lattice-tool.1

这是一款很好用的工具包

              relative to the larger of the two sets.

       --rreedduuccee--bbeeffoorree--pprruunniinngg
              Perform lattice  reduction  before  posterior-based
              pruning.  The default order is to first prune, then
              reduce.

       --pprree--rreedduuccee--iitteerraattee _I
              Perform iterative reduction prior to lattice expan-
              sion, but after pause elimination.

       --ppoosstt--rreedduuccee--iitteerraattee _I
              Perform iterative reduction after lattice expansion
              and pause node recovery.  Note: this is not  recom-
              mended  as it changes the weights assigned from the
              specified language model.

       --nnoo--nnuullllss
              Eliminate NULL nodes from lattices.

       --nnoo--ppaauussee
              Eliminate pause nodes from  lattices  (and  do  not
              recover them after lattice expansion).

       --ccoommppaacctt--ppaauussee
              Use  compact  encoding  of  pause  nodes that saves
              nodes but allows optional pauses where  they  might
              not have been included in the original lattice.

       --lloooopp--ppaauussee
              Add self-loops on pause nodes.

       --iinnsseerrtt--ppaauussee
              Insert optional pauses after every word in the lat-
              tice.  The structure of inserted pauses is affected
              by --ccoommppaacctt--ppaauussee and --lloooopp--ppaauussee.

       --ccoollllaappssee--ssaammee--wwoorrddss
              Perform  an  operation  on  the final lattices that
              collapses all nodes with  the  same  words,  except
              null nodes, pause nodes, or nodes with noise words.
              This can reduce the lattice size dramatically,  but
              also introduces new paths.

       --ccoonnnneeccttiivviittyy
              Check the connectedness of lattices.

       --ccoommppuuttee--nnooddee--eennttrrooppyy
              Compute the node entropy of lattices.

       --ccoommppuuttee--ppoosstteerriioorrss
              Compute  node  posterior  probabilities  (which are
              included in HTK lattice output).

       --ddeennssiittyy
              Compute and output lattice densities.

       --rreeff--lliisstt _f_i_l_e
              Read reference word strings from _f_i_l_e.   Each  line
              starts with a sentence ID (the basename of the lat-
              tice file name), followed by the words.   This  and
              the  next  option  triggers  computation of lattice
              word errors (minimum word error counts of any  path
              through a lattice).

       --rreeff--ffiillee _f_i_l_e
              Read  reference word strings from _f_i_l_e.  Lines must
              contain reference words only, and must  be  matched
              to input lattices in the order processed.

       --wwrriittee--rreeffss _f_i_l_e
              Write the references back to _f_i_l_e (for validation).

       --aadddd--rreeffss _P
              Add the reference words as an  additional  path  to
              the  lattice, with probability _P.  Unless --nnoo--ppaauussee
              is specified, optional pause  nodes  between  words
              are  also  added.  Note that this operation is per-
              formed before lattice reduction and  expansion,  so
              the  new path can be merged with existing ones, and
              the probabilities for the new  path  can  be  reas-
              signed from an LM later.

       --nnooiissee--vvooccaabb _f_i_l_e
              Read  a  list  of ``noise'' words from _f_i_l_e.  These
              words  are  ignored  when  computing  lattice  word
              errors,  when decoding the best word sequence using
              --vviitteerrbbii--ddeeccooddee or --ppoosstteerriioorr--ddeeccooddee, or when  col-
              lapsing nodes with --ccoollllaappssee--ssaammee--wwoorrddss.

       --kkeeeepp--ppaauussee
              Causes  the pause word ``-pau-'' to be treated like
              a regular  word.   It  prevents  pause  from  being
              implicitly added to the list of noise words.

       --iiggnnoorree--vvooccaabb _f_i_l_e
              Read a list of words that are to be ignored in lat-
              tice operations, similar to pause  tokens.   Unlike
              noise  words (see above) they are also skipped dur-
              ing LM evaluation.  With  this  option  and  --kkeeeepp--
              ppaauussee, pause words are not ignored by default.

       --sspplliitt--mmuullttiiwwoorrddss
              Split lattice nodes with multiwords into a sequence
              of non-multiword nodes.  This option  is  necessary
              to  compute  lattice  error  of  multiword lattices
              against non-multiword references, but may be useful
              in its own right.

       --sspplliitt--mmuullttiiwwoorrddss--aafftteerr--llmm
              Perform multiword splitting after lattice expansion
              using the specified LM.  This should be used if the
              LM  uses multiwords, but the final lattices are not
              supposed to contain multiwords.

       --mmuullttiiwwoorrdd--ddiiccttiioonnaarryy _f_i_l_e
              Read a dictionary from  _f_i_l_e  containing  multiword
              pronunciations  and  word boundary markers (a ``|''
              phone label).  Specifying such a dictionary  allows
              the  multiword  splitting options to infer accurate
              time marks and pronunciation  information  for  the
              multiword components.

       --mmuullttii--cchhaarr _C
              Designate  _C  as  the character used for separating
              multiword components.  The default is an underscore
              ``_''.

       --ooppeerraattiioonn _O
              Perform  a  lattice algebra operation _O on the lat-
              tice or lattices processed, with the second operand
              specified  by  --iinn--llaattttiiccee22.   Operations currently
              supported are ccoonnccaatteennaattee and oorr,  for  serial  and
              parallel lattice combination, respectively, and are
              applied after all other lattices manipulations.

       --vviitteerrbbii--ddeeccooddee
              Print out the word sequence  corresponding  to  the
              highest probability path.

       --ppoosstteerriioorr--ddeeccooddee
              Print  out  the  word sequence with lowest expected
              word error.

       --oouuttppuutt--ccttmm
              Output word sequences  in  NIST  CTM  (conversation
              time mark) format.  Note that word start times will
              be relative to the lattice start  time,  the  first
              column will contain the lattice name, and the chan-
              nel field is always 1.  The word  confidence  field
              contains  posterior  probabilities  if  --ppoosstteerriioorr--
              ddeeccooddee is in  effect.   This  option  also  implies
              --aaccoouussttiicc--mmeesshh.

       --hhiiddddeenn--vvooccaabb file
              Read  a  subvocabulary from _f_i_l_e and constrain word
              meshes to only align those words  that  are  either
              all  in  or outside the subvocabulary.  This may be
              used to keep ``hidden event''  tags  from  aligning
              with regular words.

       --ddiiccttiioonnaarryy--aalliiggnn
              Use  the  dictionary  pronunciations specified with
              --ddiiccttiioonnaarryy to induce a word distance  metric  used
              for  word mesh alignment.  See the nnbbeesstt--llaattttiiccee(1)
              --ddiiccttiioonnaarryy option.

       --nnbbeesstt--ddeeccooddee _N
              Generate the up to _N highest scoring paths  through
              a  lattice  and  write them out in nnbbeesstt--ffoorrmmaatt(5),
              along with optional additional score files to store
              knowledge  sources encoded in the lattice.  Further
              options are needed to specify the  location  of  N-
              best  lists  and score files, described below under
              "N-BEST DECODING".  Duplicated Hypotheses that dif-
              fer only in pause and words specified with --iiggnnoorree--
              vvooccaabb are removed from the N-best output.   If  the
              --mmuullttiiwwoorrddss  option is specified, duplicates due to
              multiwords are also eliminated.

       --nnbbeesstt--dduupplliiccaatteess _K
              Allow up to _K duplicate word hypotheses to be  out-
              put in N-best decoding.

       --nnbbeesstt--mmaaxx--ssttaacckk _M
              Limits the depth of the hypothesis stack used in N-
              best decoding to _M entries, which may be useful for
              limiting memory use and runtime.

       --nnbbeesstt--vviitteerrbbii
              Use  a Viterbi algorithm to generate N-best, rather
              than A-star.  This uses less memory  but  may  take
              more time.

       --ppppll _f_i_l_e
              Read  sentences  from  _f_i_l_e and compute the maximum
              probability (of any path) assigned to them  by  the
              lattice  being processed.  Effectively, the lattice
              is treated as a (deficient)  language  model.   The
              output  detail  is controlled by the --ddeebbuugg option,
              similar to  nnggrraamm  --ppppll  output.   (In  particular,
              --ddeebbuugg  22  enables  tracing of lattice nodes corre-
              sponding to sentence prefixes.)  Pause  words  in
              _f_i_l_e are treated as regular words and have to match
              pause nodes in the lattice, unless --nnooppaauussee  speci-
              fied,  in  which  case  pauses  in both lattice and
              input sentences are ignored.

       The following options control  transition  weight  assign-
       ment:

       --oorrddeerr _n
              Set the maximal N-gram order to be used for transi-
              tion weight assignment (the default is 3).

       --llmm _f_i_l_e
              Read N-gram language model from _f_i_l_e.  This  option
              also   triggers  weight  reassignment  and  lattice
              expansion.

       --mmuullttiiwwoorrddss
              Resolve multiwords in the lattice without splitting
              nodes.   This  is useful in rescoring lattices con-
              taining multiwords with a LM does  not  use  multi-
              words.

       --ccllaasssseess _f_i_l_e
              Interpret  the  LM  as an N-gram over word classes.
              The expansions of the classes are given in _f_i_l_e  in
              ccllaasssseess--ffoorrmmaatt(5).   Tokens  in the LM that are not
              defined as classes in _f_i_l_e are assumed to be  plain
              words,  so  that  the  LM can contain mixed N-grams
              over both words and word classes.

       --ssiimmppllee--ccllaasssseess
              Assume a "simple" class model: each word is  member
              of at most one word class, and class expansions are
              exactly one word long.

       --mmiixx--llmm _f_i_l_e
              Read a second N-gram model for  interpolation  pur-
              poses.   The second and any additional interpolated
              models can also be class N-grams  (using  the  same
              --ccllaasssseess definitions).

       --ffaaccttoorreedd
              Interpret the files specified by --llmm, --mmiixx--llmm, etc.
              as  factored  N-gram  model  specifications.    See
              nnggrraamm(1) for more details.

       --llaammbbddaa _w_e_i_g_h_t
              Set the weight of the main model when interpolating
              with --mmiixx--llmm.  Default value is 0.5.

       --mmiixx--llmm22 _f_i_l_e

       --mmiixx--llmm33 _f_i_l_e

       --mmiixx--llmm44 _f_i_l_e

       --mmiixx--llmm55 _f_i_l_e

       --mmiixx--llmm66 _f_i_l_e

       --mmiixx--llmm77 _f_i_l_e

       --mmiixx--llmm88 _f_i_l_e

       --mmiixx--llmm99 _f_i_l_e
              Up to 9 more N-gram models  can  be  specified  for
              interpolation.

       --mmiixx--llaammbbddaa22 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa33 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa44 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa55 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa66 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa77 _w_e_i_g_h_t

       --mmiixx--llaammbbddaa88 _w_e_i_g_h_t