15
In e na ional Jou nal o Ad ance and Applied Resea ch
www.ijaa .co.in
ISSN – 2347-7075
Impac Fac o – 8.141
Pee Re iewed
Bi-Mon hly
Vol. 6 No. 38
Sep embe - Oc obe - 2025
A Re iew o Cu en Techniques in Tex - o-Speech Syn hesis
Mahesh S. Gaikwad1 & D . Sanjay T. Wani2
1Womens College o Home Science & BCA, Loni, India
2Womens College o Home Science and BCA, Loni, India
Co esponding Au ho – Mahesh S. Gaikwad
DOI - 10.5281/zenodo.17309810
Abs ac :
Speech is he mos p ominen and na u al o m o communica ion be ween human beings.
Only a selec ew who a e li e a e in a gi en language can access he majo i y o he digi al ma e ial
a ailable oday. The applica ion o NLP p o ides solu ions in he o m o na u al in e aces, such as
TTS, allowing digi al con en o each a wide audience and acili a e he exchange o in o ma ion
among di e en people. A a ie y o echnologies, including ex - o-speech echnology, ha e been
de eloped o enhance language lea ning. The echnique o u ning ex w i en in na u al language
in o oice is called ex - o-speech syn hesis. This e iew pape includes an o e iew o ex - o-speech
sys ems de eloped o Indian languages. The wo k done in he speech domain o he Indian language
is a he p ima y s age. The esea ch wo k o Indian languages is also ca ied ou howe e , he wo k
is no able o co e he comple e phone ic a ia ion o he language.
Keywo ds: Tex - o-speech, Indian languages, Techniques, speech syn hesis.
In oduc ion:
Speech is he mos p ominen and
na u al o m o communica ion be ween
humans being. Speech has he abili y o
exp ess one’s hough s by means o a se o
signs, whe he acous ic, musical, g aphical,
ges u al. Today’s mos in o ma ion in digi al
o m is accessible o a ew who can ead o
unde s and a pa icula language. Language
echnologies can gi es solu ions in he o m o
na u al in e aces so he digi al o m con en
can each o he masses and acili a e he
exchange o in o ma ion ac oss di e en
people.
Humans ha e been d i en o de elop
compu e s wi h human-like comp ehension
and speech. In his ega d, scien is s ha e
wo ked o c ea e a sys em o speech signal
analysis and ca ego iza ion. Since, 1960s
compu e scien is s ha e been esea ching
ways and means o make compu e eco d,
in e p e and unde s and ex and con e ed
in o human speech.
A Tex - o-Speech (TTS) con e s a
aw ex in o human speech sounds. I can be
powe ul assis ance o communica ion o
isually impai ed people and also in
elecommunica ion, indus ial and educa ional
applica ions. Go e nmen o India commence
de elopmen o TTS sys ems o Indian
languages h ough TTS conso ium p ojec
unde he Minis y o Elec onics and
In o ma ion (Mei Y) [6]. So many ins i u ions
ha e been wo king on speech syn hesis such
IIIT-H, CDAC- Mumbai, SDAC-Pune and
IIT-Mad as in a ious languages. They ha e
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
16
buil applica ions as e-speak, a-speak, Sandesh
Pa hak.
The amoun o wo k o Indian
languages in Speech syn hesis has no ye
eached o a c i ical le el o be used as eal
communica ion ool, as ha in o he languages
o de eloped coun ies. Fo de eloping a eal
communica ion ool he speech syn hesis
sys em should con inuous syn hesis he ex
in o speech like human being.
Me hodology:
Figu e 1: Gene al me hodology o Speech
Syn hesis sys em.
Techniques o Speech Syn hesis:
A icula o y Syn hesis:
A icula o y syn hesis models he
physical a icula o s such as lips, jaws, ongue,
so pala e and so on [15]. In his human
speech p oduc ion sys em is modelled. I
in ol es simula ing he acous ic pa s o ocal
ac and i s dynamic mo emen . The
command pa ame e s a e sub-glo al p essu e,
ocal co d ension and he ela i e loca ion o
di e en a icula o y o gans. I p oduces
in elligible syn he ic speech bu i is a om
na u al sound and hence no widely used.
Fo man Syn hesis:
This is a ule based syn hesis
echnique, which desc ibes he esonan
equencies o he ocal ac . This me hod
uses sou ce- il e model o language ou pu .
The pa ame e s con olling he equency
esponse o he ocal ac il e and hose
con olling he sou ce signal a e upda ed a
each phoneme. Exci a ion p oduced by he
oo passes h ough he il e , is quali ied by
he esonance cha ac e izes o he il e o
c ea e language.
Hidden Ma ko Model (HMM):
The Hidden Ma ko Model is use
s a is ical models o cha ac e ize he sequence
o speech spec a and ha e success ully been
applied o speech syn hesis sys ems. This
sys em simul aneously model spec um,
exci a ion and con inuance o speech using
con en dependen HMMs and gene a es
speech wa e o ms. HMM c ea es s ochas ic
model om known u e ances and compa es
he p obabili y ha he unknown u e ance can
be engende ed by each model. This app oach
gi es good p osody ea u es wi h na u al
sound language.
Conca ena i e Syn hesis:
Conca ena i e syn hesis is he mos
uncomplica ed me hod o syn hesize he
speech which is go by conca ena ing he
di e en sen ences, wo ds, syllables, phones,
diphones and iphones. These a e al eady
s o ed o ge he desi ed ou pu language. I
equi es la ge da abases some imes i is qui e
impossible o s o e. This echnique p oduces
mo e na u al sound.
The e a e h ee sub ype o Conca ena i e
syn hesis
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
17
Uni Selec ion Speech Syn hesis (Co pus
Based Syn hesis):
Uni selec ion syn hesis uses la ge
da abase. Du ing da abase c ea ion, each
eco ded u e ance is segmen ed in o some
indi idual phones, syllables, mo phemes,
wo ds, ph ases and sen ences. An index o he
uni s in he speech da abase is hen made
based on he segmen a ion and acous ic
pa ame e s such as undamen al equency,
pi ch, du a ion, he s a e o syllable and
p e ious and nex phones. This me hod
p o ides na u alness in ou pu speech as
compa ed o o he echniques.
Diphone Syn hesis:
This echnique equi es ewe
da abases as compa ed o he uni selec ion
syn hesis. I uses wo adjacen phones o make
he speech wa e o m. Bu his echnique
su e s h ough he p oblem o co-a icula ion.
Domain Speci ic Syn hesis:
Domain speci ic syn hesis means
ela ed o he speci ic ield associa ed
syn hesis. In his syn hesis da abase consis s o
language ela ed o speci ic line o business
and ha a e used o conca ena e o c ea e he
speech.
Tex Analysis:
Tex No maliza ion:
The i s s ep in all ex - o-speech is
no malizing he inpu ex in o sen ence hen
each sen ence has o be di ided in o a
collec ion o okens ha is in o m o wo ds,
numbe s da e and so on. Non-na u al language
okens as abb e ia ions and ac onyms
ans o med o na u al language okens. [14]
Sen ence okeniza ion:
This is he i s s ep o ex
no maliza ion. In sen ence okeniza ion has
some complica ions because sen ences no
always e mina ed by pe iod (.) and some ime
sen ence e mina ed by colon (:). To iden i y
sen ence bounda ies, he inpu ex is di ided
in o okens sepa a ed by whi espace and hen
any okens such as !, .,? is selec ed hen
decision can be dependen on machine
lea ning as hese n okens indica e in end-o -
sen ence o no .
Non-S anda d Wo ds:
Second s ep o ex no maliza ion is
no malizing non-s anda d wo ds ha a e
numbe s, da es, abb e ia ions o ac onyms.
These okens ha e o be ans o med o a
sequence o na u al wo ds. Abb e ia ion
dic iona ies a e used because o he in ica e
and unclea o ms o ac onyms and
abb e ia ions.
Homog aph Resolu ion:
Thi d s ep o ex no maliza ion is
homog aph esolu ion. A homog aph is he
wo ds ha ha e same sequence o cha ac e s
bu di e in p onuncia ion.
Accen :
A e ex no maliza ion he nex s ep
is o ind he p ope accen o each. The inpu
ex can include wo ds o names ha canno be
ound in he lexicon. The name-accen lexicon
is used by so many ex - o-speech sys ems.
The accen o unknown wo ds ha is no in
accen lexicon can be p oduced h ough
g apheme- o-phoneme con e sion me hod.
Phone ic Analysis (G apheme- o-Phoneme):
Phone ic alphabe s a e used in
phone ic analysis o ansla e o hog aphic
symbols in o phonological ep esen a ions.
P osodic Analysis
P osody means hy hm o speech,
s ess pa e ns and in ona ion. Na u alness
speech is ha ing he ce ain p ope ies o
speech signal ha is ela ed wi h audible
changes in pi ch, syllabic leng h and loudness
a e collec i ely called as p osody.
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
18
Li e a u e Re iew:
Sai Sawan e al. [1] in his esea ch pape he
esea che ha e used phoneme based
conca ena i e syn hesis o English language.
This sys em is implemen ed wi h he help o
MATLAB map da a s uc u e and Ma ix
ope a ions. In his sys em esea che
phone ically 42 wo ds in English language a e
eco ded and a e ha phoneme a e ex ac ed
using PRAAT ools. Ex ac ed phoneme
compa ed wi h inpu ex phoneme and hen
conca ena ed sequen ially o econs uc he
desi ed wo d. This me hod is simple and i
equi es less memo y. Speech quali y o his
syn hesis is mo e na u alness.
Sneha C. Mad e e al. [3] in his pape e iew
shows ha he esea che has used OCR o
ex ac ing ex om image and hen
ecognized cha ac e is con e ed in o audio
using MATLAB. This sys em is easy bu
p oblem whe e ine i able like making
empla es o e e y cha ac e wi h e e y on .
Resea che desc ibe sys em is specially used
o isually challenged people.
P a in M. Gha e e al. [4] his pape
esea che has used uni selec ion speech
syn hesis o Ma a hi language. Resea che
has uses combina ions o uni s like Syllable,
Wo ds and Ba akhadi as da abases. In his
syn hesis deeply ocused on wo ds co pus.
Resea che e alua ed ha speech quali y is
mo e han 80 pe cen na u alness.
Rupinde deep Kau e al. [5] his pape
e iew shows ha he esea che has
de eloped speech syn hesis o Punjabi
language using gene al app oach such as Tex
p ocessing, Linguis ic analysis, P osodic
p edic ion and Wa e o m gene a ion. In his
pape esea che has desc ibed gene al
a chi ec u e o TTS and di e en wa e o m
gene a ion me hod i.e Fo man syn hesis,
Conca ena i e syn hesis and S a is ical
pa ame e syn hesis.
I unuoluwa Isewon e al. [8] in his pape
esea che used NLP and digi al signal
p ocessing o designing speech syn hesis in
English Nige ian language. Resea che
desc ibe ha na u al language p ocessing
module p oduces a phone ic ansc ip ion o
ex ead oge he wi h p osody and hen
Digi al signal p ocessing module ans o ms
symbolic in o ma ion om NLP in o audible
speech . He e au ho c ea e he GUI based
applica ion o speech syn hesis.
Sang amsing Kay e e al. [9] in his pape
esea che has used Hidden Ma ko Model
based speech syn hesis o Ma a hi language.
Resea che desc ibes ha ou es o speech
pa ame e s a e gene a ed om he ained
Hidden Ma ko Models. This syn hesis
consis s o wo pa s as T aining and
Syn hesis. In aining pa , spec um and
exci a ion pa ame e s a e ex ac ed om he
anno a ed speech da abase and con e ed o a
sequence o obse ed ea u e ec o s which is
modelled by a espec i e sequence o HMM.
He e spec um pa ep esen ed by me -
ceps al coe icien s and del a-del a
coe icien . Del a-del a coe icien and LogF0
deno e he exci a ion po ion. In syn hesis pa ,
inpu ex is con e ed in o sequence o
con ex ual label hen as pe he label sequence
HMM sequence is cons uc ed by
conca ena ing con ex dependen HMM. A e
his me -ceps al coe icien and F0 ou es a e
gene a ed by using he pa ame e gene a ion
algo i hm. Speech wa e o m is syn hesized
di ec ly om he gene a ed mel-ceps al
coe icien s and F0 alues by using MLSA
il e . In his speech syn hesis, speech da abase
de eloped by IIT-Hyde abad. I consis s o 12
ype o 1000 sen ence o aining.
Sang amsing N. Kay e e al. [11] in his
pape esea che deals wi h a co pus d i en
Ma a hi TTS sys em based on conca ena i e
syn hesis. Fo implemen a ion o TTS in
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
19
Ma a hi he MATLAB 2014 has been used.
This syn hesis consis s o i s ex analysis
pe o med hen sen ence spli ing i.e.
Syllabi ica ion based on Linguis ic ules. A e
ha wa e o m conca ena ion p ocess, equi ed
syllables a e e ie ed om speech co pus
based on ex analysis and a anged o p oduce
he speech. Du ing he conca ena ion p ocess
o speech uni , he e will be gli ches in he
join . These a e emo ed wi h he help o
wa e o m smoo hing p ocess. Fo he
spec um smoo hing, use ime scale
modi ica ion me hod and PRAAT so wa e
used o calcula e du a ion alue o each
syllable. Also linea p edic i e coding used o
ep esen ing he spec al en elop o digi al
signal o speech.
Nilesh Fal Dessai e al. [13] in hese
esea che s ha e used Conca ena i e syn hesis
me hod o Konkani language. The quali y o
gene a ed speech depends on he uni size.
Resea che desc ibe ha wo d syn hesis is
be e han di-phone o phoneme syn hesis.
The na u alness and in elligibili y o any
conca ena ed oice syn hesis sys em a e
c ucial ea u es. They ha e pe o med lis ene s
es , i is obse ed ha he de eloped sys em
pe o ms be e han eSpeak in e ms o
uniqueness, na u alness and in elligibili y.
When we use small size o uni he quali y o
speech no good bu co e age will be mo e.
When we inc ease size o uni i inc eases he
quali y o syn hesized speech bu we can’
co e he whole language.
Sang amsing Kay e e al. [15] his pape
e iew shows ha he esea che has used
Conca ena i e speech syn hesis o Ma a hi
language. They had only wo ked on Ma a hi
language. In his pape esea che deeply
ocused on Di-phone. Resea che s use he
Diagnos ic Rhyme Tes (DRT) o compa e wo
syn hesize s: he Fes i al TTS and he MARY
TTS sys em. The na u alness o he
syn hesized speech o bo h he syn hesize s
needs o be enhanced. The na u alness esul s
a e no so good because di-phone da abase
con ains only one ins ance o each speech uni .
So esea che says ha di-phone based speech
syn hesis quali y is no good.
Nikisha Ja iwala e al. [18] in his esea ch
pape esea che ha e used conca ena i e
syn hesis me hod using MATLAB ool o
Guj a hi language. This sys em is mos ly
de eloped o isually impai ed people. Fo
speech syn hesis esea che de eloped syllable
co pus.
Ami Kuma Jha e al. [19] Resea che ha e
used conca ena i e syn hesis me hod o
Mai hili language. Mai hili language is
syllabic in na u e. Resea che de elops he
speech co pus as a syllable. Fo speech
enhanced na u alness pu pose au ho eco d
and s o e mos equen ly occu ing wo ds.
The quali y o syn hesized speech in e ms o
in elligibili y and na u alness is e alua ed o
app oxima ely 84 pe cen . The speech co pus
consis o 930 syllable (C * V) in o al. Each
posi ion has 300 syllables and 10 independen
owels. 930 uni s o speech da a is buil om
all h ee posi ions i.e. ini ial, middle and inal
o accoun o maximum possible phone ic
co e age.
S.D. Shi bahadu ka e al. [21] in his
esea ch pape esea che desc ibe TTS sys em
using conca ena i e syn hesis me hod o
Ma a hi language wi h uni selec ion speech
da abase. Resea che de eloped Ma a hi
speech syn hesize using di e en choice o
uni s as wo d, phonemes and syllables as a
da abase. The quali y o his syn hesized
speech in e ms o in elligibili y and
na u alness is e alua ed o app oxima ely 81
pe cen ages.
Soumi a Das e al. [24] in his esea che
p oposed speech syn hesize o Ma a hi
language using Syllabic app oach. Syllabic
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
20
app oach is di iding a wo d in o syllable uni s
o na u al speaking. In his echnique wo d
di ided in o syllabic clus e o ha i s ind
ou he co ec s uc u e o wo d. Accu acy o
speech using his app oach is good.
Saadin Oyucu e al. [25] in his esea ch
pape esea che de eloped speech syn hesize
using deep lea ning o Tu kish language. This
sys em handles complex inpu ex such as
image and ideo. He de eloped Tu kish
co pus and p oposed a Taco on 2 + HiFi-
GAN s uc u e o TTS sys em. Speech quali y
o his TTS sys em is bes as pe MOS sco e
4.49.
Conclusion:
Tex - o-speech is he applica ion o
NLP. This is a bes ool o lea ning. The wo k
ca ied ou in he speech domain o English
and o he Eu opean languages ha e achie ed
an accu acy o mo e han 85% - 90%
syn hesized a e. The speech syn hesize and
hei a ious echniques ha ha e been
examined in his pape . We de e mined ha a
la ge numbe o speech syn hesizing s udies
deal wi h o eign languages as compa ed wi h
Indian languages. A TTS sys em wi h se e al
speech syn hesis echniques oge he p oduces
a highe quali y esul .
Re e ences:
1. Sai Sawan and Mangesh Deshpande.
English Tex o Speech Syn hesize
Using Conca ena ion Technique.
Sp inge Na u e Singapo e P e L d.
2018 M. Singh e al. (Eds.): ICACDS
2018, CCIS 905, pp. 471–480, 2018.
2. E .Sheilly Padda, E . Nidhi, Ms.
Rupinde deep Kau . A S ep owa ds
Making an E ec i e Tex o speech
Con e sion Sys em. In e na ional
Jou nal o Enginee ing Resea ch and
Applica ions (IJERA) ISSN: 2248-9622
Vol. 2, Issue 2, Ma -Ap 2012, pp.1242-
1244
3. Sneha.C.Mad e and P o .S.B.Gund e.
OCR Based Image Tex To Speech
Con e sion Using MATLAB.
P oceedings o he Second In e na ional
Con e ence on In elligen Compu ing
and Con ol Sys ems (ICICCS 2018)
IEEE Xplo e Complian Pa Numbe :
CFP18K74-ART; ISBN: 978-1-5386-
2842-3
4. P a in M Gha e*1 and
S.D.Shi bhadu ka . SPEECH
SYNTHESIS USING SYLLABLE FOR
MARATHI LANGUAGE.
INTERNATIONAL JOURNAL OF
ENGINEERING SCIENCES &
RESEARCH TECHNOLOGY ISSN:
2277-9655 [Gha e * e al., 7(1):
Janua y, 2018]
5. Rupinde deep Kau , M . R.K. Sha ma
and M . Pa eek Kuma . BUILDING A
TEXT-TO-SPEECH SYSTEM FOR
PUNJABI LANGUAGE. Na a ajan
Meghana han e al. (Eds): ACSIT,
SIPM, FCST, CoNeCo, CMIT - pp. 71–
87, 2017.
6. TDIL TTS, “Indian Language
Technology P oli e a ion and
Deploymen Cen e: Tex - o-speech,”
a ailable: h p:// dil-dc.in/.
7. Tex o Speech Con e sion wi h
Phonema ic Conca ena ion. Tapas
Kuma Pa a, Biplab Pa a and
Puspanjali Mohapa a. In e na ional
Jou nal o Elec onics Communica ion
and Compu e Technology (IJECCT)
Volume 2 Issue 5 (Sep embe 2012)
8. I unuoluwa Isewon, Jelili Oyelade and
Olu unke Oladipupo. Design and
Implemen a ion o Tex To Speech
Con e sion o Visually Impai ed
People. In e na ional Jou nal o Applied
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
21
In o ma ion Sys ems (IJAIS) – ISSN:
2249-0868 Founda ion o Compu e
Science FCS, New Yo k, USA Volume
7– No. 2, Ap il 2014
9. Sang amsing Kay e , Monica Mundada
and D . Cha ansing Kay e. A Ma a hi
Hidden-Ma ko Model Based Speech
Syn hesis Sys em. IOSR Jou nal o
VLSI and Signal P ocessing (IOSR-
JVSP) Volume 5, Issue 6, Ve . I (No -
Dec. 2015), PP 34-39 e-ISSN: 2319 –
4200, p-ISSN No. : 2319 – 4197
10. M. Z. Rashad, Hazem M. El-Bak y,
Islam R. Isma'il and Nikos Mas o akis.
An O e iew o Tex -To-Speech
Syn hesis Techniques. ISSN: 1792-4316
ISBN: 978-960-474-207-3 July 2010
11. Sang amsing N. Kay e, Monica
Mundada , D . Cha ansing N. Kay e ,
D .Bha i Gawali. A i icially Gene a ed
o Conca ena i e Syllable based Tex o
Speech Syn hesis Sys em o Ma a hi.
IOSR Jou nal o VLSI and Signal
P ocessing (IOSR-JVSP) Volume 5,
Issue 6, Ve . II (No -Dec. 2015), PP
44-49 e-ISSN: 2319 – 4200, p-ISSN No.
: 2319 – 4197
12. P a in M. Gha e and Su esh D.
Shi bahadu ka , “Syllable-Based
Conca ena i e Speech Syn hesis o
Ma a hi Language”. Sp inge Na u e
Singapo e P e L d. 2019 S. Fong e al.
(eds.), In o ma ion and Communica ion
Technology o Compe i i e S a egies,
Lec u e No es in Ne wo ks and Sys ems
40, pp 615-624
13. Nilesh Fal Dessai, Gau a Naik, Jyo i
Pawa . : De elopmen o Konkani TTS
Sys em using Conca ena i e Syn hesis.
In e na ional Con e ence on Con ol,
Ins umen a ion, Communica ion and
Compu a ional Technologies
(ICCICCT) 2016 pp-344-348
14. Sai Gee a1 and B.L. Mu alidha a.:
Syllable as he Basic Uni o Kannada
Speech Syn hesis. IEEE 978-1-5090-
4442-9/17, 2017
15. Sang amsing Kay e , Monica Mundada ,
D . Cha ansing Kay e. : Di-phone-
Based Conca ena i e Speech Syn hesis
Sys ems o Ma a hi Language. IOSR
Jou nal o VLSI and Signal P ocessing
(IOSR-JVSP) Volume 5, Issue 5, Ve . I
(Sep - Oc . 2015), PP 76-81 e-ISSN:
2319 – 4200, p-ISSN No. : 2319 – 4197
16. Sang amsing Kay e, Ka i a Waghma e,
D . Bha i Gawali.: Ma a hi Speech
Syn hesis: A e iew, In e na ional
Jou nal on Recen and Inno a ion
T ends in Compu ing and
Communica ion, ISSN: 2321-8169-
3708 – 3711 June 2015
17. Sang amsing Kay e , Monica Mundada ,
Jayesh Guj a hi. : Hidden Ma ko
Model based Speech Syn hesis: A
Re iew, In e na ional Jou nal o
Compu e Applica ions (0975 – 8887)
Volume 130 – No.3, No embe 2015
18. Nikisha Ja iwala and Bankim Pa el. : A
Sys em o he Con e sion o Digi al
Guja a i Tex - o-Speech o Visually
Impai ed People., Sp inge Na u e
Singapo e P e L d. 2018 S. S. Ag awal
e al. (eds.), Speech and Language
P ocessing o Human-Machine
Communica ions, Ad ances in
In elligen Sys ems and Compu ing.
19. Ami Kuma Jha, Piyush P a ap Singh,
Pankaj Dwi edi. : Mai hili Tex - o-
Speech Sys em, IEEE 978-1-7281-
2472-8/19 2019
20. Suhas R. Mache, Manasi R. Bahe i, C.
Nam a a Mahende , Re iew on Tex -
To-Speech Syn hesize , IJARCCE,
ISSN (Online) 2278-1021, ISSN (P in )
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
Mahesh S. Gaikwad & D . Sanjay T. Wani
22
2319 5940 Vol. 4, Issue 8, Augus
(2015) p. 54 – 59
21. S.D.Shi bahadu ka , D S.Bo mane,
R.L.Kazi, Subjec i e and Spec og am
Analysis o Speech Syn hesize o
Ma a hi TTS Using Conca ena i e
Syn hesis, In e na ional Con e ence on
Recen T ends in In o ma ion,
Telecommunica ion and Compu ing,
978-0-7695-3975-1/10 (2010)
22. Jayesh Tanna, Vijay Sa ani, Ami
Degada, Tex -To-Speech (TTS)
Con e sion Sys em o Guja a i
Language, JoEDT, ISSN: 2229-6980
(2013) 1-10, STM Jou nals 2013.
23. [G. D. Ram eke1 and R. J. Ram eke,
Tex -To-Speech Syn hesize o
English, Hindi and Ma a hi Spoken
Signals, B i ish Jou nal o Applied
Science & Technology 15(3): 1-16,
ISSN: 2231-0843, Ma ch 2016
24. Das, S. (2023, Ap il 27). Syllabic
Speech Syn hesis o Ma a hi Language.
2023 1s In e na ional Con e ence on
Cogni i e Compu ing and Enginee ing
Educa ion (ICCCEE).
h ps://doi.o g/10.1109/icccee55951.202
3.10424461
25. Oyucu, S. (2023, Ap il 18). A No el
End- o-End Tu kish Tex - o-Speech
(TTS) Sys em ia Deep Lea ning.
Elec onics, 12(8), 1900.
h ps://doi.o g/10.3390/elec onics12081
900
