# # Sample Jconf configuration file # for Julius library rev.4.2 # # 1) Options can also be specified in command line option. # The values are default values in Julius. # 2) For file name, relative path must be relative to THIS FILE. # 3) Texts after '#' at each line will be ignored. If you want to specify # '#', use `\#'. # 4) Each line should be no longer than 512 bytes. # # ###################################################################### #### JULIUS APPLICATION OPTION (not a part of JuliusLib) ###################################################################### #-outfile # save each result in separate file #-separatescore # print AM / LM scores separately #-callbackdebug # print callback names at call for debug #-charconv from to # print with character set conversion #-nocharconv # disable "-charconv" #-module # start in module mode #-record dir # record each inputs into dir #-logfile file # redirect logs to file #-nolog # disable all logs #-help # print help, and exit ###################################################################### #### GLOBAL OPTIONS ###################################################################### #### #### Misc. Options #### #-C jconffile # include jconf file at here #-version # print version info to stderr, and exit #-setting # print version info to stderr, and exit #-quiet # output less log #-debug # output more log for debug #-check wchmm # for debug, enter debug shell mode #-check trellis # for debug, enter debug shell mode #-check triphone # for debug, enter debug shell mode #-demo # same as "-quiet -progout" #### #### Audio Input #### #-input mic # live microphone #-input rawfile # wavefile #-input mfcfile # MFCC file (HTK Parameter file) #-input stdin # waveform from standard input #-input adinnet # waveform via network client #-input netaudio # DatLink server #-input oss # OSS API input (if available) #-input alsa # ALSA API input (if available) #-input esd # ESounD daemon input (if available) #-filelist filename # input file list #-notypecheck # does not check parameter type of input #-48 # 48kHz sampling > 16kHz conv. (16kHz only) #-NA devname # hostname for DatLink server #-adport 5530 # port number for adinnet #-nostrip # do not strip zero samples #-zmean # remove DC offset (use long input average) #-nozmean # disable "-zmean" specified before #### #### Speech segment detection by level and zero-cross #### #### default: on for microphone, off for other sources #### #-cutsilence # detection on #-nocutsilence # detection off #-lv 2000 # level threshold (0-32767) #-zc 60 # zero-cross threshold (times in sec.) #-headmargin 300 # head silence margin (msec) #-tailmargin 400 # tail silence margin (msec) #-rejectshort 0 # reject shorter input (msec) #### #### Input rejection by average power (EXPERIMENTAL) #### #### This will be enabled by "--enable-power-reject" on compilation. #### Should be used with Decoder VAD or GMM VAD. #### Valid for real-time input only. #### #-powerthres 9.0 # reject input by avg. energy #### #### Gaussian Mixture Model #### #### GMM will be used for input rejection by accumurated score, or #### for GMM-based frontend VAD when "--enable-gmm-vad" specified. #### #### NOTE: If you use MFCC for the GMM which is different from AM, you #### should also set the parameters like other AM with an option "-AM_GMM". #### If "-AM_GMM" is not used, Julius assume GMM to use the same #### parameter as the first AM. #### #-gmm hmmdefs # GMM definitions in HTK format #-gmmnum 10 # num of Gaussians to be computed per GMM #-gmmreject string # comma-separated list of GMM name to reject #### GMM_VAD #-gmmmargin 20 # head margin for GMM based VAD in frames #### #### Decoding option #### #### Real-time processing means concurrent processing of MFCC computation #### and 1st pass decoding. By default, real-time processing on the #### 1st pass is on for microphone / adinnet / netaudio input, and #### off for others. #### #-realtime # force real-time processing #-norealtime # force non real-time processing #### #### Plug-in #### #### See plugin/00readme.txt for detail #### #-plugindir ./plugin:/usr/local/share/julius/plugins ###################################################################### #### INSTANCE DEFINITION FOR MULTI DECODING ###################################################################### #### #### The following three argument will create a new configuration set #### with default parameters, and switch current set to it. Jconf #### parameters specified after the option will be set into the current #### set. #### #### To do multi-model decoding, these argument should be specified at #### the first of each model / search instances with different names. #### Any options before the first instance definition will be IGNORED. #### #### When no instance definition is found (as older version of Julius), #### all the options are assigned to a default instance named "_default". #### #### Please note that decoding with a single LM and multiple AMs is not #### fully supported. For example, you may want to construct the #### jconf file as this: #### #### -AM am_1 -AM am_2 #### -LM lm (LM spec..) #### -SR search1 am_1 lm #### -SR search2 am_2 lm #### #### This type of model sharing is not supported yet, since some part #### of LM processing depends on the assigned AM. Instead, you can #### get the same result by defining the same LMs for each AM, like this: #### #### -AM am_1 -AM am_2 #### -LM lm_1 (LM spec..) #### -LM lm_2 (same LM spec..) #### -SR search1 am_1 lm_1 #### -SR search2 am_2 lm_2 #### ## Create a new AM configuration set, and switch current to it. ## You should give a unique name. #-AM name ## Create a new LM configuration set, and switch current to it. ## You should give a unique name. #-LM name ## Create a new Search configuration set with AM and LM, and switch ## current to it. AM and LM name can be either name or ID number. #-SR name am_name_or_id lm_name_or_id ## Switch current AM to special one reserved for GMM, to specify ## analysis parameter for GMM. Be sure not to confuse with normal AM ## configuration. # -AM_GMM ## When using instance declarations, global options should be placed ## at top before any instance declaration, or after this option below. ## This option is only a switcher and can be used anywhere anytime. # -GLOBAL ## This option disables the strict section checkings and back to 4.0 # -nosectioncheck ###################################################################### #### LANGUAGE MODEL (-LM) ###################################################################### #### #### Only one type of LM can be specified for a LM configuration. #### #### #### N-gram #### #-d binary_ngram_file # N-gram in Julius binary format #-nlr ngram # forward (left-to-right) N-gram #-nrl rev_ngram # backward (right-to-left) N-gram #-v dictfile # word dictionary ## param. #-silhead "" # beginning-of-sentence (silence) word #-siltail "" # end-of-sentence (silence) word #-mapunk "" # word to which unknown words should be mapped #-iwspword # add a pause word to the dictionary #-iwspentry " [sp] sp sp" # word that will be added by "-iwspword" #-sepnum 150 # num of high freq words to linearize #-adddict dictfile # append additional word dictionary #-addword entry # append additional word entry #### #### Grammar #### #### "-gram", "-gramlist" can be used multiple times #-gram gramprefix # (comma-separated list of) grammar file prefix #-gramlist txtfile # text file containing grammar prefixes #-dfa dfafile -v dictfile # specify DFA and dictionary separately #-nogram # reset all grammar list already specified #### #### Isolated Word #### #-w dictfile # word dictionary #-wlist txtfile # text file containing dictionaries #-nogram # reset all dictfiles already specified ## param. #-wsil silB silE NULL # head / tail silence models to be appended #### #### User-defined LM #### #-userlm # declare to use user LM defined in program #### #### misc LM option #### #-forcedict # skip error word entries (no stop on error) ###################################################################### #### ACOUSTIC MODEL (-AM) (-AM_GMM) ###################################################################### #### #### Acoustic analysis parameters are included in this section, since #### the AM defines the required parameter. You can use different MFCC #### type for each AM. #### For GMM, the same parameter should be specified after "-AM_GMM" #### #### When using multiple AM, the values of "-smpPeriod", "-smpFreq", #### "-fsize" and "-fshift" should be the same among all AM. #### ## Acoustic model #-h hmmfile # acoustic HMM (ascii or Julius binary) #-hlist logicaltri # HMMList to map logical phone to physical #-tmix 2 # # of mixture to compute in a mixture PDF #-spmodel "sp" # name of a short-pause silence model #-multipath # force enable MULTI-PATH model handling #-gprune {safe|heuristic|beam|none|default} # Gaussian pruning method #-iwcd1 {max|avg|best 3} # Inter-word triphone approximation method #-iwsppenalty -1.0 # pause insertion penalty for "-iwsp" #-gshmm hmmfile # HMM for Gaussian mixture selection #-gsnum 24 # Threshold number of HMM for gshmm ## Analysis #-smpPeriod 625 # sampling period (ns) (= 10000000 / smpFreq) #-smpFreq 16000 # sampling rate (Hz) #-fsize 400 # window size (samples) #-fshift 160 # frame shift (samples) #-preemph 0.97 # pre-emphasis coef. #-fbank 24 # number of filterbank channels #-ceplif 22 # cepstral liftering coef. #-rawe # use raw energy #-norawe # disable "-rawe" (this is default) #-enormal # normalize log energy #-noenormal # disable "-enormal" (this is default) #-escale 1.0 # scaling log energy for enormal #-silfloor 50.0 # energy silence floor in dB for enormal #-delwin 2 # delta window (frames) #-accwin 2 # acceleration window (frames) #-hifreq -1 # cut-off hi frequency (Hz) (-1: disable) #-lofreq -1 # cut-off low frequency (Hz) (-1: disable) #-zmeanframe # frame-wise DC offset removal (same as HTK) #-nozmeanframe # disable "-zmeanframe" (this is default) ## Cepstral mean / variance normalization #-cmnload filename # load initial cep. mean / variance on startup #-cmnsave filename # save cep. mean / variance at each input end #-cmnupdate # update beginning cep. data at each input #-cmnnoupdate # keep initial mean, disable "-cmnupdate" #-cmnmapweight 100.0 # weight for MAP-CMN #-cvn # enable variance normalization ## Vocal tract length normalization (VTLN) #-vtln 1.0 300 4800 # enable VTLN (alpha, lowerfreq, upperfreq) ## Spectral subtraction (default: disabled) #-sscalc # do SS, estimate noise from head sil #-sscalclen 300 # length of head silence for "-sscalc" (msec) #-ssload filename # do SS, load noise spectrum saved by "mkss" #-ssalpha 2.0 # alpha coef. for spectral subtraction #-ssfloor 0.5 # spectral floor coef. ## Others #-htkconf configfile # load analysis settings from HTK Config file ###################################################################### #### RECOGNIZER (-SR) ###################################################################### #### #### Default values for beam width and LM weights will change #### according to compile-time setup of JuliusLib and model specification. #### Please see the startup log for the actual values. #### #### #### parameter (common) #### #-inactive # start this process with inactive status #-1pass # perform only the 1st pass, omit 2nd pass #-no_ccd # switch off the phone context dependency #-force_ccd # force on the phone context dependency #-cmalpha 0.05 # CM alpha value #-iwsp # append a skippable sp at all word ends #-transp 0.0 # transition penalty for transparent words #### #### parameter (1st pass) #### #-lmp weight penalty # LM weight and word insertion penalty (pass1) #-penalty1 penalty # word insertion penalty for grammar (pass1) #-b width # beam width (# of nodes) #-bs score # beam width (score) #-nlimit 3 # with enable-wpair-nlimit, set max N at nodes #-progout # progressive output while decoding #-proginterval 300 # output interval in msec for "-progout" #### #### parameter (2nd pass) #### #-lmp2 weight penalty # LM weight and word insertion penalty (pass2) #-penalty2 penalty # word insertion penalty for grammar (pass2) #-b2 width # envelope beam width of 2nd pass (#word) #-sb 80.0 # envelope score width at 2nd pass #-s 500 # hypotheses stack size on 2nd pass (#hypo) #-m 2000 # hypotheses overflow threshold (#hypo) #-n n # num of sentences to find #-output 1 # num of sentences to output as result #-lookuprange 5 # hypo. lookup range at word expansion (#frame) #-looktrellis # expand only trellis words in grammar #-fallback1pass # output 1st pass result when 2nd pass fails #### #### short-pause segmentation (and decoder-based VAD) #### #-spsegment # enable sp segmentation (or decoder VAD) #-spdur 10 # # of frames to detect a short pause #-pausemodels string # comma-separated pause model names #### for decoder-VAD #-spmargin 40 # backstep margin at trigger up (frame) #-spdelay 4 # decision delay at trigger up (frame) #### #### lattice output #### #-lattice # output result in word graph (aka -graphout) #-graphrange 0 # merge same words nearby, -1 to disable merge #-graphcut 80 # graph depth cut threshold (in depth) #-graphboundloop 20 # max itertations for boundary adjustment loop #-graphsearchdelay # activate an alternate generation algorithm #-nographsearchdelay # disable "-graphsearchdelay" #### #### confusion network output #### #-confnet # enable confusion network output #-noconfnet # disable confusion network output #### #### multi-grammar output (for grammar and isolated word) #### #-multigramout # output max hypo for each grammar #-nomultigramout # disable "-multigramout" #### #### forced alignment #### #-walign # enable alignment for result at word level #-palign # enable alignment for result at phoneme level #-salign # enable alignment for result at state level ################################################################# end of file