Design And Implementation Of Voiceprint Recognition On Android Platform

With the development of the times,personal information plays a more and more important role in people's life.Fingerprint,iris,and face become the mainstream personal identification features,and their technology is mature and widely used.As fingerprint,iris and face,voiceprint is unique,economical and convenient.It is a potential biometric method.Mobile devices are commonly used recognition carriers.The above recognition methods ultimately play their practical value and utility on the carriers.Android is a commonly used system for mobile devices,and its economic prospect is good.In this dissertation,Android is used as a platform to study voiceprint recognition model The application performance of recognition(VR)on Android platform.At the same time,because the computing power of mobile device is worse than that of PC,prune and compress the model,reduce the calculation amount of the model,and accelerate the recognition speed of the model while ensuring the acceptable accuracy.In this dissertation,the common feature parameters and recognition algorithms of voiceprint recognition are analyzed.The Mel frequency cepstrum coefficient(MFCC)is used as the speaker's speech parameters.The extraction process of Mel cepstrum coefficient is introduced in detail,and its advantages as parameters are analyzed.Deep learning is studied The application of learning(DL)in voiceprint recognition,using deep learning network model to extract features of speech parameters,and explore the recognition effect.The voiceprint recognition algorithm studied in this dissertation applies the deep learning technology.After training and adjusting on the PC,it has achieved good recognition effect.It has achieved good accuracy in the open set and closed set tests.After the compression of the model,its calculation amount has been reduced.After transplanting to the Android platform,the recognition accuracy and recognition speed are suitable for the application of mobile devices.This dissertation has verified the portability and practicability of the voiceprint recognition model on Android devices through research,and has good application prospects and economic benefits.At the same time,due to recording,illness,noise,etc.,voiceprint recognition will have security,stability and other issues to be resolved,which also makes the application of voiceprint recognition will encounter more obstacles,requiring the joint efforts of relevant researchers.