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

Voice key information separation method based on deep learning

A technology of voice information and key information, which is applied in the field of separation of key voice information based on deep learning, can solve problems such as complex processes, numerous steps, and decreased accuracy, and achieve the effects of avoiding cumulative errors, reducing manual intervention, and improving effects

Pending Publication Date: 2020-11-03
厦门熙重电子科技有限公司
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a method for separating key speech information based on deep learning, and its main purpose is to overcome the problems of the existing speech retrieval methods with numerous steps and complicated processes, which lead to accumulation of errors and decrease in accuracy

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
  • Voice key information separation method based on deep learning
  • Voice key information separation method based on deep learning
  • Voice key information separation method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0028] refer to Figure 1 to Figure 4 , a method for separating key speech information based on deep learning, comprising the following steps:

[0029] S1, CNN training: the voice signal sample set is used as training data, and the key information to be tested is used as the label, and the CNN convolutional neural network is used to train the voice signal sample set to obtain a voice information classification model, and the voice information obtained after training The classification model can distinguish whether different speech signals contain key information that needs attention, such as determining whether there is "ID card" related information in a speech.

[0030] S2. Voice information calibration: based on the trained voice information classification model, pass the voice signal to be tested through the voice information classification model, and use the reverse gradient activation average algorithm and feature weighted activation mapping algorithm to automatically cal...

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 invention discloses a voice key information separation method based on deep learning. The method comprises the following steps of CNN training, voice information calibration, voice information feature clustering and voice information separation. According to the artificial intelligence method based on deep learning and clustering, the key voice signals in the voice signals can be automaticallyseparated under the condition that manual intervention is reduced as much as possible. In the process, semantic analysis is not involved, accumulated errors in a traditional voice separation processing flow can be avoided, the method can also be used as a preprocessing means of a traditional method, and the voice separation effect is further improved.

Description

technical field [0001] The present invention relates to speech processing technology, and specifically refers to a method for separating key speech information based on deep learning. Background technique [0002] As a key multimedia data, speech plays an important role in information expression, storage, and human-computer interaction. Speech signals contain rich information, and speech information retrieval is an important research hotspot at present. [0003] At present, the common means of voice intelligent retrieval are keyword retrieval, sentence retrieval, and document retrieval. Most of the existing retrieval methods rely on semantic analysis in speech. The retrieval methods have many steps and complicated processes, which lead to the accumulation of errors and the decline in accuracy. Contents of the invention [0004] The present invention provides a method for separating key speech information based on deep learning, and its main purpose is to overcome the prob...

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
IPC IPC(8): G10L25/54G10L25/78G10L25/30G10L25/03
CPCG10L25/54G10L25/78G10L25/30G10L25/03Y02T10/40
Inventor 张建国叶家艺茅剑
Owner 厦门熙重电子科技有限公司
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