Click here to register.

Julian Startup Output

$julian -input mic -C julian.jconf
include config: julian.jconf
###### check configurations
###### initialize input device
fragment size = 1024 bytes (10 msec)
AD-in thread created
###### build up system
Reading in HMM definition...(ascii)...
limit check passed
   defined HMMs:    50
  logical names:   506 in HMMList
    base phones:    44 used in logical
done
Making pseudo bi/mono-phone for IW-triphone...369 added as logical...done
reading [sample.dfa] and [sample.dict]...
Reading in dictionary...
18 words...done
Reading in DFA grammar...done
- Gram #0: read
[grammars]
  # 0: [active     ]   18 words,   6 categories,    6 nodes (new) "sample"
gram "sample" registered
- Grammar update check
Mapping dict item <-> DFA terminal (category)...done
- Gram #0: installed
Building HMM lexicon tree.................153+0=153 nodes
  coordination check passed
done
now beam width = 63 (guess)
- update completed
[grammars]
  # 0: [active     ]   18 words,   6 categories,    6 nodes "sample"
  Grobal:              18 words,   6 categories,    5 nodes
Generating addlog table...1953 kb...done
All init successfully done

------------- System Info begin -------------
Julian rev.3.5.2-multipath (standard)

Engine configuration:
 - Base setup : standard
 - Tunings    : DFA, StrictIWCD2, IconvOutput
 - Compiled by: gcc -O6 -fomit-frame-pointer

Continuous Speech Recognition Parser based on automaton grammar

Files:
    hmmfilename=hmm15/hmmdefs
    hmmmapfilename=tiedlist
    grammar #1:
        dfa  = sample.dfa
        dict = sample.dict

Acoustic analysis condition:
           sample period =  208 ns      (48000 Hz)
             window size =  400 samples (8.3 ms)
             frame shift =  160 samples (3.3 ms)
            delta window =    2 frames  (6.7 ms) around
                hi freq. = OFF
                lo freq. = OFF
          energy measure = (_0) 0'th cepstral
    spectral subtraction = off

HMM Info:
    50 models, 135 states, 135 mixtures are defined
              model type = context dependency handling ON
      training parameter = MFCC_N_D_Z_0
           vector length = 25
        cov. matrix type = DIAGC
           duration type = NULLD
             mixture num = 1
           max state num = 5

Dictionary Info:
        vocabulary size  = 18 words, 51 models
        average word len = 2.8 models, 8.5 states
       maximum state num = 15 nodes per word
       transparent words = not exist
       words under class = not exist

Lexicon tree info:
         total node num =    153
          root node num =     18
          leaf node num =     18

DFA grammar info:
      5 nodes, 6 arcs, 6 terminal(category) symbols
      category-pair matrix size is 4 bytes

Weights and words:
        (-penalty1) IW penalty1 = +0.0
        (-penalty2) IW penalty2 = +0.0
        (-cmalpha)CM alpha coef = 0.050000
        (-sp)shortpause HMM name= "sp" specified, "sp" applied (physical)
          found sp category IDs =

Search parameters:
              1st pass decoding = on-the-fly with last CMN
                1st pass method = 1-best approx. generating indexed trellis
        (-b) trellis beam width = 63 (-1 or not specified - guessed)
        (-n)search candidate num= 10
        (-s)  search stack size = 500
        (-m)    search overflow = after 2000 hypothesis poped
                2nd pass method = searching sentence, generating N-best
        (-b2)  pass2 beam width = 100
        (-lookuprange)lookup range= 5  (tm-5 <= t <tm+5)
        (-sb)2nd scan beamthres = 80.0 (in logscore)
        (-gprune)Gauss. pruning = none (full computation)
        (-n)        search till = 10 candidates found
        (-output)    and output = 1 candidates out of above
         IWCD handling:
           1st pass: approximation (use average prob. of same LC)
           2nd pass: strict (apply when expanding hypo. )
         all possible words will be expanded in 2nd pass
         build_wchmm2() used
         lcdset limited by word-pair constraint
         output word confidence measure based on search-time scores

System I/O configuration:
            speech input source = microphone
                 sampling freq. = 48000 Hz
                threaded A/D-in = supported, on
          zero frames stripping = on
                silence cutting = on
                    level thres = 2000 / 32767
                zerocross thres = 60 / sec.
                    head margin = 300 msec.
                    tail margin = 400 msec.
               remove DC offset = off
             reject short input = off
        short pause segmentation= off
               result output to = tty (standard out)
           output charset conv. = disabled

------------- System Info end -------------

        ************************************************************
        * NOTICE: The first input may not be correctly recoginized *
        *         since no CMN parameter is available on startup.  *
        ************************************************************

------
### read waveform input
<<< please speak >>>