Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A kind of speech recognition error correction method and man-machine dialogue system

A technology of speech recognition and error correction method, which is applied in the field of speech recognition error correction, which can solve problems such as the inability to directly modify the speech recognition model, and achieve the effects of improving the quality of man-machine dialogue, reducing the impact of jumps, and improving accuracy

Active Publication Date: 2022-05-03
CHONGQING COLLEGE OF ELECTRONICS ENG
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in specific interactive scenarios, the speech recognition technology of Internet giants still has great shortcomings, especially short text recognition under a large amount of environmental noise
At this time, since the speech recognition model cannot be directly modified, we can only focus on mapping the speech recognition results to the scene-related text

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 kind of speech recognition error correction method and man-machine dialogue system
  • A kind of speech recognition error correction method and man-machine dialogue system
  • A kind of speech recognition error correction method and man-machine dialogue system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] such as Figure 1 As shown, a speech recognition error correction method includes the following steps:

[0035] Acquiring a speech recognition result and candidate semantic segments, and pinyin of the speech recognition result and pinyin of the candidate semantic segments;

[0036] Combining the speech recognition result with the candidate semantic segments to form several new error-correcting texts, calculating the phonetic similarity between the speech recognition result and the candidate semantic segments and the combination score of the error-correcting texts, and screening the error-correcting texts according to the combination score to obtain an error-correcting candidate list;

[0037] Language model optimization: calculating language model scores by using language models, and generating a final error correction list according to the combined scores and the language model scores.

[0038] Among them, calculating the phonetic similarity score of each semantic text spec...

example 1

[0069] Robot: Sir, today is the repayment date of your car loan. Please repay it on time.

[0070] Userq: I'm already pregnant.

[0071] Corrections 0: I have already returned it.

[0072] Shot: Repayment.

example 2

[0074] Robot: Is it convenient for you now?

[0075] Userq: I can measure it again.

[0076] Corrections 0: I'm driving

[0077] Shot: inconvenient

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

The present invention relates to the technical field of speech recognition error correction, in particular to a speech recognition error correction method and a man-machine dialogue system, the method comprising the following steps: acquiring speech recognition results and candidate semantic segments, as well as pinyin and candidate semantic segments of the speech recognition results Pinyin; combine speech recognition results and candidate semantic fragments to form several new error-correcting texts, calculate the combination score of the error-correcting text, and filter the error-correcting text according to the combination score to obtain a list of error-correcting candidates; the language model optimization step uses the language model Compute the language model score and generate a final error correction list based on the combined score and the language model score. The invention provides a voice recognition error correction method and a man-machine dialogue system, which can more accurately match the error correction text after a voice recognition error, and effectively reduce the impact of the voice recognition result on the Chinese man-machine dialogue flow. , improve the quality of man-machine dialogue.

Description

Technical field [0001] The invention relates to the technical field of speech recognition error correction, in particular to a speech recognition error correction method and a man-machine dialogue system. technical background [0002] The research of man-machine dialogue technology can be traced back to the 1960s. Since alan turing put forward the idea of testing whether a machine has human intelligence through Turing test, researchers have been devoted to the research of man-machine dialogue system. At present, there are many human-machine dialogue products on the market, such as intelligent voice assistants, telephone robots, etc. Among these products, speech recognition technology is a very important module. Because of the user's nonstandard expression, dialect, environmental noise and other factors, it leads to wrong speech recognition results, which is the main reason for the failure of man-machine dialogue. [0003] The research of speech recognition in China started in 195...

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): G10L15/18G10L15/22G10L15/26G10L15/04G06F40/232
CPCG10L15/1815G10L15/1822G10L15/22G10L15/26G10L15/04
Inventor 兰飞
Owner CHONGQING COLLEGE OF ELECTRONICS ENG
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com