Simultaneous interpretation processing method, device and equipment

Abstract

The present invention proposes a simultaneous interpretation processing method, device and equipment, wherein the method includes: acquiring the speech recognition result of the first language, and performing word segmentation processing on the speech recognition result to generate word segmentation results; receiving the word segmentation results in sequence , when the number of received word segmentation results is equal to the preset value K, input the received word segmentation results into the pre-trained translation model from the first language text to the second language text for processing, and decode the first A translation result, where K is a positive integer; receive the Pth word segmentation result, input the Pth word segmentation result into the translation model for processing, and decode the P‑K+1th translation result, where P greater than K and P is an integer. As a result, the translation delay of simultaneous interpretation is reduced, and the delay is controlled at the word level.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a simultaneous interpretation processing method, device and equipment. Background technique [0002] At present, simultaneous interpretation is widely used in various conferences, media activities, broadcasting lectures and other fields. Simultaneous interpretation is a major challenge to speech recognition technology and machine translation technology. To evaluate the effect of simultaneous interpretation needs to comprehensively consider the delay and accuracy of translation. [0003] The simultaneous interpretation system in the related art usually invokes translation based on the sentence segmentation and punctuation results of the speech recognition result, resulting in a delay of one or more clauses in the translation result compared with the speech recognition result, and the delay is relatively high. Contents of the invention [0004] For this r...

AI Technical Summary

Problems solved by technology

[0004] For this reason, the first purpose of the present invention is to propose a simultaneous interpretation processing method, by cutting the speech recognition results into words, and controlling the translation delay of K words, which solves the problem of simultaneous interpretation translation delay in the related art. High problem, reducing the translation delay of simultaneous interpretation, and controlling the delay at the word level

Images