version using the first version conversion module; and transmitting the second feature vector to the external device.

9 . The method of claim 8 , wherein the obtaining the compatibility information associated with the compatible version comprises:
identifying a highest version included in both the at least one first version and the at least one second version as the compatible version.

10 . The method of claim 8 , wherein, based on the highest version among the at least one first version being lower than a highest version among the at least one second version, a version of the first feature vector and a version of the second feature vector are identical.

11 . The method of claim 8 , further comprising:
receiving, from the external device, a third feature vector corresponding to a voice of a second user obtained by the external device and is based on the compatible version; determining whether the version of the first decoder is identical with the compatible version; and obtaining a text corresponding to the voice of the second user based on the determination result.

12 . The method of claim 11 , wherein the obtaining the text comprises:
based on the version of the first decoder being identical with the compatible version, obtaining a text corresponding to the third feature vector using the first decoder; and based on the version of the first decoder being different from the compatible version, converting the third feature vector to a fourth feature vector corresponding to the version of the first decoder using the first version conversion module, and obtaining a text corresponding to the fourth feature vector using the first decoder.

13 . The method of claim 8 , further comprising:
determining compatibility with the external device based on a comparison of the at least one first version and the at least one second version; and based on determining that the electronic apparatus is not compatible with the external device, outputting a message requesting update of the first interpretation model.

14 . A non-transitory computer readable medium for storing computer readable program code or instructions for carrying out operations, when executed by a processor, for providing real-time interpretation during a conversation, the operations comprising:
storing first information associated with each of: a first encoder, a first decoder, a first interpretation model including a first version conversion module, and at least one first version which the first version conversion module is able to convert, wherein a version of the first encoder and a version of the first decoder is a highest version among the at least one first version; communicating with

Assignees

• Samsung Electronics Co Ltd

Inventors

• Park Soyoon
• Lim Sungjun
• Kim Jonghyun
• Kim Jiwan
• Kim Hakjung
• Kim Hyunkyung
• Park Sohyun
• Lee Indong

Classifications

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• G06F40/44

  Statistical methods, e.g. probability models · CPC title

• H04M1/65

  Recording arrangements {for recording a message from the calling party (in the exchange H04M3/50)} · CPC title

• H04M1/60

  including speech amplifiers · CPC title

• G10L15/26

  Speech to text systems (G10L15/08 takes precedence) · CPC title