Data processing method based on simultaneous interpretation, computer device, and storage medium

Patent metadata

Field	Value
Publication number	US-12087290-B2
Application number	US-202016941503-A
Country	US
Kind code	B2
Filing date	Jul 28, 2020
Priority date	May 10, 2018
Publication date	Sep 10, 2024
Grant date	Sep 10, 2024

Abstract

A data processing method based on simultaneous interpretation, applied to a server in a simultaneous interpretation system, including: obtaining audio transmitted by a simultaneous interpretation device; processing the audio by using a simultaneous interpretation model to obtain an initial text; transmitting the initial text to a user terminal; receiving a modified text fed back by the user terminal, the modified text being obtained after the user terminal modifies the initial text; and updating the simultaneous interpretation model according to the initial text and the modified text.

First claim

What is claimed is:

1. A data processing method, applied to a server in an interpretation system, the method comprising:
obtaining audio transmitted by an interpretation device;
processing the audio by using an interpretation model to obtain an initial text;
transmitting the initial text to a first user terminal and a second user terminal to be respectively displayed on a first interpretation auxiliary page and a second interpretation auxiliary page;
receiving a first modified text fed back by the first user terminal and a second modified text fed back by the second user terminal, the first modified text being obtained after the first user terminal modifies the initial text based on the first interpretation auxiliary page, and the second modified text being obtained after the second user terminal modifies the initial text based on the second interpretation auxiliary page;
determining a weighted cumulative value S based on a weight q1 corresponding to an identifier of the first user terminal, a weight q2 corresponding to an identifier of the second user terminal, a number of times t1 of text modification corresponding to the first user terminal, and a number of times t2 of text modification corresponding to the second user terminal, wherein the weighted cumulative value S includes a sum of t1 multiplies q1 and t2 multiplies q2;
determining the weighted cumulative value S is greater than a threshold;
in response to determining the weighted cumulative value S being greater than the threshold, updating the interpretation model according to the initial text, the first modified text, and the second modified text;
generating a modified text based on the updated interpretation model, and automatically assigning a sequence number to the modified text to be the same as a sequence number of the initial text; and
sending the modified text and the corresponding sequence number to a user terminal;
wherein the sequence number indicates an arrangement or a storage position of the initial text, for the user terminal to locally search and replace the initial text with the modified text.

2. The method according to claim 1 , wherein the processing the audio comprises:
performing noise reduction processing on the audio; and
using the interpretation model to obtain the initial text after performing the noise reduction processing.

3. The method according to claim 1 , wherein the interpretation model comprises a universal voice model and an auxiliary voice model, and processing the audio comprises:
performing, using the universal voice model, speech recognition on the audio to obtain a recognized text; and
updating the recognized text by using the auxiliary voice model to obtain a recognition update text, the initial text comprising at least one of the recognized text and the recognition update text; and
updating the interpretation model according to the initial text and the modified text comprises: updating the auxiliary voice model according to the initial text and the modified text.

4. The method according to claim 1 , wherein the interpretation model comprises a translation model, the initial text comprises a translation text, and the modified text comprises a modified translation text; and the updating the interpretation model comprises:
updating the translation model according to the translation text and the modified translation text.

5. The method according to claim 1 , wherein the method further comprises:
receiving a video transmitted by the interpretation device and corresponds to the audio; and
embedding the initial text into the video; and the transmitting the initial text to the user terminal comprises:
transmitting the video embedded with the initial text to the user terminal.

6. The method according to claim 1 , wherein the audio corresponds to a group identifier; the transmitting the initial text comprises:
transmitting the initial text to the user terminal accessing via the group identifier.

7. The method according to claim 1 , wherein the audio corresponds to a group identifier; and the method further comprises: storing the initial text and the group identifier.

8. The method according to claim 1 , further comprising:
obtaining a voice component from the audio;
obtaining, from the voice component, an audio component whose energy value is greater than or equal to an energy threshold; and
processing the audio component by using the interpretation model to obtain the initial text.

9. The method according to claim 1 , wherein the first user terminal and the second user terminal both correspond to a conference number, and the method further comprises:
creating a document;
adding the initial text to the document; and
establishing a mapping relationship between the document and the conference number.

10. The method according to claim 1 , further comprising:
detecting accuracy rates of the text modifications;
increasing the weight q1 in response to determining that the accuracy rate of the text modification corresponding to the first terminal reaches a modification accuracy threshold and that the number of times t1 of the text modification corresponding to the first terminal reaches the modification times threshold; and
increasing the weight q2 in response to determining that the accuracy rate of the text modification corresponding to the second terminal reaches the modification accuracy threshold and that the number of times t2 of the text modification corresponding to the second terminal reaches the modification times threshold.

11. The method according to claim 1 , wherein the number of times t1 that the first user terminal performs the text modification is an integer greater than 1, and each time of the t1 times that the text modification performed by the first user terminal contributes to the weighted cumulative value S is the same weight q1.

12. The method according to claim 1 , further comprising:
transmitting, to the first user terminal, a simultaneous interpretation auxiliary page configuration file corresponding to a child application identifier of a child application, the child application being implemented in an environment provided by a parent application run on the first user terminal, wherein the simultaneous interpretation auxiliary page configuration file is for configuring the first interpretation auxiliary page presented by the child application on the first user terminal.

13. A computer device, comprising a memory and a processor, the memory storing a computer program, the computer program, when executed by the processor, causing the processor to perform:
transmitting the initial text to a first user terminal and a second user terminal to be respectively displayed on a first interpretation auxiliary page and a second interpretation auxiliary page;
receiving a first modified text fed back by the first user terminal and a second modified text fed back by the second user terminal, the first modified text being obtained after the first user terminal modifies the initial text based on the first interpretation auxiliary page, and the second modified text being obtained after the second user terminal modifies the initial text based on the second interpretation auxiliary page;
determining a weighted cumulative value S based on a weight q1 corresponding to an identifier of the first user terminal, a weight q2 corresponding to an identifier of the second user terminal, a number of times t1 of text modification corresponding to the first user terminal, and a number of times t2 of text modification corresponding to the second user terminal, wherein the weighted cumulative value S includes a sum of t1 multiplies q1 and t2

Assignees

• Tencent Tech Shenzhen Co Ltd

Inventors

• Bai Jingliang
• Ouyang Caisheng
• Liu Haikang
• Chen Lianwu
• Chen Qi
• Zhang Yulu
• Luo Min
• Su Dan

Classifications

• G10L2025/783

  based on threshold decision · CPC title

• G10L2015/0636

  Threshold criteria for the updating · CPC title

• G10L25/84

  for discriminating voice from noise · CPC title

• G10L25/21

  the extracted parameters being power information · CPC title

• G10L21/0232

  Processing in the frequency domain · CPC title