Run Acoustic Model Creation Script
Make sure you have the following HTK scripts to your 'voxforge/HTK_scripts'
folder:
maketrihed;
mkclscript.prl; prompts2mlf; prompts2wlist.
Create a new folder in your voxforge and call it 'scripts'.
Extract the following tar file to your 'voxforge/auto/scripts folder:
scripts.tgz
Run the
HTK_Compile_Model_sh
script:
$ cd voxforge/auto/scripts
$./HTK_Compile_Model.sh
This will create the following files (located in the auto/acoustic_model_files folder) that make up your acoustic model:
Note: the hmmdefs file shown here was trained with my voice. It will not work with your
voice. Your file will have a the same structure as this one, but the
statistics will be different.
Comments
Click the 'Add' link to add a comment to this page; click the 'Read More' link to view replies to a posted comment.
Add
•
Search
Error running HTK_Compile_Model.sh
By 9yards
-
7/21/2009
- 1 Replies
I am running HTK_Compile_Model.sh and getting the following error message.
Step 4 - Creating Transcription Files ============================================================== writing to mlf file ./interim_files/words.mlf writing to ./interim_files/words.mlf file done Step 5 - Coding the (Audio) Data ============================================================== ERROR [+6210] OpenWaveInput: Cannot open waveform file ../train/wav/sample1.w av FATAL ERROR - Terminating program C:\cygwin\HTK\htk-3.3-windows-binary\htk\HCop y.exe
I have checked the directories and the code, but can't figure out where I am going wrong. Can anyone give me some pointers?
Much appreciated
error after HTK_Compile_Mode1.sh
By rahulgupta.87
-
5/4/2009
- 11 Replies
sample.grammar is :
S : NS_B SENT NS_E
SENT: COM VAR
sample.voca
% NS_B
<s> sil
% NS_E
</s> sil
% COM
INPUT in pu t
OUTPUT au t pu t
% VAR
A a
B b
RUNNING THE SCRIPTS GIVE :
$ ./HTK_Compile_Model.sh
init
==============================================================
Step 1 - Task Grammar
==============================================================
already completed manually
Step 2 - Pronunciation Dictionnary
==============================================================
sorting:./interim_files/wlist to:./interim_files/wlist1
Found voxforge_lexicon
***Please review the following HDMan output***:
WARNING: no script file ./input_files/voxforge_lexicon.ded
Missing Words
-------------
*/sample1
5.41.15.1507;}viewkind4uc1pardf0fs20
A
ABALON
ABDOMINALS
ABOLISHpar
ABOUNDING
ABOUT
ACCOUNT
ACHIEVE
ACTUAL
ACUPUNCTURE
ADVENTUREpar
AEROBIC
AFGHAN
AGAINST
AGATHApar
AHEAD
AIRPLANE
ALGORITHM
ALLENTOWNpar
ALRIGHT
ALTHOUGH
ALTOGETHER
AMNESIA
AND
ANOTHERpar
ARTHRITIS
ASIA
AVERSION
Apar
B
BATTLE
BEATLE
BEIGE
BEIJINGpar
BELOW
BELT
BEND
BITTEN
BLATANT
BODY
BOOKENDS
BRIGHT
BRIGHTEN
BRITAINpar
BROOKHAVEN
BROUHAHA
BULLHEADSpar
BUSBOYS
Bpar
CALL
CHAMPLAIN
CHOICE
COILS
COINpar
COLLECTION
COLORATION
COMBINATION
COMMERCIALpar
COMPLAIN
COPYRIGHT
COUPLE
CRADLE
CRITERIA
CRUMBLEpar
CUBA
CUBE
CUMULATIVEpar
CURING
CURLING
CYCLINGpar
CYNTHIA
Courier
DANFORTH
DASHBOARD
DATABASEpar
DEPTHpar
DIGEST
DIGITAL
DILIGENTpar
ELAINE
EXPLAINpar
GATEWAY
GATORADE
GAZEBO
HELLO
HELMET
HELP
HELPLESS
HELPpar
HOOD
INPUT
INPUTpar
KENNEL
KENNETH
KENYA
LEISURE
LITTLE
MEASURE
MEASUREpar
METALpar
MIDDLE
Msftedit
NEEDLE
New;}}
OUTPUT
OUTPUTpar
PHONE
POODLE
RIGHTpar
SADDLEpar
SENT-END
SENT-START
STEVE
WEEKENDpar
YOUNG
YOUNGpar
par
Dictionary Usage Statistics
---------------------------
Dictionary TotalWords WordsUsed TotalProns PronsUsed
voxforge_lex 0 0 0 0
dict 0 0 0 0
119 words required, 119 missing
Dictionary ./interim_files/dict created
Step 3 - Recording the Data
==============================================================
already completed manually
Step 4 - Creating Transcription Files
==============================================================
writing to mlf file ./interim_files/words.mlf
writing to ./interim_files/words.mlf file done
ERROR [+1232] NumParts: Cannot find word Courier in dictionary
FATAL ERROR - Terminating program C:\my\cygwin\HTK\htk-3.3-windows-binary\htk\H
LEd.exe
ERROR [+1232] NumParts: Cannot find word Courier in dictionary
FATAL ERROR - Terminating program C:\my\cygwin\HTK\htk-3.3-windows-binary\htk\H
LEd.exe
HellRazorr@HellRazorrr ~/new/voxforge/auto/scripts
$
HowTo Error - directory acoustic_model_files not being created
By kmaclean
-
1/26/2009
See this entry in trac: Ticket #449 - HowTo error
Just thought I would share a minor issue with you regarding the howto -
in one of my projects I neglected to establish a directory called
acoustic_model_files to receive hmmdefs and tiedlist. Interestingly the
scripts created a single file called acoustic_model_files instead of
the directory, and the error is not found until you run julius to test
and he complains he cannot find hmmdefs etc. It looks as though the
directory is there but in fact it is not. When I deleted the bogus
file, created the directory and ran the setup again all was well.
Execution stops at step 5
By colbec
-
3/5/2008
- 1 Replies
The first time I ran HTK_Compile_Model.sh it stopped at step 5 and complained that it could not find sample1.wav. The message indicated it was looking in a directory one level up, but the Howto calls for the wav files to be stored two levels up in voxforge/train/wav directory.
I found that after:
mkdir auto/train and mkdir auto/train/wav,
copy all sample*.wav files to this new wav directory
mkdir auto/train/mfcc
then the script ran perfectly to the end. Maybe I missed something, but hope it helps.
STEEP 5 MISSING
By kat_wills
-
12/28/2007
- 1 Replies
Hi,
Step 4 made a reference on step 5,but it does not appear as if it exist.So when I ran the 'script' step, /HTK_Compile_Model.sh,it gave error.The step made reference to 'auto/acoustic_model_files folder' which did not exists.It appears some steps are missing in the linux tutorial.
Any help will be appreciated.
Thanks