German

Flat
Newbie braucht einen Anstoss :-)
User: swa00
Date: 10/27/2016 10:44 am
Views: 6482
Rating: 0

Hallo liebe Gemeinde,

ich möchte euch an dieser Stelle bitten mir ein wenig

Hilfestellung zu geben.

Ich entwickle derzeit unter Android und habe erfolgreich  das

DemoProject zum Laufen gebracht ( eng) (pocketsphinx)

Nun komme  ich  mit meinem Verständniss ein wenig an meine Grenzen , in dem ich versuche  die deutsche  Sprache einzubinden.

Ich  habe  gesehen , das  "günther"  auch  was  schönes Neues hochgeladen hat (Danke  an dich) . Allerdings  , kann ich  mir dem Inhalt der Daten  nicht  so  wirklich etwas  anfangen , so  dass ich sie  ins  gleiche  Format, wie die  derzei englischen  bekomme

Im  englischen habe  ich  die mdef,means,noisedict, variances  etc...   und  dazu  noch  eine  dict.

Günthers  Beispiel  habe  ich  dann mal  auch so  genommen, (cmusphinx-voxforge-der20160922)

allerdings  scheint das  Ganze  zu  gross  zu sein und  die  App  verweigert   ihren  Dienst  weil  zu  viel  Speicher benötigt wird.

Eigentlich  möchte  ich  NUR ein KeyPhrase ausführen, weiter  ist  nichts  geplant.

Wer ist  also  so  lieb  und  könnte  mir  auf  die  Sprünge  helfen  , was  ich  falsch  mache  oder wo  ich etwas  laden kann , was ggf  lauffähig  ist .

P.S.  mir steht  leider  kein Linux  rechner  derzeit  zur  verfügung, (nur  win7) 

 

Lieben Dank im Voraus für  euere Mühe ..

Stefan

 

 

 

 

 

 

 

 

 

 

 

Re: Newbie braucht einen Anstoss :-)
User: nsh
Date: 10/29/2016 4:48 am
Views: 35
Rating: 0

You only need etc/voxforge.dic and files in model_parameters/voxforge.cd_cont_3000. Total size is 16Mb. You can also convert mdef to binary format with pocketsphinx_mdef_convert to save space.

Unfortunately Günther trained continuous model for accurate recognition, it is too big for mobile. You need to train PTM model yourself or ask Günther to do that. It's enough to train a model from the downloadable voxforge data, accuracy for few words shoudl be enough. You can add german-speechdata-v2 too.

Re: Newbie braucht einen Anstoss :-)
User: swa00
Date: 10/29/2016 7:39 am
Views: 5
Rating: 0

Hi nsh,

thank you very much for  your answer. i will try to follow your instructions , otherwise  i will contact Guenther.

Have a  nice weekend

 

 

Re: Newbie braucht einen Anstoss :-)
User: guenter
Date: 10/30/2016 4:52 am
Views: 4
Rating: 0

Hi there!

I am currently preparing the next release of my german model which will include a much larger portion of german-speechdata-v2 besides the voxforge material (all in all >200h of speech!).

I had never heard of PTM models before, I can definitely try to train one of those. Is there a sphinx_train.cfg file available somewhere that I could use as a starting point?

 

Thanks,

   Guenter

 

Re: Newbie braucht einen Anstoss :-)
User: nsh
Date: 10/30/2016 5:17 am
Views: 88
Rating: 0

Hello Guenter

To learn about model types you can check

http://cmusphinx.sourceforge.net/wiki/acousticmodeltypes

to train PTM model you could try to set the following in config:

$CFG_HMM_TYPE  = '.ptm.'; # PocketSphinx (larger data sets)

} elsif ($CFG_HMM_TYPE eq '.ptm.') {

  $CFG_DIRLABEL = 'ptm';

  $CFG_FEATURE = "1s_c_d_dd";

  $CFG_NUM_STREAMS = 3;

  $CFG_SVSPEC = "0-12/13-25/26-38";

  $CFG_INITIAL_NUM_DENSITIES = 128;

  $CFG_FINAL_NUM_DENSITIES = 128;

  die "For phonetically tied models, the initial and final models have the same density"

    if ($CFG_INITIAL_NUM_DENSITIES != $CFG_FINAL_NUM_DENSITIES);

$CFG_N_TIED_STATES = 5000; 

The resulting model will be less accurate but more suitable for mobile. 

US English models is distributed in both PTM and continuous version for example.

 

Re: Newbie braucht einen Anstoss :-)
User: guenter
Date: 11/7/2016 12:33 pm
Views: 25
Rating: 0

nsh, thanks for the quick response! I have built a PTM model using these settings which seems to have worked without any issues, very nice work :)

stats for these models (trained on 200 hours):

cmusphinx cont model: SENTENCE ERROR: 25.4% (1523/6006)   WORD ERROR RATE: 5.7% (4147/73022)
cmusphinx ptm model: SENTENCE ERROR: 24.7% (1481/6006)   WORD ERROR RATE: 7.1% (5199/73022)
I have uploaded both models along with the language model in arpa format to:
http://goofy.zamia.org/voxforge/de/ 
look for files that carry the release tag r20161107 in their name.
I will announce the availability of the new 200h models in a separate forum post soon (release will include kaldi models as well), so please consider this a preview release for now. 
Feedback - as always - is welcome.
Thanks again for your support and suggestions! :)
   Guenter
Re: Newbie braucht einen Anstoss :-)
User: swa00
Date: 11/7/2016 3:17 pm
Views: 2841
Rating: 0

Dear Guenter,

THANK YOU VERY MUCH for your PTM Model.

i have  tested it under PocketSphinx / Android and it works great.

and getting  results  nearly  the english PTM Model.

 

 

 

 

 

PreviousNext