Automatic Speech Therapy For Kids Using Frequency Comparison

Automatic Speech Therapy For Kids Using Frequency Comparison

AutomaticSpe chTherapyForKidsUsing FrequencyComparison

JoswinValerianConcesao*1,De kshaKundabarandady*2,JArpana*3,NidhiPrakash*4,SarikaHegde#

*UGStudent,NMAM InstituteofTechnology

Nite-576201,India

#Asistantprofesor,ComputerScienceEngine ring

NMAM InstituteofTechnology, Nite-576201,India

Keyword-Spe chtherapy,Interactivefe dback,Game,Children, Voicedisorder

I. INTRODUCTION

Languagetrainingisamediationbenefitthatspotlights on enhancing a tyke's discourse and capacities to comprehend and expres dialect,includingnonverbal dialect.Language teachers,ordiscourse and dialect pathologists(SLPs),are the expertswho give these administrations.Languageinstructionincorporatestwo parts:1)organizingthemouthtocreatesoundstoshape words and sentences (to adres explanation, familiarity, and voice volume control); and 2) understandingandcommunicatingdialect(toadres theutilizationofdialectthroughcomposed,pictorial, body,andsignstructures,andtheutilizationofdialect through elective corespondence frameworks, for example,web-basedlife,PCs,andiPads).

Discourseisueisextensivelydelegatedtakesafter: Variousauxiliaryisuescancausevoiceisue.Contact ulcers,growths,knobs,polyps,andsoforthareaportion

ofthenotablebasicsorescausingvoiceisue.These soresaditionalycausetormentinthethroatdistrict. Thesituationofthesoresinfluencesthenatureofvoice delivered.Thisiscaledasstructuralvoicedisorder.

FunctionalVoiceDisordersiscausedwhenmusclesof thelarynxmaynotworklegitimately,withnospecific auxiliaryornaturalreason.Theflawedutilizationofthe musclesmayresultbecauseofunreasonablepresure. Along these lines, utilitarian voice isue is comprehensivelynamedasmusclepresuredysphonia, orMTD.Theterm likewiseincorporatesaconditionin whichthevoicesoundstypical,yetaimsagony,distres orexhaustiontothevoice.

Neurogenicvoicedisordersarevoiceisuecausedby isuesinthefocalorfringesensorysystemsfalunder thisclas.Los ofmotionoftheentirebody,facialos ofmotion,hemiplegia and comparable neurological conditionsinfluencesthenervesofthelarynx,whichthus bothersvoicecreation.Los ofmotionofthevocalfolds causesfe ble,raspyorunpleasantvoice oronlya whisper.

Conversion dysphonia isa voice isue comingabout becauseofphysicalap earanceofamentalinjuryor strife.Theindividualmayhaveencountered\orse na stuningocasion,forexample,mishaporpasingofa dearone,orsexualmanhandle.Practicaltreatmentfor voicecontrolwilbeefectivejustifthehidenmental conditionisinvestigated.Thisisknownaspsychogenic voicedisorder.

Strategiesforlanguagetrainingisextensivelydelegated demonstratedasfolows: Sensoryfe dback causesyourkid to wind up more mindfulofthesoundsthatheiscreatingandhowheis deliveringthem.Forinstance,theSLPmayutilizesound-

related input.She mayrecord an example ofyour youngster'svocalizationsandplaythem back.Shewould then playa chronicle ofsimilarsounds,articulated efectively,withthegoalthathecanhearthedistinction. Youryoungstercanlikewiseprofitbyvisualcriticism or watchinghimselforothersdeliversounds.

Articulationspe chtherapyactivitiesnotwithstanding materialinput,youryoungster'slanguageteacherwil utilize other language instruction exercises for enunciationtoenablehim toarticulatecertainsounds. TheSLPmayobviouslyexpres awordandaskthatyour adolescent rehash it.This wil be done diferent circumstancestogiveyourkidalotoftraining.

Oralmotortherapy techniques are some discourse isuesportrayed byporcontrolofthe oralengine muscles.Thiscanmedlewithyourkid'sdiscourse,as welashisgulpingandsustainingcapacitiesalso.Oral engineactivitiesandfacialbackrubmightbeutilizedto enhance muscle tone.The language instructor may likewisedealwithpresentingsustenancesofdiferent surfacesandtemperatureske pinginmindtheendgoal to raise oral mindfulnes.Oral engine treatment proceduresregularlyacentuate language instruction exercisesthatmatchdevelopmentswithsounds.For instance,yourkidmaybe requestedtoconsiderthe adjustingofhislipsashecreatesthe"sh"sound.

Language Intervention are spe ch and dialectisue canotjustafectakid'scapacitytoverbalizeplainly, yetaditionalyhiscapacitytocomprehenddialectand communicateproperly.Afewchildrenmayhavetrouble articulatingtheirmusings,folowingorgivingheadings, orportrayingastory.Youryoungster'sSLPwilenergize dialectsecuringbyutilizingboks,pretendamusements, andcomparativelanguagetrainingexercises.[1] Thesoftwaredesignedisagametohelpchildrenhaving spe chdisordersbyaskingthem to repeatthe pre- recorded words.The score is produced acordingly. Contingentonthescoreofthechild,therewilbean advanceinthegame.

1. METHODOLOGY

   Thisspe chtherapysoftwareisagamewhichfocuseson thechildsdevelopmentinspe ch.Ittrainschildrenby askingthemtorepeatprerecordedwordsbyalowingthe childtolistentotheprerecordedwordsfirstandthen repeatitindividualy.then,oncethechildispretygod ateach word,the score ofthe child in the game increases.bubleblasterisagamewhichisusedinthis projectwhichisplayedbythechildwhichisaninteractive gamethatmakesthelearningproces moreinteresting andfunforthechild.Thesteps,includerecording,noise removal,extractingfrequency,generatinghistogramand comparison.

   1. RecordingandNoiseremoval:

      Theaudioofthepatientisrecordedusingpyaudio libraryinPython.Therecordingstartswhentheuser clickstherecordbutonandcontinuesrecordinguntil theuserstopsspeaking.Therecordingautomaticaly stopswhenitdetectssilence.Noiseremovalisdoneon silentchunksofaudiodata.Thesesilentchunksorwind noisefrequenciesarereplacedbyNaN(Notanumber) values.

   2. FrequencyExtraction:

      The numberofocurencesofa repeatingevent forunittimeiscaledasfrequencyalsoreferedas temporal frequency.It emphasizes the contrast to spatialfrequency and angular frequency.It is an importantparameterusedinthefieldofscienceand engine ring to specify the rate ofoscilatoryandvibratoryphenomena, such as mechanical vibrations,audio signals(sound),radio waves,andlight.

      In communication,the usable voice frequency band rangesfromaround30 Hzto340 Hz.Itisthusthatthe ultra-lowfrequencybandoftheelectromagneticrange somewhere in the range of 30 and 300 Hz is aditionalyaluded to asvoice frequency.The data transfercapacity ap ortioned fora solitude voice- frequency transmision chanelis normaly 4 kHz,

      includingguardbands,permitingasamplingrateof8 kHztobeutilizedasthepremiseofthepulsecode modulationsystemutilizedforthedigitalPSTN.Perthe Nyquist Shanon sampling theorem,the sampling frequency(8kHz)mustbeatleastdoublethehighest component of the voice frequency through proper filteringpriortosamplingatdiscretetimes(4kHz)for efectivereconstructionofthevoicesignal.

      Usingthepyaudiolibrary,frequencycanbeextracted froma16-bitmonochanel.wavfilewithapresetchunk size.

      Thefrequencyextractioncodehasaformulausingwhich thefrequencyiscalculatedforeachchunk(frame)of audiodata.

      Theextractedfrequencyisap endedtoanaraychunk bychunk.

      Formula:

      x1=(y2-y0)*.5/(2*y1-y2-y0)

      TheFTofthedatatimesawindow(ablackmanwindow), theFTissquared,thebinthathas thehighestvalueisfoundandaquadraticinterpolation aroundthepeakisusedusingthelog ofthemaxvalueanditstwoneighbouringvaluestofind thefundamentalfrequency.

      The acuracy can be increased or decreased by increasingoreducingthechunksizerespectively.

      Thechunksizeshouldbeamultipleof2tomakefuluse oftheFT.

      Thisproces isrepeatedforaltheprerecordedaudio samplespresentinthesoftwaredatabase.Theaverage ofthesearaysiscalculatedandstoredinanewaray. Now,thechildsaudioisused.Thefrequencyfrom this fileisextractedandstoredinanaray.

   3. Histograms:

      Histogramisusedtofindthenumberoffequencies thatocurbetwe n some presetranges.Matplotlib Libraryis used to getthe histogram features.The frequencyaraysaretheinputsforthehistogram code andthehistogram isploted.The2-histogram datais furtherstoredin2arays.

      Histogram is used to represent the numeric data acurately.Itisdiferentfromabargraph,whilethebar graphrelatestwovariables,histogram relatesonlyone variable.Inordertoconstructahistogram,thefirststep isto"bin"therangeofvalues,whichistodividethe entirerangeofvaluesintoaseriesofintervalswhich ne dnotnecesarilybeofsamesizeandthencounthow manyvaluesfalintoeachinterval.Ifthebinsareof equalsize,wehavearectanglewhichisformedoverthe binwithheightproportionaltothefrequency.Ifthebins

      arenotbeofequalwidth,wehavearectangleformed whichisdefinedtohaveitsareaproportionaltothe frequency.Theverticalaxisisfrequencydensityandnot frequency,where frequencydensityisthe numberof casesperunitofthevariableonthehorizontalaxis.

   4. Comparisonoftwohistograms:

   The2histogram araysarecompared,andascoreis generateddependingonhowacuratethevaluesare. This score wilbe used to telthe child his/her performance.

2. RESULTANDDISCUSION

   Thedatasetispreparedbyrecordingtheselectedwords, listed as,Ap le,Mango,Orange,Pineap le,Banana, Watermelon,Grapes,Strawbery,Chery,Papaya.Five speakersofage-group5to15areuteredtwosamples ofeachword,totalyconstituting14samplesofeach word.Themeanofthefrequencyfornormalspeakerand patientare,296.42and241.72respectively.Afteralthe background calculations and computations are completed,thescoregetsdisplayedonthedisplayas shownasinFig.1.

   Thecomputedfrequencyiscompareswiththethreshold frequency.Ifthefrequencyisabovethreshold,thenthe childgetstoplaythebubleblastergame.Ifthechild wantstotrythegameagainthenhe/shewilhavetoredo thespe chtherapyandthegamecontinues.

   Fig.1Frequencydistributionofthepatientafterperforming spe chtherapy

   Whentheplaybutonisclicked,afile(fruitname)is randomlyselectedandplayed.Onclickingtherecord buton,therecordingbeginsandrecordsuntilsilenceis detected.Beforesavingtherecordedfile,thenoiseis removed.Histogramsareplotedforalthe14sample filesoftherandomlyselectedfruitandtheaverage histogram isplotedusingthe14histograms.Histogram isthenplotedforthepatientsvoice(therecordedfile) andboththehistogramsarecomparedwithavariation of+/-3.Ifthe diference isles than 3 then itis consideredassuces forthatrange.Thepercentageis calculatedconsideringthetotalnumberofsuces.If thescorepercentageisgreaterthan70thenthegame opens.Once the patientlosesthe game,he getsa chancetoretryandimprovehisscoreandthisproces repeats.

   Letusconsideratestrunofthissoftware.Folowingare theresultswegotfromthetestrun.

   Fig.2Frequencydistributionofpatientwithoutdisorder

   Fig.4Averagefrequencydistributionofprerecordedaudiosample

   Fig.3Frequencydistributionofpatientwithdisorder

   1. CONCLUSION

      Wecanusevoicebandsandothervoicefeaturestotrain achildwithspe chdisorderbywiththehelpofanother human.

      Atpresent,apatientmustbepersonalygivenspe ch therapyandthereisaconstraintoftimewhichleadsto some patientsnotreceivingspe chtherapyproperly. Spe ch therapysoftware wilhelp manypatientsto improvetheirspe ch(speakingskils).A patientcan trainthemselvestoimprovetheirspe chataconvenient time.Spe chtherapycanbedoneforfre ofcost.

   2. REFERENCES

1. TheParentswebsite.[online].Available: htps:/w w.parents.com/kids/development/learning- disabilities/what-is-spe ch-therapy/

2. American Spe ch-Language-Hearing Asociation(ASHA) website. [Online].Available:htps:/w w.asha.org/practice- portal/clinical-topics/voice-disorders/