Automatic error correction method for real-time court hearing speech recognition, storage medium and computing device

Abstract

Provided are a computer-executed automatic error correction method for real-time court hearing speech recognition, a computer-readable storage medium, and a computing device. The method includes: discovering hot words and adding them to a hot word library in real-time in response to detecting a modification operation of a character recognized by a user in a courtroom system; carrying out real-timespeech recognition; and performing real-time hot word matching operation on the recognized text and performing hot word replacement operation. The court hearing speech recognition system according tothe embodiment of the present invention realizes the self-learning function of the court hearing system by automatically discovering hot words and constructing and updating a hot word library in realtime during the running process. Through ts_code technology in hot-word storage and hot-word substitution, different settings can be generated automatically for different accents through fuzzy soundconfiguration, so that the problem of speech recognition errors caused by multiple local accents can be easily and efficiently corrected.

Description

technical field [0001] The present invention generally relates to speech recognition technology in court, in particular to a computer-executed real-time trial speech recognition automatic error correction method, storage medium and computing device. Background technique [0002] With the technological breakthrough of artificial intelligence, more and more artificial intelligence-related products are applied in people's lives. The trial speech recognition system is an application in the legal industry, which involves natural speech processing. A hybrid application of speech recognition technology. [0003] Natural language processing is an important direction in the field of computer science and artificial intelligence. It studies various theories and methods that can realize effective communication between humans and computers using natural language. Natural speech processing involves technologies such as automatic word segmentation, part-of-speech tagging, syntactic analy...

AI Technical Summary

Problems solved by technology

[0008] (1) Users are required to manually set hot words, increasing user workload;

[0009] (2) It has a certain delay, and hot words need to wait for a while to take effect;

[0010] (3) It is impossible to solve the local dialect problem. If the user just cannot pronounce some words correctly, then the existing technical solutions will be completely ineffective;

[0011] (4) There is a certain limit on the number of hot words. If too many hot words are set, the original model of the speech recognition engine may be broken. Therefore, the setting of hot words in the existing speech recognition engine has certain restrictions on the length and quantity of hot words;

[0012] (5) Automatic typesetting is not possible, for example, when the Contract Law of the People's Republic of China is recognized, punctuation marks cannot be automatically added

Images