Robustness speech recognition method for agricultural product market element information collection

A technology of element information and speech recognition, which is applied in speech recognition, speech analysis, data processing applications, etc., and can solve problems such as inability to recognize speech

Active Publication Date: 2014-11-19
AGRI INFORMATION INST OF CAS
View PDF4 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem in the prior art that the terminals of agricultural product market information collection cannot be recognized by speech, a r

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
  • Robustness speech recognition method for agricultural product market element information collection
  • Robustness speech recognition method for agricultural product market element information collection
  • Robustness speech recognition method for agricultural product market element information collection

Examples

Experimental program
Comparison scheme
Effect test

example 1

[0102] Example 1: Test in the environment of a large-scale agricultural product wholesale market. The test set recorded 3 males and 3 females with 50 sentences each, a total of 300 sentences, which were recorded by mobile phones in a relatively quiet environment as approximately pure speech, and the speakers were not in the training set. Then artificial noise is added to the noise of the large-scale agricultural product wholesale market environment, and finally the noisy speech with signal-to-noise ratios of -5dB, 0dB, 5dB, 10dB, 15dB, 20dB, and 25dB is obtained. The test voice is 300 sentences, a total of 2100 sentences. For the baseline system, various spectral subtraction algorithms are used alone, and various algorithms combined with CMVN are compared and tested, and the recognition rates shown in Table 1 are obtained. Among them, this algorithm is SSMMSE (spectral minus minimum mean square error) + CMVN, and its recognition rate curve is as attached figure 2 shown.

...

example 2

[0106] Example 2: Similar to the above example 1, the above experiment was carried out in the community farmer's market environment, and the recognition rate results obtained are shown in Table 2 below, and the recognition rate result curve is shown in the attached image 3 shown.

[0107] Table 2 Recognition rate in community farmer's market environment

[0108]

[0109] It can be seen from the above examples that the anti-noise robust speech recognition algorithm proposed by the present invention has a higher recognition rate in the field of agricultural product market information collection, especially in a low signal-to-noise ratio environment, and its improved performance is more obvious.

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 relates to the technical field of speed recognition, in particular to a robustness speech recognition method for agricultural product market element information collection. The method includes the steps that initial speech signals are collected; denoising is conducted on the initial speech signals through an MMSE spectral subtraction algorithm, and approximately-pure speech signals are obtained; the feature value of the approximately-pure speech signals is extracted; CMVN compensation is conducted on the feature value, and an HMM is trained according to a speech feature vector obtained after the compensation is conducted. By means of the method, the algorithm adopted for a continuous speech recognition system facing non specific people with medium vocabulary is simple, and the method is easy to implement and small in calculation amount.

Description

technical field [0001] The invention relates to the technical field of voice recognition, in particular to a robust voice recognition method for collecting agricultural product market element information. Background technique [0002] Agricultural product market information is related to agriculture and social stability, and is the basis for ensuring stable and healthy economic development. In response to the importance of agricultural product market information, relevant national departments and local governments have also established various forms of agricultural product market information collection platforms. The information collection methods often use traditional manual transcription and then secondary input into computers, telephone quotations or emails. However, this type of information collection method involves a lot of repetitive labor, low efficiency, and poor timeliness. For this reason, many institutions and researchers have proposed information collection met...

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): G10L21/02G10L15/14G06Q50/02
Inventor 诸叶平许金普
Owner AGRI INFORMATION INST OF CAS
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