VoxForge
Hi I am struck here
HERest -A -D -T 1 -C config -I phones0.mlf -t 250.0 150.0 1000.0 -S train.scp -H hmm0/macros -H hmm0/hmmdefs -M hmm1 monophones0
HTK Configuration Parameters[10]
Module/Tool Parameter Value
# NUMCEPS 12
# CEPLIFTER 22
# NUMCHANS 26
# PREEMCOEF 0.970000
# USEHAMMING TRUE
# WINDOWSIZE 250000.000000
# SAVEWITHCRC TRUE
# SAVECOMPRESSED TRUE
# TARGETRATE 100000.000000
# TARGETKIND MFCC_0_D_N_Z
HERest ML Updating: Transitions Means Variances
System is PLAIN
40 Logical/40 Physical Models Loaded, VecSize=25
2 MMF input files
Pruning-On[250.0 150.0 1000.0]
Processing Data: sample1.mfc; Label sample1.lab
ERROR [+6510] LOpen: Unable to open label file ../train/mfcc/sample1.lab
FATAL ERROR - Terminating program HERest
I tried to google and could not find an answer. Earlier was struck with sp file not found issue. I removed sp entry from monophones0
Any help?
>Unable to open label file ../train/mfcc/sample1.lab
this looks like a path issue...make sure the label file ids in your phones0.mlf start with '*; (e.g. "*/sample1.lab") and that your train.scp file points to the actual location of your feature files (e.g. ../train/mfcc/sample1.mfc)
Some more details based on your input.
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$ cat phones0.mlf
#!MLF!#
"*/words.lab"
sil
....
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$ cat train.scp
../train/mfcc/sample1.mfc
../train/mfcc/sample2.mfc
....
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$ ls ../train/mfcc/
sample10.mfc sample13.mfc sample16.mfc sample19.mfc sample21.mfc sample24.mfc sample27.mfc sample2.mfc sample32.mfc sample35.mfc sample38.mfc sample40.mfc sample6.mfc sample9.mfc
sample11.mfc sample14.mfc sample17.mfc sample1.mfc sample22.mfc sample25.mfc sample28.mfc sample30.mfc sample33.mfc sample36.mfc sample39.mfc sample4.mfc sample7.mfc
sample12.mfc sample15.mfc sample18.mfc sample20.mfc sample23.mfc sample26.mfc sample29.mfc sample31.mfc sample34.mfc sample37.mfc sample3.mfc sample5.mfc sample8.mfc
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$
I am just following step 6 menioned here
http://www.voxforge.org/home/dev/acousticmodels/linux/create/htkjulius/tutorial/monophones/step-6
>phones0.mlf
>#!MLF!#
>"*/words.lab"
>sil
the label entries in your phones0.mlf file should contain the same root name as in the entries in your train.scp file....
For Example, your phones0.mlf file should contain a label name whose root (i.e. "sample1" in this case):
#!MLF!# "*/sample1.lab" sil d ay ah l w
...
which must correspond to the feature file with the same root name in your train.scp file:
../train/mfcc/sample1.mfc ../train/mfcc/sample2.mfc
...
Looks like I missed some steps. I will restart from step1. Because got into this issue
HERest -A -D -T 1 -C config -I phones0.mlf -t 250.0 150.0 1000.0 -S train.scp -H hmm0/macros -H hmm0/hmmdefs -M hmm1 monophones0
HERest -A -D -T 1 -C config -I phones0.mlf -t 250.0 150.0 1000.0 -S train.scp -H hmm0/macros -H hmm0/hmmdefs -M hmm1 monophones0
HTK Configuration Parameters[10]
Module/Tool Parameter Value
# NUMCEPS 12
# CEPLIFTER 22
# NUMCHANS 26
# PREEMCOEF 0.970000
# USEHAMMING TRUE
# WINDOWSIZE 250000.000000
# SAVEWITHCRC TRUE
# SAVECOMPRESSED TRUE
# TARGETRATE 100000.000000
# TARGETKIND MFCC_0_D_N_Z
HERest ML Updating: Transitions Means Variances
System is PLAIN
40 Logical/40 Physical Models Loaded, VecSize=25
2 MMF input files
Pruning-On[250.0 150.0 1000.0]
Processing Data: sample1.mfc; Label sample1.lab
Unable to traverse 2907 states in 1410 frames
WARNING [-7324] StepBack: File ../train/mfcc/sample1.mfc - bad data or over pruning
in HERest
Processing Data: sample2.mfc; Label sample2.lab
ERROR [+6510] LOpen: Unable to open label file ../train/mfcc/sample2.lab
FATAL ERROR - Terminating program HERest
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$
Now it gives error for sample2, but have only one phones0.mlf and where to define it for sample2?
Got this error in step8
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$ HVite -A -D -T 1 -l '*' -o SWT -b SENT-END -C config -H hmm7/macros -H hmm7/hmmdefs -i aligned.mlf -m -t 250.0 150.0 1000.0 -y lab -a -I words.mlf -S train.scp dict monophones1
HVite -A -D -T 1 -l * -o SWT -b SENT-END -C config -H hmm7/macros -H hmm7/hmmdefs -i aligned.mlf -m -t 250.0 150.0 1000.0 -y lab -a -I words.mlf -S train.scp dict monophones1
HTK Configuration Parameters[10]
Module/Tool Parameter Value
# NUMCEPS 12
# CEPLIFTER 22
# NUMCHANS 26
# PREEMCOEF 0.970000
# USEHAMMING TRUE
# WINDOWSIZE 250000.000000
# SAVEWITHCRC TRUE
# SAVECOMPRESSED TRUE
# TARGETRATE 100000.000000
# TARGETKIND MFCC_0_D_N_Z
ERROR [+6551] LoadMasterFile: MLF file header is missing
FATAL ERROR - Terminating program HVite
sekar@sekar-J1900NH2:~/speechrecognition/voxforge/tutorial$ more words.mlf
ABALON
ABDOMINALS
ABOLISH
>ERROR [+6551] LoadMasterFile: MLF file header is missing
assuming that you are at "Step 8 - Realigning the Training Data", it seems like there's something wrong in your words.mlf file