VoxForge
VoxForge
Does anyone have a standard 40 phone transcription for TIMIT database? The original came with a 60 phone transcription which is not much of an interest.
--- (Edited on 3/13/2015 8:22 am [GMT-0500] by Terminator) ---
#h sil
aa ao
ae ae
ah ah
ao ao
aw aw
ax ah
ax-h ah
axr er
ay ay
b b
bcl sil
ch ch
d d
dcl sil
dh dh
dx dx
eh eh
el el
em m
en en
eng ng
epi sil
er er
ey ey
f f
g g
gcl sil
h# sil
hh hh
hv hh
ih ih
ix ih
iy iy
jh jh
k k
kcl sil
l el
m m
n en
ng ng
nx en
ow ow
oy oy
p p
pau sil
pcl sil
q t
qcl sil
r r
s s
sh sh
t t
tcl sil
th th
uh uh
uw uw
ux uw
v v
w w
y y
z z
zh sh
--- (Edited on 3/13/2015 9:07 am [GMT-0500] by Visitor) ---
Hi Visitor,
I have this. I need to change the phone label files to new phones, the *.phn files. There is 'tcl and t in sequence for a transcription, If i replace them there will be 't t', but pronunciation according to 40 phone does not warranty two t's, but only one 't'. My task is to build a monophone model and measure accuuracy.
--- (Edited on 3/13/2015 9:21 am [GMT-0500] by Terminator) ---
maybe i dont understand you or maybe you dont have spend time to look at the map table i posted to you. but i can ensure this maps the 62 timit models to the 39 lee and hon phone models.
this table was provided by Joseph Keshet
--- (Edited on 3/13/2015 9:37 am [GMT-0500] by cigp) ---
ok, got it. tcl -> sil seems ok. But how can I use this map to build acoustic models. My acoustic model needs to be only 40 phones and all my label files (*.phn) contain 60 phones. Is there any tool which takes each *.phn in 60 format and gives out *.phn in 40 format? My request.
--- (Edited on 3/13/2015 9:49 am [GMT-0500] by Terminator) ---
ok, got it. tcl -> sil seems ok. But how can I use this map to build acoustic models. My acoustic model needs to be only 40 phones and all my label files (*.phn) contain 60 phones. Is there any tool which takes each *.phn in 60 format and gives out *.phn in 40 format? My request.
--- (Edited on 3/13/2015 9:51 am [GMT-0500] by Terminator) ---