Unlock instant, AI-driven research and patent intelligence for your innovation.

A text error correction method after speech conversion

A text error correction and speech conversion technology, which can be used in speech analysis, speech recognition, text database query, etc., and can solve the problem of low speech recognition accuracy.

Active Publication Date: 2021-09-07
AVIC HUADONG OPTOELECTRONICS (SHANGHAI) CO LTD
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problem that the accuracy rate of existing speech recognition is not high

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A text error correction method after speech conversion
  • A text error correction method after speech conversion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] refer to figure 1 with figure 2 , a text error correction method after voice conversion of the present embodiment, comprising the following steps:

[0038] S100, input voice information, perform voice feature extraction to obtain LPCC (Linear Prediction Cepstral Coefficient), Mel cepstral coefficient, output feature vector;

[0039] S200. Input the feature vector into the HMM (Hidden Markov Model) acoustic model, and output the phoneme information;

[0040] S300. Carry out tone-by-tone scanning in the form of single tone, two tone, three tone, and four tone, each word is first matched with the network thesaurus T1, if the match is successful, then output the text, if the match fails, then match with Standard thesaurus T2 for matching;

[0041] S400. Match each word with the thesaurus T2, if the matching is successful, output the text result, if the matching fails, continue to step S500;

[0042] S500. Judging the pronunciation of the word that fails to match, scann...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A text error correction method after voice conversion of the present invention comprises the following steps: S100, input voice information, output feature vector; S200, input feature vector into HMM acoustic model, output to obtain phoneme information; S300, after performing tone-by-tone scanning Match with the network thesaurus T1, if the match is successful, then output the text, if the match fails, then match with the standard thesaurus T2; S400, match each word with the thesaurus T2, if the match is successful, then output the text result, If the matching fails, continue to step S500; S500, judge the pronunciation of the word that fails to match, scan the word word by word and convert the approximate sound; S600, repeat steps S300 and S400 for the word after conversion. In the speech recognition text matching process, the words that fail to be matched are replaced by pronunciation to form new words to correct the speech recognition text conversion error caused by the accent.

Description

technical field [0001] The invention belongs to the technical field of speech recognition conversion, in particular to a text error correction method after speech conversion. Background technique [0002] Talking to a machine so it can understand what you say has been a long-held dream. The image of the China Internet of Things School-Enterprise Alliance compares speech recognition to the auditory system of a machine. Speech recognition technology is a technology that allows machines to convert voice signals into corresponding text or commands through the process of recognition and understanding. [0003] Speech recognition technology mainly includes three aspects: feature extraction technology, pattern matching criteria and model training technology. The voice recognition technology of the Internet of Vehicles has also been fully cited. For example, it is safe and convenient to set the destination and direct navigation just by dictating. [0004] Speech recognition is an...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06F40/284G06F40/30G06F16/33G06K9/62G10L15/02G10L15/14G10L15/26G10L25/24
CPCG06F40/30G06F40/284G06F16/3344G10L15/02G10L15/144G10L15/26G10L25/24G10L2015/025G06F18/295G06F18/214
Inventor 徐涵许召辉马翼平陈年生范光宇饶蕾章弘凯
Owner AVIC HUADONG OPTOELECTRONICS (SHANGHAI) CO LTD