mirror of
https://github.com/rhasspy/espeak-ng.git
synced 2026-10-31 05:35:35 +00:00
579 lines
9.4 KiB
Plaintext
579 lines
9.4 KiB
Plaintext
// This file is UTF8 encoded
|
||
// Spelling-to-phoneme words for Latvian
|
||
|
||
// names of Latvian letters
|
||
a ,a_:
|
||
ā ,a::_!
|
||
b b,e:_:
|
||
c ts,e:_:
|
||
č tS,e:_:
|
||
d d,e:_:
|
||
e ,e_:
|
||
ē ,e::_!
|
||
f ,ef:_:
|
||
g g,a:_:
|
||
ģ J,e:_:
|
||
h h,a:_:
|
||
ḩ h,e:_:
|
||
i ,i_:
|
||
ī ,i::_!
|
||
j j,e:_:
|
||
k k,a:_:
|
||
ķ c,e:_:
|
||
l ,el:_:
|
||
ļ ,el^:_:
|
||
m ,em:_:
|
||
n ,en:_:
|
||
ņ ,en^:_:
|
||
o ,uo_:
|
||
ō ,o::_!
|
||
p p,e:_:
|
||
q k,u:_:
|
||
r ,eR::_!
|
||
ŗ ,er::_!
|
||
s ,es::_!
|
||
š ,eS::_!
|
||
t ,te:_:
|
||
u ,u_:
|
||
ū ,u::_!
|
||
v v,e:_:
|
||
w d,ubv,e:_:
|
||
x ,iks_:
|
||
y ,ig,Rek_:
|
||
z z,e:_:
|
||
ž Z,e:_:
|
||
|
||
// diphthong consonants
|
||
dz dze:
|
||
dž dZe:
|
||
|
||
// numbers
|
||
_0 n'ulle
|
||
_1 v'iens
|
||
_2 d'ivi
|
||
_3 tRi:s
|
||
_4 tS'etRi
|
||
_5 p'ietsi
|
||
_6 s'eSi
|
||
_7 s'eptin^i
|
||
_8 'astuon^i
|
||
_9 d'evin^i
|
||
_10 d'esmit
|
||
_11 v'ienpadsmit
|
||
_12 d'ivpadsmit
|
||
_13 tR'i:spadsmit
|
||
_14 tS'etRpadsmit
|
||
_15 p'ietspadsmit
|
||
_16 s'eSpadsmit
|
||
_17 s'eptin^padsmit
|
||
_18 'astuon^padsmit
|
||
_19 d'evin^padsmit
|
||
_2X d'ivdesmit
|
||
_3X tR'i:sdesmit
|
||
_4X tS'etRdesmit
|
||
_5X p'ietsd,esmit
|
||
_6X s'eSdesmit
|
||
_7X s'eptin^desmit
|
||
_8X 'astuon^desmit
|
||
_9X d'evin^desmit
|
||
_0C s'imti_
|
||
_1C s'imts
|
||
_0M1 t'u:kstuo:Si
|
||
_1M1 t'u:kstuo:tis
|
||
_0M2 m'iljo:ni
|
||
_1M2 m'iljo:ns
|
||
_dpt k'omats_
|
||
|
||
// ordinal numbers
|
||
_ord ais // default ending
|
||
_1o p'iRmais
|
||
_2o 'uotRais
|
||
_3o tR'eSais
|
||
_4o ts'etuRtais
|
||
_5o p'iektais
|
||
_6o s'estais
|
||
_7o s'epti:tais
|
||
_8o 'astuotais
|
||
_9o d'evi:tais
|
||
_0Co s'imtais
|
||
|
||
|
||
// accent names
|
||
_lig l'igatu:Ra
|
||
_acu 'aku:ts
|
||
_ac2 d'ubultaku:ts
|
||
_brv 'i:suma
|
||
_ced ts'edi:la
|
||
_cir d'iakRitiska:
|
||
_dia 'umlauts
|
||
_dot p'unkts
|
||
_grv gR'a:v
|
||
_hac h'atSeks
|
||
_mcn m'akRon
|
||
_ogo 'ogonek
|
||
_rng R'ing
|
||
_stk str'o:k
|
||
_tld t'ilde
|
||
|
||
|
||
// character names
|
||
_cap lielais
|
||
_?A buRts
|
||
_?? Rakstzi:me
|
||
_#9 tabulato:Rs
|
||
_#32 tukSums
|
||
|
||
_! izsaukumzi:me
|
||
_" pe:din^as
|
||
_# resti:te
|
||
_' apostro:fs
|
||
_( 'iekava
|
||
_) 'iekava||ts'iet
|
||
_, komats
|
||
_- defise
|
||
_– mi:nuss
|
||
_— duomuzi:me
|
||
_. punkts
|
||
_: ko:ls
|
||
_; semiko:ls
|
||
_< m'aza:ks||p'aR
|
||
_> l'iela:ks||p'aR
|
||
_? jauta:jumzi:me
|
||
_@ at
|
||
_[ kv'adRa:tiekava
|
||
_] kv'adRa:tiekava||ts'iet
|
||
_^ sakne
|
||
__ pasvi:tRuojumzi:me
|
||
_` 'gravis
|
||
_{ f'igu:Riekava
|
||
_| sta:vsvi:tRa
|
||
_} f'igu:Riekava||ts'iet
|
||
_~ tilde
|
||
_¦ s'adali:ta||st'a:vsvi:tRa
|
||
_« p'e:din^as
|
||
_´ aku:ts
|
||
_» p'e:din^as||ts'iet
|
||
_¿ 'apgRiezta:||j'auta:jumzi:me
|
||
_‒ duomuzi:me
|
||
_‖ d'ubultsta:st,a:vsvi:tRa
|
||
|
||
= _viena:ds_! $max3
|
||
* zvaigzni:te_! $max3
|
||
% pRotsenti_! $max3
|
||
‰ pRomiles_!
|
||
+ plus_! $max3
|
||
@ at_:
|
||
& _!%un
|
||
/ sl,i:psvi:tRa_! $u $max3
|
||
\ ,apve:Rsta:||sl,i:psvi:tRa_! $u $max3
|
||
$ dola:Ri
|
||
€ eiRo:
|
||
& un
|
||
¢ tsenti
|
||
£ ma:Rtsin^as
|
||
¥ je:nas
|
||
₨ Ru:pi:jas
|
||
§ sektsija
|
||
¶ paRagRa:fs
|
||
© auto:Rtiesi:bas
|
||
® ReJistRe:ts
|
||
° gRa:di
|
||
± plusmi:nuss
|
||
µ mikRo:
|
||
¼ ts'etuRtd,al^a
|
||
½ puse
|
||
¾ tR'i:s||ts,etuRtdal^as
|
||
× Reiz
|
||
÷ dali:ts
|
||
♯ die:zs
|
||
♭ bemo:ls
|
||
|
||
// unstressed words and/or with pause
|
||
aiz $u
|
||
apakš $u
|
||
ap $u
|
||
arī $u+ $brk
|
||
ar $u+
|
||
atkal $u+
|
||
bet $u+ $pause
|
||
bez $u+
|
||
bija $u+
|
||
būt $u+
|
||
būs $u+
|
||
caur $u+
|
||
cik $u+
|
||
dēļ $u+
|
||
gan $u+ $brk
|
||
gar $u
|
||
iekš $u
|
||
ir $u+
|
||
iz $u
|
||
ja $u+
|
||
jau $u+
|
||
jeb $u
|
||
jo $u
|
||
kad $u+
|
||
kam $u
|
||
kas $u+
|
||
ka $u
|
||
kā $u+ $brk
|
||
kur $u+
|
||
kopš $u
|
||
ko $u+
|
||
labad $u
|
||
lai $u+
|
||
līdz $u+
|
||
ļoti $u+
|
||
nav $u+
|
||
ne $u+
|
||
nedz $u $brk
|
||
ne $u+ $brk
|
||
nevis $u $brk
|
||
no $u+
|
||
nu $u+
|
||
par $u
|
||
par $u+
|
||
pa $u
|
||
pār $u
|
||
pēc $u+
|
||
pat $u
|
||
pie $u
|
||
pirms $u+
|
||
piem $u $pause
|
||
piemēram $u $pause
|
||
protams $u+
|
||
pret $u+
|
||
priekš $u
|
||
starp $u
|
||
šo $u+
|
||
taču $u $pause
|
||
tad $u+
|
||
tas $u+
|
||
tagad $u+
|
||
te $u+
|
||
tikt $u+
|
||
tiek $u+
|
||
tika $u+
|
||
tikai $u+
|
||
tiklab $u $brk
|
||
tomēr $u $pause
|
||
turpretim $u $pause
|
||
turpretī $u $pause
|
||
un $u $pause
|
||
uz $u
|
||
vai $u+ $brk
|
||
vairs $u
|
||
var $u+
|
||
vien $u $brk
|
||
virs $u
|
||
zem $u
|
||
|
||
// pronouns
|
||
es $u+
|
||
jums $u+
|
||
jūsos $u+
|
||
jūs $u+
|
||
jūsu $u+
|
||
kādai $u+
|
||
kādam $u+
|
||
kādas $u+
|
||
kāda $u+
|
||
kādā $u+
|
||
kāds $u+
|
||
kādu $u+
|
||
kurai $u+
|
||
kuram $u+
|
||
kuras $u+
|
||
kura $u+
|
||
kurā $u+
|
||
kurš $u+
|
||
kuru $u+
|
||
kuŗai $u+
|
||
kuŗam $u+
|
||
kuŗas $u+
|
||
kuŗa $u+
|
||
kuŗā $u+
|
||
kuŗš $u+
|
||
kuŗu $u+
|
||
manis $u+
|
||
mani $u+
|
||
manī $u+
|
||
man $u+
|
||
mēs $u+
|
||
mums $u+
|
||
mūsos $u+
|
||
mūs $u+
|
||
mūsu $u+
|
||
pašai $u+
|
||
pašam $u+
|
||
pašas $u+
|
||
paša $u+
|
||
pašā $u+
|
||
pašu $u+
|
||
pati $u+
|
||
pats $u+
|
||
savai $u+
|
||
savam $u+
|
||
savas $u+
|
||
sava $u+
|
||
savā $u+
|
||
savs $u+
|
||
savu $u+
|
||
šai $u+
|
||
šajai $u+
|
||
šajā $u+
|
||
šim $u+
|
||
šis $u+
|
||
šīs $u+
|
||
šī $u+
|
||
šo $u+
|
||
tai $u+
|
||
tajās $u+
|
||
tajā $u+
|
||
tajos $u+
|
||
tam $u+
|
||
tas $u+
|
||
tām $u+
|
||
tās $u+
|
||
tā $u+
|
||
tevis $u+
|
||
tevi $u+
|
||
tevī $u+
|
||
tev $u+
|
||
tiem $u+
|
||
tie $u+
|
||
tik $u+
|
||
tos $u+
|
||
to $u+
|
||
tu $u+
|
||
viņai $u+
|
||
viņam $u+
|
||
viņas $u+
|
||
viņa $u+
|
||
viņām $u+
|
||
viņās $u+
|
||
viņā $u+
|
||
viņiem $u+
|
||
viņi $u+
|
||
viņos $u+
|
||
viņš $u+
|
||
viņus $u+
|
||
viņu $u+
|
||
|
||
|
||
// exception words with stress on 2nd syllable
|
||
aizvien $2
|
||
arvien $2
|
||
joprojām $2
|
||
kautko $2
|
||
labdien $2
|
||
labrīt $2
|
||
labvakar $2
|
||
mazliet $2
|
||
nekad $2 $u+
|
||
nekam $2 $u+
|
||
nekas $2 $u+
|
||
nekā $2 $u+
|
||
neko $2 $u+
|
||
nekur $2 $u+
|
||
nemaz $2 $u+
|
||
pagalam $2
|
||
paldies $2
|
||
palaikam $2
|
||
papilnam $2
|
||
paretam $2
|
||
patiešām $2
|
||
patlaban $2
|
||
pavisam $2
|
||
tāpat $2
|
||
tepat $2
|
||
tikpat $2
|
||
tikvien $2
|
||
turpat $2
|
||
turpretī $2
|
||
vismaz $2
|
||
|
||
// 1st word unstressed, 2nd word stressed
|
||
(kaut cik) $2 $u+
|
||
(kaut kad) $2 $u+
|
||
(kaut kam) $2 $u+
|
||
(kaut kas) $2 $u+
|
||
(kaut kā) $2 $u+
|
||
(kaut kāda) $2 $u+
|
||
(kaut kādai) $2 $u+
|
||
(kaut kādam) $2 $u+
|
||
(kaut kādas) $2 $u+
|
||
(kaut kādā) $2 $u+
|
||
(kaut kāds) $2 $u+
|
||
(kaut kādu) $2 $u+
|
||
(kaut ko) $2 $u+
|
||
(kaut kur) $2 $u+
|
||
(nezin kad) $3 $u+
|
||
(nezin kur) $3 $u+
|
||
(nez kad) $2 $u+
|
||
(tai pašai) $2 $u+
|
||
(tai pašā) $2 $u+
|
||
(tai tur) $2 $u+
|
||
(tajā pašā) $2 $u+
|
||
(tajā tur) $3 $u+
|
||
(tam pašam) $2 $u+
|
||
(tam tur) $2 $u+
|
||
(tas pats) $2 $u+
|
||
(tas tur) $2 $u+
|
||
(tā paša) $2 $u+
|
||
(tā pati) $2 $u+
|
||
(tās pašas) $2 $u+
|
||
(tās tur) $2 $u+
|
||
(tā tur) $2 $u+
|
||
(tik pat) $2 $u+
|
||
(to pašu) $2 $u+
|
||
(to tur) $2 $u+
|
||
|
||
// stress on last syllable
|
||
neparko $3
|
||
itnekur $3
|
||
itnemaz $3
|
||
|
||
//stress on 2nd syllable (international words)
|
||
adadžo $2
|
||
kakao $2
|
||
maestro $2
|
||
piano $2
|
||
solfedžo $2
|
||
fiasko $2
|
||
|
||
//stress on last syllable (international words)
|
||
ateljē $3
|
||
dekoltē $3
|
||
foajē $3
|
||
frikasē $3
|
||
galifē $3
|
||
kanapē $3
|
||
komunikē $4
|
||
portjē $2
|
||
protežē $3
|
||
|
||
// Specific words which has bad default spelling
|
||
tvnet te:ve:net
|
||
|
||
///////////////////
|
||
// Abbreviations //
|
||
///////////////////
|
||
a/s $abbrev
|
||
as $abbrev
|
||
b-ba biedRi:ba
|
||
dr $abbrev
|
||
dz $abbrev
|
||
lp $abbrev
|
||
ls $abbrev
|
||
lu $abbrev
|
||
lvu $abbrev
|
||
na $abbrev
|
||
nr $abbrev
|
||
pr-ks $abbrev
|
||
psrs $abbrev
|
||
rpi $abbrev
|
||
sia sia: $pause
|
||
st $abbrev
|
||
sv $abbrev
|
||
šķ $abbrev
|
||
tml te:em,el: $u $pause
|
||
utml ute:e,mel: $u $pause
|
||
utt ute:te: $u $pause
|
||
|
||
// words with o
|
||
adadžo $alt $2
|
||
aikido $alt
|
||
ambo $alt
|
||
anno $alt
|
||
automoto $alt
|
||
bandžo $alt
|
||
bendžo $alt
|
||
bingo $alt
|
||
bolero $alt
|
||
bordo $alt
|
||
borneo $alt
|
||
bravo $alt
|
||
bruno $alt
|
||
bruto $alt
|
||
depo $alt
|
||
dingo $alt
|
||
domino $alt
|
||
džudo $alt
|
||
eldorado $alt
|
||
embargo $alt
|
||
esperanto $alt
|
||
flamenko $alt // can't use "f" as indicator, as 6 letters between are to far from "o"
|
||
indigo $alt
|
||
inkognito $alt
|
||
intermeco $alt
|
||
kakao $alt
|
||
kastro $alt
|
||
kazino $alt
|
||
kimono $alt
|
||
kioto $alt
|
||
kolombo $alt
|
||
kolorado $alt
|
||
kolorādo $alt
|
||
kongo $alt
|
||
kvatročento $alt
|
||
lego $alt
|
||
leporello $alt
|
||
lesoto $alt
|
||
loto $alt
|
||
lumbago $alt
|
||
maestro $alt
|
||
majordomo $alt
|
||
mecotinto $alt
|
||
meteo $alt
|
||
metro $alt
|
||
monako $alt
|
||
mono $alt
|
||
montekarlo $alt
|
||
montevideo $alt
|
||
moto $alt
|
||
nato $alt
|
||
odo $alt
|
||
ontārio $alt
|
||
oregano $alt
|
||
orinoko $alt
|
||
oslo $alt
|
||
otello $alt
|
||
oto $alt
|
||
panno $alt
|
||
pianīno $alt
|
||
pikolo $alt
|
||
polo $alt
|
||
pončo $alt
|
||
porto $alt
|
||
puertoriko $alt
|
||
rančo $alt
|
||
retro $alt
|
||
riodežaneiro $alt
|
||
rodeo $alt
|
||
rodrigo $alt
|
||
rokoko $alt
|
||
roks $alt
|
||
rondo $alt
|
||
saldo $alt
|
||
sambo $alt
|
||
santodomingo $alt
|
||
siroko $alt
|
||
solo $alt
|
||
sombrēro $alt
|
||
sonera $alt
|
||
sorgo $alt
|
||
stereokino $alt
|
||
storno $alt
|
||
tobago $alt
|
||
toronto $alt
|
||
triko $alt
|
||
trimo $alt
|
||
uno $alt
|
||
veto $alt
|
||
čello $alt
|
||
ūdenspolo $alt
|
||
žabo $alt
|
||
žigolo $alt
|
||
|