like the topic say: how to create means, variances and mixture_weights?
TIA
Jens
root@debian:/opt/SpinxBase/bin/jkallup/model-de# mk_s2sendump -moddeffn mdef -mixwfn mixture_weights -sendumpfn sendump
Current configuration: [NAME][DEFLT][VALUE]
-example no no
-help no no
-mixwfn mixture_weights
-moddeffn mdef
-mwfloor 0.00001 1.000000e-05
-pocketsphinx no no
-sendumpfn sendump
INFO: model_def_io.c(573): Model definition info:
INFO: model_def_io.c(574): 301 total models defined (7 base, 294 tri)
INFO: model_def_io.c(575): 1204 total states
INFO: model_def_io.c(576): 903 total tied states
INFO: model_def_io.c(577): 21 total tied CI states
INFO: model_def_io.c(578): 7 total tied transition matrices
INFO: model_def_io.c(579): 4 max state/model
INFO: model_def_io.c(580): 4 min state/model
INFO: senone.c(210): Reading senone mixture weights: mixture_weights
WARN: "senone.c", line 318: Weight normalization failed for 1 senones
INFO: senone.c(331): Read mixture weights for 3123 senones: 1 features x 8 codewords
FATAL: "mk_s2sendump.c", line 283: #senones different in mdef(903) and mixw(3123) files
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You need to use mdef file which goes together with mixture_weights, you should have used some different file. Maybe something went wrong in training beforehand, it's better to start wtih a clean folder.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
like the topic say: how to create means, variances and mixture_weights?
TIA
Jens
root@debian:/opt/SpinxBase/bin/jkallup/model-de# mk_s2sendump -moddeffn mdef -mixwfn mixture_weights -sendumpfn sendump
Current configuration:
[NAME] [DEFLT] [VALUE]
-example no no
-help no no
-mixwfn mixture_weights
-moddeffn mdef
-mwfloor 0.00001 1.000000e-05
-pocketsphinx no no
-sendumpfn sendump
INFO: model_def_io.c(573): Model definition info:
INFO: model_def_io.c(574): 301 total models defined (7 base, 294 tri)
INFO: model_def_io.c(575): 1204 total states
INFO: model_def_io.c(576): 903 total tied states
INFO: model_def_io.c(577): 21 total tied CI states
INFO: model_def_io.c(578): 7 total tied transition matrices
INFO: model_def_io.c(579): 4 max state/model
INFO: model_def_io.c(580): 4 min state/model
INFO: senone.c(210): Reading senone mixture weights: mixture_weights
WARN: "senone.c", line 318: Weight normalization failed for 1 senones
INFO: senone.c(331): Read mixture weights for 3123 senones: 1 features x 8 codewords
FATAL: "mk_s2sendump.c", line 283: #senones different in mdef(903) and mixw(3123) files
You need to use mdef file which goes together with mixture_weights, you should have used some different file. Maybe something went wrong in training beforehand, it's better to start wtih a clean folder.