I tried running the vtln training on the an4 database and i get this error in
the 12.vtln_align log:
INFO:feat.c(1205):Atdirectory/home/tomy/tutorial/an4/featINFO:feat.c(1022):Readingmfcfile:'/home/tomy/tutorial/an4/feat/an4_clstk/fash/an253-fash-b.0.80.mfc'[0..-1]SYSTEM_ERROR:"pio.c",line504:Failedtostatfile'/home/tomy/tutorial/an4/feat/an4_clstk/fash/an253-fash-b.0.80.mfc';retrying...:NosuchfileordirectoryERROR:"feat.c",line1031:Failedtoopenfile'/home/tomy/tutorial/an4/feat/an4_clstk/fash/an253-fash-b.0.80.mfc'forreading:NosuchfileordirectoryERROR:"main_align.c",line908:Uttan253-fash-b:Inputfileread(an4_clstk/fash/an253-fash-b) with dir (/home/tomy/tutorial/an4/feat)andextension(.0.80.mfc)failedINFO:corpus.c(662):an253-fash-b:0.0secCPU,10.0secClk;TOT:0.0secCPU,20.0secClk
I this the problem ? I'm not sure if this is it or I ran into something else
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Did any oe tried for the vtln training using sphinx train recently
developed.There is some issue while extracting features with warp factor
the error is it is not able to create mfc files with warp facor like 0.80.mfc
0.85.mfc and so on
This is a bug which we need to fix soon. Your help on that is welcome. The
feature extraction should be moved to 000.comp_feat stage.
The solution for the problem is
Log("Phase 5: Extracting features with warp factor $warp\n");
if ($have_feats) {
Log("Phase 5: Skipped, feature files already exist\n");
}
elsif (-x "$ST::CFG_SCRIPT_DIR/make_feats.pl") {
my $logfile = "$logdir/${ST::CFG_EXPTNAME}.extract.$warp.log";
my $rv = RunTool('../scripts/make_feats.pl',
$logfile, 0,
-c => $ST::CFG_LISTOFFILES,
-eo => "$warp.$ST::CFG_FEATFILE_EXTENSION",
-ei => $ST::CFG_WAVFILE_EXTENSION,
-di => $ST::CFG_WAVFILES_DIR,
-do => $ST::CFG_FEATFILES_DIR,
-mswav => $ST::CFG_VTLN,
-cfg => $ST::CFG_FILE,
-warp_params => $warp);
if ($rv != 0) {
Log("Failed in warp $warp");
exit 1;
}
}
else {
Log("Failed to find a script for feature extraction");
exit 1;
}
just remove the whole code between
Extract features (unless they're already there)
Wait for them all to finish, and collect their likelihoods
and paste the above code in between those
the vtln problem is solved
I tried running the vtln training on the an4 database and i get this error in
the 12.vtln_align log:
I this the problem ? I'm not sure if this is it or I ran into something else
yes