Changes between Version 18 and Version 19 of NewLanguageSupport


Ignore:
Timestamp:
12/14/09 16:53:04 (15 years ago)
Author:
marcela_charfuelan
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • NewLanguageSupport

    v18 v19  
    287287 
    288288#Usage: java DatabaseSelector -locale language -mysqlHost host -mysqlUser user -mysqlPasswd passwd -mysqlDB wikiDB  
    289 #        -tableName selectedSentencesTableName -featDef file -stop stopCriterion  
    290 #        [-coverageConfig file -initFile file -selectedSentences file -unwantedSentences file ] 
     289#        -tableName selectedSentencesTableName  
     290#        [-stop stopCriterion] 
     291#        [-featDef file -coverageConfig file] 
     292#        [-initFile file -selectedSentences file -unwantedSentences file ] 
    291293#        [-tableDescription a brief description of the table ] 
    292294#        [-vectorsOnDisk -overallLog file -selectionDir dir -logCoverageDevelopment -verbose] 
     
    296298#    generating several selection sets. FINAL name will be: "locale_name_selectedSenteces".  
    297299#    where name is the name provided for the selected sentences table. 
    298 #-tableDescription : short description of the selected sentences table. (default: empty) 
     300#-tableDescription : short description of the selected sentences table.  
     301#    Default: empty 
    299302#-featDef file : The feature definition for the features 
     303#    Default: [locale]_featureDefinition.txt for example for US English: en_US_featureDefinition.txt 
     304#            this file is automatically created in previous steps by the FeatureMaker. 
    300305#-stop stopCriterion : which stop criterion to use. There are five stop criteria.  
    301 # They can be used individually or can be combined: 
    302 #  - numSentences n : selection stops after n sentences 
    303 #  - simpleDiphones : selection stops when simple diphone coverage has reached maximum 
    304 #  - simpleProsody : selection stops when simple prosody coverage has reached maximum 
     306#    They can be used individually or can be combined: 
     307#    - numSentences n : selection stops after n sentences 
     308#    - simpleDiphones : selection stops when simple diphone coverage has reached maximum 
     309#    - simpleProsody : selection stops when simple prosody coverage has reached maximum 
     310#    Default: "numSentences 90 simpleDiphones simpleProsody" 
    305311#-coverageConfig file : The config file for the coverage definition.  
    306 #   Default config file is ./covDef.config. 
     312#    Default: there is a default coverage config file in MARY_BASE/java/marytts/tools/dbselection/covDef.config 
     313#             this file will be copied to the current directory if no file is provided. 
     314#-initFile file : The file containing the coverage data needed to initialise the algorithm. 
     315#    Default: /current-dir/init.bin 
     316#-overallLog file : Log file for all runs of the program: date, settings and results of the current 
     317#    run are appended to the end of the file. This file is needed if you want to analyse your results  
     318#    with the ResultAnalyser later. 
     319#-selectionDir dir : the directory where all selection data is stored. 
     320#    Default: /current-dir/selection 
    307321#-vectorsOnDisk: if this option is given, the feature vectors are not loaded into memory during  
    308 # the run of the program. This notably slows down the run of the program! 
    309 #-initFile file : The file containing the coverage data needed to initialise the algorithm. 
    310 #   Default init file is ./init.bin 
    311 #-overallLog file : Log file for all runs of the program: date, settings and results of the current 
    312 # run are appended to the end of the file. This file is needed if you want to analyse your results  
    313 # with the ResultAnalyser later. 
    314 #-selectionDir dir : the directory where all selection data is stored. 
    315 #   Standard directory is ./selection 
    316 #-logCoverageDevelopment : If this option is given, the coverage development over time  
    317 # is stored. 
    318 #-verbose : If this option is given, there will be more output on the command line 
    319 # during the run of the program. 
    320  
     322#    the run of the program. This notably slows down the run of the program! 
     323#    Default: no vectorsOnDisk 
     324#-logCoverageDevelopment : If this option is given, the coverage development over time is stored. 
     325#    Default: no logCoverageDevelopment 
     326#-verbose : If this option is given, there will be more output on the command line during the run of the program. 
     327#    Default: no verbose 
    321328 
    322329export MARY_BASE="[PATH TO MARY BASE]" 
     
    332339-tableName "test" \ 
    333340-tableDescription "Testing table: English wikipedia short set. " \ 
    334 -featDef "/current-dir/en_US_featureDefinition.txt" \ 
    335341-stop "numSentences 90 simpleDiphones simpleProsody" \ 
    336 -coverageConfig "/current-dir/covDef.config" \ 
    337 -initFile "/current-dir/init.bin" \ 
    338 -overallLog "/current-dir/overallLog.txt" \ 
    339 -selectionDir "/current-dir/selection" \ 
    340342-logCoverageDevelopment \ 
    341343-vectorsOnDisk 
    342344 
    343345}}} 
    344 The following is an example of covDef.config file:[[BR]] 
    345  
    346 {{{ 
    347 # 
    348 # Template settings file for selection algorithm 
    349 # Change the settings according to your needs  
    350 # A comment starts with # 
    351 # 
    352 #simpleDiphones true means units are phone+nextPhone+prosody 
    353 #(This is the only one supported for the moment) 
    354 simpleDiphones true  
    355 # 
    356 #possible frequency weights: normal, 1minus, inverse and none 
    357 frequency inverse  
    358 # 
    359 #sentenceLength none ignores sentence length 
    360 #sentenceLength <maxValue> <minValue> restricts sentence length 
    361 sentenceLength 150 30 
    362 # 
    363 #the wanted weights for features phone, nextPhone/nextPhoneClass and prosody 
    364 wantedWeight 25 5 1  
    365 # 
    366 #the number by which the wanted weight is divided each time a unit with the 
    367 #appropriate value is added to the cover 
    368 wantedWeightDecrease 1000  
    369 # 
    370 #the phones that are known to be missing in the database and should be ignored 
    371 #missingPhones  
    372 }}} 
     346 
    373347'''Output:'''[[BR]] - Several log information in "/current-dir/selection/" directory 
    374348