Speech Recognition using Fast Fourier Transform Algorithm

DOI : 10.17577/IJERTCONV10IS08007

Download Full-Text PDF Cite this Publication

Text Only Version

Speech Recognition using Fast Fourier Transform Algorithm

Mr. P. Manikandan

Assistant Professor

M.I.E.T Engineering College Trichy,India

M. Kiruthika


M.I.E.T Engineering College Trichy,India

K. Shrimathi


        1. Engineering College Trichy,India

          1. Mubeena


        1. Engineering College Trichy,India

          AbstractVoice-based systems permit users access to info on the net over a voice interface. Previous studies on examination systems that create use of voice interface don't sufficiently exhibit intelligent variety of assessment that diminishes the rigor of examination. the target of the system is to boost on the achievements of previous studies by providing a framework that can guide the development of a voice-based examination like government exams like TNPSC cluster II, cluster IV here we tend to develop the project for blind students United Nations agency all have an interest to participate the examination, so case knowledgeable system for the visually impaired students therein suggests that the queries are prepared by the language as they chose within the examination portal. So, the question are on scan mode, the scan mode can enhance the speech so queries are scan and therefore the choices also will admit defeat the scan mode in order that they will simply perceive the question and choices they'll answer it manner. The study employs a mixture of technologies like system style, server-side scripting, voice-based system development, knowledge management and rule-based reasoning in developing the system.

          KeywordsAI, Fast Fourier Transform, OGG/MP3

          1. INTRODUCTION

            Humans have distinctive options that distinguish one person from another. Biometric recognition or just biometrics refers to automatic ways of recognizing a person supportedpsychological or activity characteristics. The use of bioscience for human identification incorporates a heap of advantages as a result of the options explored area part of the personal data that, in most cases, is not possible to fake, share or forget like, for instance pas,swords or PINs. The way a personal speak is one of those distinctive optionsthat may be used for recognition.Voice,typically thought of as a form of activity biometric is really a mixture of each physiological and activity statistics.A voice biometric "voice print," is as distinctive to a private as a finger or palm print. Any Authentication application that employs a voice channel throughout the Authentication session is in a position to feature voice biometric identification to the method for even higher levels of authentication and security.Voice verification technology uses the various characteristics of an individual's voice to discriminate

            between speakers. Speech recognition allows providing input to associate application with voice Speaker Recognition is among the wide used statistics when it involves our activity characteristics.

          2. EASE OF USE

            1. An Automated System for Scoring Short Answers Essays and different sorts of writing practices are extensivelyused for analysis functions. Graduate Record Examination(GRE), Scholastic ability check (SAT), SeniorSchool Examinations like Zhongkao in China and every oneIndia Senior faculty Certificate Examination (AISSCE) inIndia square measure some of the numerous examples. The stakes forgetting high grades within the essays and thus in these examsare tremendous for pupils, academics and faculties alike. Theessays and short answers written by the scholars within the examsdetermine their future faculties and thus have a careerwide impact.Under the No kid Left Behind rules,

              U.S. Stateshave been asked touse uniform and controlled check scoresfor analysis of academics for determinant their salaries andtenures (Higgins 2014). This underlines the importance ofgetting sensible scores in these writing practices. A schoolsreputation is usually determined by the weekday variant its graduatingstudents, that successively is compactby however well they need been schooled to put in writing their essays and short answers(Dale and Krueger 2002).

            2. TEXT MINING

              Text mining, conjointly called text data processing or data discovery from matter databases, refers to the process of extracting fascinating and non-trivial patterns or data from text documents. Regarded by many because the next wave of data discovery, text mining has terribly high business values. Last count reveals that there ar quite

              10 sophisticated firms providing merchandise for text mining.Has text mining evolved thus chop-chop to become a mature field? this text tries to shed some lights to the question. Wefirst gift a text mining framework consisting of 2 components: Text processing that transforms unstructured text documents into associate degree intermediate form; and data distillation that deduces patterns or knowledge from theintermediate type. we have a tendency to then survey the progressive text mining

              products/applications and align them supported the text processing and data distillation functions additionallyas the intermediate type that they adopt. lastly, we have a tendency to highlight the future challenges of textmining and also the opportunities it offers.

            3. Automated Essay Grading

              Assessment is taken into account to play a central role inthe educational method. The interest within the developmentand in use of Computer-based Assessment Systems (CbAS) has fully grown exponentially in the previous few years, due each to the rise of the number of scholars attending universities and to the possibilities provided by e-learning approaches to asynchronous and omnipresent education.

              According to our findings (Valenti, Cuc-chiarelli, & Panti., 2002) quite forty business CbAS are presently out there onthe market. Most of those tools are supported the utilization of the questionable objective-type questions:

                1. multiple selections, multiple answer, short answer, selection/ association, hot spot and visual identification (Valenti et al., 2000).Most researchers during this field agree on the thesis that some aspects of complicated action are tough to measure victimisation objective-type queries. Learning outcomes implying the flexibility to recall organize and integrate concepts, the flexibility to precise oneself in writing and therefore the ability to provide simply than determineinterpretation and application of knowledge needless structuring of response than that obligatory by objective test things (Gronlund, 1985). Its within the measuring of such outcomes, reminiscent of the upper levels of the Blooms (1956) taxonomy (namely analysis and synthesis) that the essay question serves itsmost helpful purpose. One of the difficulties of grading essays is that theperspicacity, or a minimum of the perceived perspicacity, of the grading method. Several researchers claim that the subjective nature of essay assessment results in variation in grades awarded by totally different human assessors, which is perceived by students as an excellent supply of unfairness.Furthermore essay grading could be a time intense activity. in keeping with Mason (2002), about30% of teachers time in nice kingdom is dedicated to marking. So, if we wish to liberate that half-hour (worth 3 billion Britain Pounds/year to the payer by the approach) then we tend to should realize an efficient way, that teacher will trust, to mark essays and short text responses.This issue could also be sweet-faced through the adoption of machine- drven assessment tools for essays. A system for automated assessment would a minimum of be consistent within the approach it scores essays, and massive price and time savings may be achieved if the system are often shown to grade essays among the vary of theseawarded by human bureaucrat. Moreover, in keeping with publisher (2000) victimisation computers to extend ourunderstanding of the matter options and psychological feature skills concerned within the creation and within the comprehension of writtentexts, can give variety of advantages to the academic community. in reality it canhelp North American

                  country develop more practical educational materials for up reading, writing and different communication abilities. it'll conjointly facilitate North American country develop more practical technologies like search engines and question respondent systems for providing universal access to electronic data.Purpose of this paper is to gift a survey of current approaches to the machine-driven assessment of freetext answers. Thus, within the next section, the subsequent systems are discussed: Project Essay Grade (PEG), Intelligent Essay bureaucrat (IEA), instructional Testing service I, Electronic Essay Rater (ERater), C-Rater, BETSY, Intelligent Essay Marking System, SEAR, Paperless faculty free text marking Engine and Auto mark. of these systems are presently out there either as business systems or because the result ofanalysis during this field. For every system, the overallstructure and therefore the performance claimed by the authors arebestowed. In the last section, we'll attempt tocompare these systems and to spot problems that will fosterthe analysisin the field.

            4. Comparison of JSON and XML Data InterchangeFormats

              Data interchange formats evolved from being markup and display-oriented to additional support the cryptography of meta-data that describes the structural attributes of the data. The necessities to support knowledge interchange of Java applications LED to the event of standard knowledge interchange formats. JSON and XML area unit 2 knowledge interchange formats with distinctive purposes. Sections 2 and 3 offer background for JSON and XML. Section four describes the case study and methodologywont to compare speed and resource utilizations. Section 5 describes results and section six identifies the threats to the validity of this study. we tend to conclude in sectionseven and supply directions for attainable refinements tothe present study. This case study measures transmission times and resource utilizations. The null hypothesis statesthat there is no distinction in transmission times and resource utilization between JSON and XML. The operational setting for this case study consists of a client/server program. The shopper is setup in isolation andsends JSON and XML objects to the server in order to live performance and resource utilization.

            5. Biometric Recognition

            A wide kind of systems needs reliable personal recognition schemes to either ensure or confirm the identity Of a personal requesting their services. The aim of such schemes is to make sure that the rendered services square measure accessed solely by a legitimate user and nobody else. Samples of such applications include secure access to buildings, pc systems, laptops, cellular phones, and ATMs. Within the absence of strong personal recognition schemes, these systems square measure susceptible to the wiles of ansham. Biometric recognition or, simply, bioscience refers to the automated recognition of people supported their physiological and/or activity characteristics. By exploitation bioscience, it is possible to substantiate or establish associate degree individuals identity primarily based on who she is, instead of by

            what she possesses (e.g., an ID Card) or what she remembers (e.g., a password). During this paper, we provide a temporarysummary of the sector of bioscience and summarize some of its blessings, disadvantages, strengths, limitations, and related privacy considerations. Humans have used body characteristics like face, voice, and gait for thousands of years to acknowledge every other. Bertillon, chief of the criminal identification division of the department of local government in Paris, developed then practiced the thought of employing a variety of body measurements to identify criminals within the mid-19th century. Even as his plan wasgaining quality, it had been obscured by away more important and sensible discovery of the distinctiveness of the human fingerprints within the late nineteenth century. Shortly when this discovery, many major enforcement departments embraced the thought of 1st booking the fingerprints of criminals and storing it in a info (actually, a card file). Later, the leftover (typically, fragmentary) fingerprints (commonly stated as latents) at the scene of crime may be lifted and matched with fingerprints within the info to work out the identity of the criminals. Though bioscience emerged from its intensive use in enforcement to spot criminals (e.g., misappropriated aliens, security clearance for workers for sensitive jobs, fatherhood determination, forensics, and identification of convicts and prisoners), it's being progressively used nowadays to ascertain person recognition in an exceedingly sizable amount of civilian applications.

          3. ADMIN

            Only the admin can navigate to the project. Only the known person will enter by giving valid information. If the user provides invalid information then permission is denied to navigate to other pages. This authentication module concentrates on the security of the project from unauthorized users. Admin can authenticate only if the cloud authority provides permission else the access is denied to the user.

                1. Add Subject

                  The admin has the responsibility to add subject for TNPSC and UPS exams. The module will be enhanced thelevel of the subjects and the determination among they using with the purpose of the add subject in the era. Subjectdetails will be added and view to the portal and it was maintain to the admin side.

                2. Add Question

                  The add question module will be declared that the questions are added by the admin side, they will evolve the termination along with the examination for TNPSC and UPS examination, there will be the questions will be added in the era. Type of subject the type and questions will be added at the model so the based on subject question will beadd.

                3. View Question

                  The view question module will be declared that the, portalof the questions will be added by the admin side, would be viewed at the view question portal, questions willbe view at the and recognize the questions along the

                  add subjects and add questions. After added the question by theadmin the questions will be viewed at the portal.

                4. Voice Notification

                  The notifications of the voice message will be sent to the student, whenever the exam notification will be announcing by the government the details and the updates will be send as notification to the student side. The notification of the link will be also sent to the student the link will send as the video.

                5. View Result

            The result will be shown as the portal as they declared t\and work on the examination that TNPSC and UPS examinations with the result. Admin will collect all the needed information that will be entered with the application. Department wise timetable details will be submitted as report. Admin workload during class period will also report. The module will display the timetable information notification to the end user that will help to enhance the admin to know the weather information, management notification.

          4. STUDENT

            1. Registration

              In the registration module the student will register for the authentication purpose. By this registration the admin will view the student details and retrieve for the further verification. This module is the first module fr the studentwhich is the gateway for the other module

            2. Login

              In the login module the each admin can enter the valid student name and therefore the positive identification to enter within the home page. This module are going to be accessed by the licensed user United Nations agency is aware of the positive identification that is developed. These modules are going to be the entryway module for the project which will facilitate to enter the information.

            3. View Subject

              The admin has the responsibility to add subject for TNPSC and UPS exams. The module will be enhanced thelevel of the subjects and the determination among they using with the purpose of the add subject in the era. The view portal will shows that subjects will be added by the admin.

            4. View Test

            The view test portal will be enhanced the level of the testing that which will be added by the admin. The test details will be view at the website.

          5. VOICE RECOGNIZE (OGG/MP3)

            OGG Vorbis has higher sound quality than MP3. The compressed enter OGG Vorbis format is smaller than the compressed file of the MP3 format. The bit rate of compression in OGG format varies in keeping with the need of the file whereas the bit rate of compression is constant in MP3 format. The voices are going to be recognized and developed by the portal.

          6. FAST FOURIER TRANSFORM ALGORITHM As the name implies, the quick Fourier remodel (FFT)

            is associate degree formula that determines distinct Fourier remodel of associate degree input considerably quicker than computing it directly. In engineering science slang, the

            FFT reduces the quantity of computations required for a tangle of size N from O (N^2) to O (NlogN).

            The FFT operates by rotten associate degree N purpose time domain signal into N time domain signals every composed of one purpose. The second step is to calculate the N frequency spectra resembling these N time domain signals. Lastly, the N spectra square measure synthesized into one frequency spectrum.

            a quick Fourier remodel (FFT) is associate degree formula that computes the distinct Fourier remodel (DFT) of a sequence, or its inverse (IDFT). Harmonic analysis converts a sign from its original domain (often time or space) to a illustration within the frequency domain and the other way around.

            AD MIN



            Upload Exam Details

            Exam Details Voice Recognition

            Speech Recognition Fast Fourier Algorithm

            Function FFT (A, )

            Input: constant illustration of a polynomial A(x) of degree

            n 1, wherever n could be a power of two

            Output: price illustration A (^0). . . A (^n1) if = 1: come back A(1)

            categorical A(x) within the kind Ae(x^2) + xAo(x^2) decision FFT(Ae, ^2) to judge Ae at even powers of decision FFT(Ao, ^2) to judge Ao at even powers of for j = zero to n 1:

            cipher A(^j) = Ae(^2j) + ^jAo(^2j) come back A (^0). . .A (^n1)


            A system design or systems design is that the abstract model that defines the structure, behavior, and a lot of views of a system. Associate in Nursing design description could be a formal description and illustration of a system, organized during a manner that supports reasoning concerning the structures and behaviors of the system. System design will comprise system elements, the outwardly visible properties of these elements, the


            Voice recognition

            Exam Detail View

            Regist ration



            relationships (e.g. the behavior) between them. Fig.1 Architecture Diagram

          8. RESULT The voice-based examination system would improve the accessibility of examination in distance learning for learners with visual impairment, as well as other able- bodied learners.

            Fig.2 Question View and Recognize


To conduct an Online Examination using voice which will be helpful for people who dont want to use keyboard for interaction with the system. The proposed method will help for voice recognition where we take voice as input through microphone and then register for online examination using the concept of Key generation. The system will analysis the voice based examinations like government exams TNPSC exam, UPS exam, the online tests will be provided by the voice recognition along with perspective with add subjects and add and view the subjectsdetails, with MP3 and ogg format . The questions and the answers will given like an options. So it will be develop the regular systems with online voice based examination. In this paper, an e-examination voice interface for the visually impaired learners in ODL has been provided. The developed system was realized using a framework, system design with pseudo code dialogue sequence and algorithm. A usability evaluation of the system was also conducted.


[1] Bandura, A. Psychological Modeling. New York:

Aldine Atherthon.1971

[2] L.Lamel and J.L.Gauvain,"Speech recognition", inMitkov, R. (Ed.) The Oxford Handbook of ComputationalLinguistics.

Oxford: Oxford University Press. 2003 [3]National Center for Improved Practice in SpecialEducation (NCIP) Update onVoice/Speech Recognition. New York: NCIP.2003

[3] Y.Y. Wang, L. Deng, A. Acero, "Spoken language understanding", IEEE Signal Processing Magazine 22, 5 16-

31. 2005.

[4] Biggs, M.L. Learning Theories for Teachers. Ed.ke-3. New York: Harper and Row Publishers. 1976.

[5] P.R. Atkins and Collins, T. Physical Constraints in Sonar Design, The Journal of the Acoustical Society of America, 109, 5, Part 2, 2001, 2285-2286

[6] Reed, S.K. Cognition. Fifth edition California: AddisonWesley, California, 2000.

[7] Graevenitz, G. A. von (n.d.). About Speaker Recognition Technology. Bergdata Biometrics GmbH,

[8] Germany

[9] Rozmiarek, D.J.Continuous Speech Recognition and Computers: A Written Communication Tool for Students with learning diabilities. Delaware: University of Delaware. 1998.

[10] Dragon Systems Unveils Revolutionary Breakthrough with Continuous Speech Recognition. New York: Dragon Systems. 2001.

[11] Snaidoo, A. Automatic Voice Recognition System. California: University of California. 2003.

[12] Ainsworth, W.A. Speech Recognition by Machine. London: Peter Peregrinus Ltd on behalf of the IEEE (IEEE Computing Series, 12.) 1988.