CN108628835A

CN108628835A - Using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app

Info

Publication number: CN108628835A
Application number: CN201710172487.5A
Authority: CN
Inventors: 邱念
Original assignee: Hunan Original Culture Development Co Ltd
Current assignee: Hunan Original Culture Development Co Ltd
Priority date: 2017-03-22
Filing date: 2017-03-22
Publication date: 2018-10-09

Abstract

The invention discloses a kind of using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app, including：1）User terminal mobile phone app softwares；2）The convolutional neural networks model that cloud computing center carries；3）The voice large database concept of Guangdong language accent；4）The voice large database concept of Spanish pronunciation；5）The large database concept of middle Spanish intertranslation；6）The proper noun large database concept of several industry types of Spanish；7）Guangdong language talks about syntax data；8）Spanish syntax data；After convolutional neural networks by voice data by carrying out audiograph drafting, input convolutional neural networks training, convolutional neural networks after the completion of training the Guangdong language or spanish audio that app is transmitted can be translated into Spain with or Guangdong language pass app back again, it is reached for the purpose that user is translated, and instead of translator, greatly reduce the translation cost of user.

Description

Using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app

Technical field

The present invention relates to the application field that convolutional neural networks algorithm and cell phone application technology be combined with each other, more particularly to Using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app.

Background technology

With the quickening of internationalization process, go to the Chinese Guangdong user that Spain travels more and more, more and more frequently, But it is not that each goes to Hispanic Cantonese all and can say Spanish and standard mandarin；Before not being each yet Spain's translation retinue can be afforded toward the Hispanic Chinese for saying Guangdong language accent, and engages and understands that the Spain of Guangdong language is turned over It translates that its searching process is just cumbersome, is unfavorable for the convenience of trip.

Invention content

Convolutional neural networks algorithm translation Guangdong language and Xi Ban are used the invention mainly solves the technical problem of providing a kind of The mobile phone app of tooth language can substitute the advanced simultaneous interpretation translation of high wages, and providing to the user will not be because of translation time length and because tired Translation error caused by labor, and can identify the Guangdong language accent of user, avoid user from not speaking standard Chinese pronunciation, translator is ignorant of again The difficult situation of Guangdong language accent.

In order to solve the above technical problems, one aspect of the present invention is：It provides a kind of using convolutional Neural net Network algorithm translation Guangdong language and Spanish mobile phone app, which is characterized in that including：1）User terminal mobile phone app softwares；2）Cloud meter The BP neural network model that calculation center is carried；3）The voice large database concept of Guangdong language accent；4）The big number of voice of Spanish pronunciation According to library；5）The large database concept of middle Spanish intertranslation；6）The proper noun large database concept of several industry types of Spanish；7） Guangdong language talks about syntax data；8）Spanish syntax data, by above-mentioned 3）、4）、5）、6）、7）、8）In the database and cloud computing of item The BP neural network model that the heart carries carries out data and links, then carries out the data training of time period, so that intelligence friendship is passed and turns over Rate of accuracy reached is translated to 95% or more, intelligent simultaneous interpretation translates rate of accuracy reached to 70% or more, if translation accuracy rate is less than the value, extends Its accuracy rate is set to improve to the translation quality phase of the high level translation personnel with international standard the cycle of training of BP neural network model Closely, until after the completion of BP neural network model training, it is connected with the app of user terminal by network, simultaneous interpretation is needed to translate The user of Spanish takes earphone, allows the people that the needs of Guangdong language accent are translated to speak against the app, voice data will be same Step is transmitted to the translation model of the accent of BP neural network Guangdong language and Spanish simultaneous interpretation intertranslation of cloud computing center, BP neural network For model by translated content Spanish voice output, last output data passes through the earphone transmission that is connected on user mobile phone Give the user；User can independently the audio of switching input and output be the national language or Spanish of Guangdong language accent on app.

The convolution process of convolutional neural networks is：It is deconvoluted an audio sound inputted with a trainable filter fx Wave image（It is input picture at C1 layers, the input of convolutional layer later is then the convolution characteristic pattern of preceding layer）, activated by one Function (generally uses Sigmoid functions), and then plus one biases bx, obtains convolutional layer Cx.Concrete operation such as following formula, formula Middle Mj is the value of input feature vector figure：

The process of sub-sampling includes：The m pixel per neighborhood（M is to be manually set）Summation becomes a pixel, then passes through mark Wx+1 weightings are measured, biasing bx+1 is further added by, Feature Mapping figure is then generated by activation primitive Sigmoid；From a plane to The mapping of next plane can be regarded as making convolution algorithm, and S layers are considered as fuzzy filter, play Further Feature Extraction Effect；Spatial resolution between hidden layer and hidden layer is successively decreased, and the number of planes contained by every layer is incremented by, and can be used for detecting more in this way More characteristic informations；For sub-sampling layer, there is N number of input feature vector figure, just has N number of output characteristic pattern, only each feature The size of figure has obtained corresponding change, concrete operation such as following formula, down in formula（）Indicate down-sampling function；

。

Claims

1. using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app, which is characterized in that including：1）User Hold mobile phone app softwares；

2）The convolutional neural networks model that cloud computing center carries；

3）The voice large database concept of Guangdong language accent；

4）The voice large database concept of Spanish pronunciation；

5）The large database concept of middle Spanish intertranslation；

6）The proper noun large database concept of several industry types of Spanish；

7）Guangdong language talks about syntax data；

8）Spanish syntax data.

2. it is according to claim 1 using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app, it is special Sign is：The mobile phone app softwares of user terminal are necessarily mounted on the smart mobile phone that can be surfed the Internet for having earphone access device, and The smart mobile phone of user need to be connected with internet；The built-in function of converting audio data to audiograph of cell phone application.

3. it is according to claim 1 using convolutional neural networks algorithm translation Guangdong language and Spanish mobile phone app, it is special Sign is to translate and include the following steps：

Step 1 builds convolutional neural networks model in cloud computing center；

Step 2, that all kinds of and Guangdong language stored in cloud computing center and Spanish relevant big data are carried out classification point library is whole Reason, and every audio data is converted to sound wave picture；

The sound wave picture classification of each large database concept is transmitted to the training that convolutional neural networks carry out big data by step 3；

Step 4 randomly selects 1000 groups of spanish audios and 1000 groups of Guangdong language audios, inputs trained convolutional Neural respectively Network tests it and translates accuracy rate；If simultaneous interpretation translates accuracy rate and is higher than 70%, biography translation accuracy rate is handed over to be higher than 95%, then convolution god Through network training success；If being less than the value, repeatedly step 3, and the big data sample size that step 3 inputs is increased, then carried out Training of the convolutional neural networks to Guangdong language and Spanish Translator, until rate of accuracy reached is designated as stopping；

Step 5, formal in use, say Guangdong language or Spanish against the smart mobile phone for installing APP of the present invention, APP is by sound Frequency is according to audiograph is converted to, and then the convolutional neural networks by audiograph by network transmission to cloud computing center translate mould Type translates corresponding Spanish or Guangdong language after carrying out analysis and identification to audiograph by convolutional neural networks, then will be after translation Audio transmission to cell phone application, user listens to the audio content after translating by wearing the handset earphone being connected with the APP.