RESOURCE MANAGEMENT TEST: sphinx - auxillary test

ORIGINAL TEST DATE: prior to May 1988

RECOGNITION SYSTEMS: speaker-independent continuous speech transcription 

SPEAKER-INDEPENDENT TRAINING MATERIAL:
     . (((all 80 speakers from Speaker-Independent Training set) X
         (40 sentence utterances/speaker)) +
        ((25 speakers from Speaker-Independent Development-Test set) X
         (40 sentences utterances/speaker)))        
       = 4200 training utterances from CD2-3.1: /rm1/ind_trn, 
         CD2-4.2: /rm1/ind/dev_aug and CD2-4.2: /rm1/ind/dev_exc
     . files are listed in CD2-4.2: /rm1/doc/train/80_indtr.ndx and
       CD2-4.2: /rm1/doc/train/25_indtr.ndx

TEST PARAMETERS:
     . grammar: 
          (1) no grammar - full branching
          (2) "standard" non-probabilistic Word-Pair grammar developed at BBN
          (3) "bigram" grammar with estimated "tag transition" probabilities

SPEAKER-INDEPENDENT TEST MATERIAL:
     . (((10 speakers from Speaker-Independent Development-Test set) X
         (10 speakers X (10 sentence utterances/speaker)) +
        ((5 speakers from Speaker-Independent Development-Test set) X
         (10 sentence utterances/speaker)))
       = 150 test utterances from CD2-4.2: /rm1/ind/dev_aug
     . files are listed in CD2-4.2: /rm1/doc/test/sphinx88/x_indtst.ndx

SCORING SOFTWARE SETTINGS:
     The following commands should be used when scoring this test:

     . in the no-grammar case:
          ngscore hypothesis-file [system-code] [system-description]

     . in the Word-Pair and bigram grammar case:
          wgscore hypothesis-file [system-code] [system-description]