Method and device for detecting keywords in speech

A keyword and voice technology, applied in the field of audio processing, can solve problems such as multiple redundant connection arcs, complex grid structure, and high keyword missed detection rate, so as to improve search efficiency, reduce missed detection rate, and occupy a small space Effect

Inactive Publication Date: 2016-09-07
BEIJING JETSEN TECH
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the complex structure of the grid, there are many redundant connection arcs that are not conducive to data processing and storage; therefore, the missed detection rate of keywords is relatively 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
  • Method and device for detecting keywords in speech
  • Method and device for detecting keywords in speech
  • Method and device for detecting keywords in speech

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0045] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0046] The flow chart of the method schematic diagram for a kind of multi-screen synchronous broadcasting of the present invention, as figure 1 As shown, the method includes:

[0047] Step 101: Acoustically decoding the input speech, determining a candidate word sequence corresponding to the input speech according to the acoustic features corresponding to the acoustic encoding, and generating a word candidate grid according to the candidate word sequence and the corresponding backtracking path and matching score;

[0048] Using speech recognition technology, the input speech is acoustically decoded to generate a word grid with multiple candidates. In speech recognition, the input speech signal needs to go through the process of preprocessing and acoustic feature ...

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 embodiment of the invention discloses a method and device for detecting keywords in voice and belongs to the field of audio processing. The method comprises the steps that arcs in a character lattice are optimized and are clustered to be a confusion network; an index is established, so that the position of each target phonic letter in a keyword to be retrieved in the confusion network is located; the character lattice is converted into the confusion network, so that forced competing words are in the same group and words appearing in the same period in the lattice are naturally ordered according to time; the posterior probability of the words in the confusion network is directly used as a confidence coefficient mark; due to the fact that the occupied space of the confusion network is smaller than that of the lattice, the structure is simple and indexing is easy; the phonic letter of different tones are regarded as the same phonic letter for searching, so that the detection leakage rate is effectively reduced and the searching efficiency is improved.

Description

technical field [0001] The invention relates to the field of audio processing, in particular to a method and device for detecting keywords in speech. Background technique [0002] Among the existing detection methods, they are mainly divided into detection methods based on filler models and speech recognition; the first method cannot freely add and delete keywords, and the second method cannot detect foreign words. [0003] The speech keyword detection method based on the filler model needs to define the keyword and the search network of the filler model before the keyword recognition. The detection speed is fast, but it does not use the knowledge of the language model. When the keyword changes, it needs to be redefined. Key words and filler models are re-identified by the search network. In the news broadcast public opinion information retrieval task, the keywords to be queried often change, so it is not suitable for news broadcast public opinion information retrieval. ...

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): G10L15/08G10L15/183
Inventor 李伟浩苏鹏宇
Owner BEIJING JETSEN TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap