Data acquisition and model training method and device for isolated word speech recognition

A technology for data collection and model training, applied in speech recognition, speech analysis, instruments, etc., to solve problems such as the impact of recognition performance

Pending Publication Date: 2021-03-02
北京紫光青藤微系统有限公司
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

If you only collect speech samples of specific environmental factors and train the speech recognition model based on them, the recognition performance will be affected

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
  • Data acquisition and model training method and device for isolated word speech recognition
  • Data acquisition and model training method and device for isolated word speech recognition
  • Data acquisition and model training method and device for isolated word speech recognition

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present application.

[0036] The embodiment of the present invention discloses a data collection and model training method and device for speech recognition of isolated words, including: collecting speech samples in batches, and the speech samples collected in the first batch or the first few batches include "noisy" The speech of isolated words in "environment" and the speech of isolated words in "fixed environment", the subsequent batches can only collect the speech of isolated w...

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 data acquisition and model training method and device for isolated word speech recognition, relates to the technical field of speech recognition, and is used for reducing thecost of isolated word speech sample acquisition and improving the acquisition efficiency on the premise of ensuring the robustness of speech recognition. The method comprises the steps of collectingisolated word voices in batches, and training a Y-shaped network by utilizing the isolated word voices collected in a noisy environment and the isolated word voices collected in a fixed environment ina first batch or several previous batches; and only collecting fixed environment isolated word voices in subsequent batches, and only updating the model parameters of the semantic feature sub-networks during network training (see Figure 1 in the specification).

Description

technical field [0001] The invention relates to the technical fields of speech processing and speech recognition, in particular to a data collection and model training method and device for speech recognition of isolated words. Background technique [0002] At present, in some fields (such as mobile phone applications, smart furniture, industrial control, etc.), it may involve device wake-up and on-demand change of device status. If adopt button mode, realize above-mentioned function, then convenience is not strong. [0003] Using a specific voice to wake up the device or change the device status by voice commands has the advantages of non-contact and strong real-time performance, which improves the user's application experience. [0004] Due to differences in application environments and voice acquisition equipment, voice signals will be affected by factors such as environmental noise, surrounding voices, and channel distortion. A successful speech recognition system must...

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): G10L15/06G10L15/02G10L15/08G10L15/16
CPCG10L15/063G10L15/02G10L15/16G10L15/08
Inventor 徐彧毋磊续素芬
Owner 北京紫光青藤微系统有限公司
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