Voice and speech recognition in Tamil language

Kiran, Ravi; Nivedha, K; Devi, S Pavithra; Subha, T.D.

doi:10.1109/iccct2.2017.7972293

Cited by 9 publications

(3 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In other words, speech recognition can be defined as speech transcription using a computer [1]. ASR can be applied to numerous practical areas, such as controlling software [2,3], dialing numbers [4], internet searches [5,6], etc. It is a difficult problem, so several recognition techniques have been proposed, including linear-time-scaled word-template matching [7], hidden Markov models (HMMs) [8,9,10], deep neural networks [11], etc.…”

Section: Introductionmentioning

confidence: 99%

Isolated Word Recognition by Recursive HMM Parameter Estimation Algorithm

Vaičiulytė

Sakalauskas

2021

cai

View full text Add to dashboard Cite

Automatic speech recognition (ASR) technologies enable humans to communicate with computers. Isolated word recognition (IWR) is an important part of many known ASR systems. Minimizing the word error rate in cases of incremental learning is a unique challenge for developing an on-line ASR system. This paper focuses on on-line IWR using a recursive hidden Markov model (HMM) multivariate parameter estimation algorithm. The maximum likelihood method was used to estimate the unknown parameters of the model, and an algorithm for the adapted recursive EM algorithm for HMMs parameter estimation was derived. The resulting recursive EM algorithm is unique among its counterparts because of state transition probabilities calculation. It obtains more accurate parameter estimates compared to other algorithms of this type. In our experiment, the algorithm was implemented and adapted to several datasets for IWR. Thus, the recognition rate and algorithm convergence results are discussed in this work.

show abstract

Section: Introductionmentioning

confidence: 99%

Isolated Word Recognition by Recursive HMM Parameter Estimation Algorithm

Vaičiulytė

Sakalauskas

2021

cai

View full text Add to dashboard Cite

show abstract

“…The Voice recognition process involves speech signal/pre-processing, feature extraction, feature matching, and voice recognition (Tymchenko et al, 2020;Krčadinac et al, 2021;Ye & Yang, 2021). The primary aim of voice recognition is to enhance security during identification and verification (Imario et al, 2017;Kiran et al, 2017). It is applied in various fields, including telebanking, video games, household appliances, surveillance and security systems, mobile application, voice dialling, online services, personal assistant services, and access control (Khotimah et al, 2020;Thomas et al, 2020;Krčadinac et al, 2021;Ye & Yang, 2021).…”

Section: (A) Voice Recognitionmentioning

confidence: 99%

Development of a secure multi-factor authentication algorithm for mobile money applications

Ali

View full text Add to dashboard Cite

With the expansion of industry 4.0, financial technology (FinTech) has become paramount in this era. Mobile money as one of the FinTech has immensely contributed to improving financial inclusions among the unbanked population in many developing countries. Several mobile money schemes were developed to ensure easy access to mobile money services. However, they have suffered severe authentication security challenges since implementing two-factor authentication (2FA). Therefore, this research developed a secure multi-factor authentication (MFA) algorithm for mobile money applications that combines personal identification number (PIN), one-time password (OTP), and biometric fingerprints to authenticate the mobile money subscribers. It also used the customer’s biometric fingerprints and the agent’s quick response (QR) code to authorise money withdrawal. The PINs and OTP are secured by secure hashing algorithm-256 (SHA-256) and biometric fingerprints by Fast IDentity Online (FIDO), where the Rivest-Shamir-Adleman (RSA) encryption protects the public/private key pair and the fingerprint templates. The QR codes, confidential financial information in the databases, and all the data before transmission to the remote databases are secured using Fernet encryption. A design science research approach was employed in the research using a mixed-method. The review results identified and grouped the threat models into attacks against privacy, authentication, confidentiality, integrity, and availability. The cryptographic functions and personal identification were the countermeasures. The survey identified authentication attacks, identity theft, phishing attacks, and PIN sharing as the crucial security issues Uganda’s mobile money systems encountered. The security analysis of the designed algorithm and developed native genuine mobile money (G-MoMo) applications proved that it provided robust security during authentication and ensured data confidentiality, integrity, privacy and user anonymity. It is highly effective against several security attacks and resilient to non-repudiation. The performance analysis results showed that the algorithm enhanced security but had high communication overhead and computational cost. Lack of a forward navigation button, lack of uniformity in the applications menu title, lack of search field options, lack of actions needed for recovery, and lack of help & documentation, were identified as the results of the usability issues with the native G-MoMo applications’ user interfaces. While the results of the usability testing showed that the native G-MoMo applications were learnable, effective, efficient, memorable, had few errors, satisfaction, ease of use, aesthetic, helpful, easy to integrate, and understandable. In conclusion, implementing a secure mobile money authentication using the ii novel approach combining multiple factors helps mobile money subscribers and other stakeholders trust the mobile money industry since the security goals are highly maintained.

show abstract

“…The phases involve speech signal/pre-processing, feature extraction, feature matching, and voice recognition (Tymchenko et al, 2020;Krčadinac et al, 2021;Ye & Yang, 2021). The primary aim of voice recognition is to enhance security during identification and verification (Imario et al, 2017;Kiran et al, 2017).…”

Section: (V) Voice Recognitionmentioning

confidence: 99%

Development of a secure multi-factor authentication algorithm for mobile money applications

Guma¹

View full text Add to dashboard Cite

With the evolution of industry 4.0, financial technologies have become paramount and mobile money as one of the financial technologies has immensely contributed to improving financial inclusion among the unbanked population. Several mobile money schemes were developed but, they suffered severe authentication security challenges since they implemented two-factor authentication. This study focused on developing a secure multi-factor authentication (MFA) algorithm for mobile money applications. It uses personal identification numbers, one-time passwords, biometric fingerprints, and quick response codes to authenticate and authorize mobile money subscribers. Secure hash algorithm-256, Rivest-Shamir-Adleman encryption, and Fernet encryption were used to secure the authentication factors, confidential financial information and data before transmission to the remote databases. A literature review, survey, evolutionary prototyping model, and heuristic evaluation and usability testing methods were used to identify authentication issues, develop prototypes of native genuine mobile money (G-MoMo) applications, and identify usability issues with the interface designs and ascertain their usability, respectively. The results of the review grouped the threat models into attacks against privacy, authentication, confidentiality, integrity, and availability. The survey identified authentication attacks, identity theft, phishing attacks, and PIN sharing as the key mobile money systems’ security issues. The researcher designed a secure MFA algorithm for mobile money applications and developed three native G-MoMo applications to implement the designed algorithm to prove the feasibility of the algorithm and that it provided robust security. The algorithm was resilient to non-repudiation, ensured strong authentication security, data confidentiality, integrity, privacy, and user anonymity, was highly effective against several attacks but had high communication overhead and computational costs. Nevertheless, the heuristic evaluation results showed that the G-MoMo applications’ interface designs lacked forward navigation buttons, uniformity in the applications’ menu titles, search fields, actions needed for recovery, and help and documentation. Similarly, the usability testing revealed that they were easy to learn, effective, efficient, memorable, with few errors, subscriber satisfaction, easy to use, aesthetic, easy to integrate, and understandable. Implementing a secure mobile money authentication and authorisation by combining multiple factors which are securely stored helps mobile money subscribers and other stakeholders to have trust in the developed native G-MoMo applications.

show abstract

Voice and speech recognition in Tamil language

Cited by 9 publications

References 1 publication

Isolated Word Recognition by Recursive HMM Parameter Estimation Algorithm

Isolated Word Recognition by Recursive HMM Parameter Estimation Algorithm

Development of a secure multi-factor authentication algorithm for mobile money applications

Development of a secure multi-factor authentication algorithm for mobile money applications

Contact Info

Product

Resources

About