This chapter describes the linguistic resources included in the file
pes.zip of the lang folder.
The iso-639-3 code of Persian language is PES. The resources are
under-construction. Any help is welcome.
List of phonemes
Consonant Plosives
SPPAS
IPA
Description
Examples
b
b
voiced bilabial
d
d
voiced alveolar
k
k
voiceless velar
g
g
voiced velar
p
p
voiceless bilabial
q
q
voiceless uvular
t
t
voiceless alveolar
G\
ɢ
voiced uvular
?
ʔ
glottal stop
Consonant Fricatives
SPPAS
IPA
Description
Examples
f
f
voiceless labiodental
h
h
voiceless glottal
s
s
voiceless alveolar
S
ʃ
voiceless postalveolar
v
v
voiced labiodental
x
x
voiceless velar
z
z
voiced alveolar
Z
ʒ
voiced postalveolar
Consonant Nasals
SPPAS
IPA
Description
Examples
m
m
bilabial
n
n
alveolar
Consonant Liquids
SPPAS
IPA
Description
Examples
l
l
alveolar lateral
r
r
alveolar trill
Affricates
SPPAS
IPA
Description
Examples
dZ
d͡ʒ
voiced postalveolar
tS
t͡ʃ
voiceless postalveolar
Semivowels
SPPAS
IPA
Description
Examples
j
j
voiced palatal
Vowels
SPPAS
IPA
Description
Examples
a
a
open front unrounded
A
ɒ
open back rounded
e
e
close-mid front unrounded
i
i
close front unrounded
o
o
close-mid back rounded
u
u
close back rounded
y
y
close front rounded
Fillers
SPPAS
Description
laugh
laughter
noise
noises, unintelligible speech
fp
filled pause (euh)
dummy
un-transcribed speech
Acoustic Model
The acoustic model was created by Brigitte Bigi from the HMM
prototypes extracted from other languages (mainly French and Spanish).
The model was then trained with 3 minutes of manually time-aligned data
and 26 minutes of manually phonetized data.
UBPA at 40ms of the initial model based on prototypes is 89.83% and
UBPA of the final model is 89.96%.
The model was created using a Python script available in the SPPAS package:
acmtrain.py.