Font Size: a A A

Design And Implementation Of Mobile Phone Security System Based On Speech Recognition

Posted on:2013-11-04Degree:MasterType:Thesis
Country:ChinaCandidate:S H CuiFull Text:PDF
GTID:2248330377951924Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the advent of the mobile Internet era, more and more handheld devices beganto enter people’s lives. Mobile phones, especially smart phones play an increasingrole in people’s lives. People complete a variety of information exchange throughmobile phones; therefore, the confidentiality of the phoneitself has become increasingly important.For this problem, combined with research focus of current identification, in thispaper, voiceprint recognition technology in the biometric technology is used to mobilephone owner identification. At the same time, phones have a specific characteristicthat it would be in a variety of noise occasions at any time; this paper usesmicrophone array speech enhancement technology for the front-end processing. Andthe system is implemented on the WinCE embedded operating system. Overall, thispaper studied from the flowing aspects.First, this paper studied the development of related technology as well as thecomparison of several technology applications and chooses one of them as the focusof this paper. Followed by the transplantation of an embedded operatingsystem WinCE and elaborated its architecture and transplantation process. Form ARMand peripherals being connected with mature connection to ensure the stable ofsystem bottom. Then studied the theoretical basis of the speech recognition,including the pretreatment of the sub-frame, windowed, and focus on the endpointdetection technology of the pretreatment. Voice activity detection is to ensurethat the result is correct to identify an important prerequisite for, so the focus ofthis study. Next, this paper studied Mel Frequency Cepstrum Coefficient(MFCC) extraction process in the parameter extraction stage, this process is describedin detail. During speech recognition using vector quantization (VQ) and thistechnology avoids dynamic adjustment when the length of templates is inconsistentwith the matched voice in dynamic time warping (DTW) technology. Finalizedsystems software development based on the Embedded C++, and introduced the development and use of GPRS module.The experiment results show that the system can meet the expected results andprovide the possibility of transplanting to other platforms theoretical basis for future.
Keywords/Search Tags:Mobile phone security, WinCE, Speech recognition, Mel FrequencyCepstrum Coefficient (MFCC), Vector quantization (VQ)
PDF Full Text Request
Related items