Method and system for generating mixed voice data

一种混合语音、语音数据的技术，应用在语音分析、语音识别、仪器等方向，能够解决人工采集周期长不利研发、音频数据不足、提高研发成本等问题，达到提高收敛速度、提高收集速度、提高性能的效果

Pending Publication Date: 2019-10-11
XIAMEN YEALINK NETWORK TECH
View PDF7 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] At present, the application of deep learning in the field of speech recognition is constantly developing, but in the existing technology, the lack of audio data required for deep learning is a major problem. Traditional solutions usually collect data through manual collection, but in the actual process , it is difficult to cover all scenes through artificially collected noise. The long cycle of manual collection is not conducive to research and development, and it also increases the cost of research and development

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 system for generating mixed voice data
  • Method and system for generating mixed voice data
  • Method and system for generating mixed voice data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] combine figure 1 As shown, a method for generating mixed voice data of the present invention first collects pure voice and noise, then normalizes the collected voice data, then randomizes the processed data, and then performs GAIN on the data processing, and finally the mixed voice data is obtained through filter processing. Specific steps are as follows:

[0033] Step 1. Raw data collection

[0034] Gather pure voice data and noise data earlier; Pure voice is to collect in anechoic chamber among the present embodiment, and pure voice is the voice (such as low noise floor, high signal-to-noise ratio) of noise floor. figure 2 shown). Acquisition of noise is carried out in two ways: on-site collection and network download collection. It is worth noting that it is necessary to collect noise in different scenarios, such as noise collection in offices, streets, and stations (such as image 3 shown).

[0035] Step 2, normalization processing

[0036] First convert the ...

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 method and system for generating mixed voice data, and belongs to the technical field of voice recognition. The method for generating the mixed voice data comprises the stepsthat first pure voice and noise are collected, then the collected voice data is normalized, then the processed data is randomized, then GAIN processing is performed on the data, and finally the mixedvoice data is obtained through filter processing. The system for generating the mixed voice data includes a collection unit, a calculation unit and a storage unit. The collection unit is electricallyconnected with the calculation unit, and the calculation unit is connected with the storage unit through a data transmission unit. The method and system for generating the mixed voice data aim to overcome the shortage of audio data required for deep learning in the prior art, the mixed voice data can be automatically generated, and the data requirements of deep learning can be met.

Description

technical field [0001] The present invention relates to the technical field of voice recognition, and more specifically, to a method and system for generating mixed voice data. Background technique [0002] With the development of science and technology, speech recognition has become a key point in the application of artificial intelligence. It is simple and convenient to control equipment through voice, and there has been an upsurge of research and application in various fields. Data, algorithms, and chips are the three keys to speech recognition technology. A large amount of high-quality data, accurate and fast algorithms, and high-performance speech recognition chips are the core to improve speech recognition. Speech recognition technology is a high technology that allows machines to convert speech signals into corresponding text or commands through the process of recognition and understanding. Speech recognition technology mainly includes three aspects: feature extractio...

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 Applications(China)
IPC IPC(8): G10L15/26G10L25/27
CPCG10L15/26G10L25/27G10L21/00G10L15/063G10L21/0208G10L25/93G10L2025/935
Inventor 康元勋方泽煌冯万健
Owner XIAMEN YEALINK NETWORK 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