Simon having trouble with Julius

Peter Grasch peter at grasch.net
Thu Jun 19 10:45:23 UTC 2014


On Thursday, June 19, 2014 02:00:31 PM Shantanu Tushar Jha wrote:
> > > Did one more training and it did do "compiling model" and stuff, but the
> > > same error still persists, and the tiedlist file is still absent.
> > 
> > Great! That means it's reproducible. Now I need to find out what could go
> > wrong
> > with a QFile::copy()...
> > 
> > Anyway, I have a few tight deadlines until next week, but I'll try to have
> > a
> > look at it after the worst is over. May be next Friday, though.
> 
> No problemo. Who knows, maybe if I find time I'll try compiling from source
> and debugging a bit ;)
If you do, this is what you're looking for: 
simonlib/speechmodelcompilation/modelcompilerhtk.cpp:1593:
ModelCompilerHTK::adaptBaseModel()

Specifically, line 1623 is a good start:
  if ( (QFile::exists(hmmDefsPath) && (!QFile::remove(hmmDefsPath))) ||
       (!QFile::copy(tempDir+"/xforms/"+expectedHmmPath+".mfc.mllr2", 
hmmDefsPath)) ||
       (QFile::exists(tiedListPath) && (!QFile::remove(tiedListPath))) ||
       (!QFile::copy(baseTiedlistPath, tiedListPath)) ) {
    analyseError(i18n("Could not switch to new model."));
    return false;
  }

Then you'd track it back trhough the modelcompilationmanagerhtk.cpp (that 
initiates the compilation at line 161) and so on.

If you don't have the time, I'll take a look next week.

Best regards,
Peter


More information about the Kde-speech mailing list