-
Notifications
You must be signed in to change notification settings - Fork 40
Default Pocketsphinx Configuration
watsonbox edited this page Oct 20, 2014
·
1 revision
require 'awesome_print'
ap Pocketsphinx::Configuration.default.details
[
[ 0] {
:name => "agc",
:type => :string,
:default => "none",
:required => false,
:value => "none",
:info => "Automatic gain control for c0 ('max', 'emax', 'noise', or 'none')"
},
[ 1] {
:name => "agcthresh",
:type => :float,
:default => 2.0,
:required => false,
:value => 2.0,
:info => "Initial threshold for automatic gain control"
},
[ 2] {
:name => "allphone",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Perform phoneme decoding with phonetic lm"
},
[ 3] {
:name => "allphone_ci",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Perform phoneme decoding with phonetic lm and context-independent units only"
},
[ 4] {
:name => "alpha",
:type => :float,
:default => 0.97,
:required => false,
:value => 0.97,
:info => "Preemphasis parameter"
},
[ 5] {
:name => "ascale",
:type => :float,
:default => 20.0,
:required => false,
:value => 20.0,
:info => "Inverse of acoustic model scale for confidence score calculation"
},
[ 6] {
:name => "aw",
:type => :integer,
:default => 1,
:required => false,
:value => 1,
:info => "Inverse weight applied to acoustic scores."
},
[ 7] {
:name => "backtrace",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Print results and backtraces to log file."
},
[ 8] {
:name => "beam",
:type => :float,
:default => 1.0e-48,
:required => false,
:value => 1.0e-48,
:info => "Beam width applied to every frame in Viterbi search (smaller values mean wider beam)"
},
[ 9] {
:name => "bestpath",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Run bestpath (Dijkstra) search over word lattice (3rd pass)"
},
[ 10] {
:name => "bestpathlw",
:type => :float,
:default => 9.5,
:required => false,
:value => 9.5,
:info => "Language model probability weight for bestpath search"
},
[ 11] {
:name => "bghist",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Bigram-mode: If TRUE only one BP entry/frame; else one per LM state"
},
[ 12] {
:name => "ceplen",
:type => :integer,
:default => 13,
:required => false,
:value => 13,
:info => "Number of components in the input feature vector"
},
[ 13] {
:name => "cmn",
:type => :string,
:default => "current",
:required => false,
:value => "current",
:info => "Cepstral mean normalization scheme ('current', 'prior', or 'none')"
},
[ 14] {
:name => "cmninit",
:type => :string,
:default => "8.0",
:required => false,
:value => "8.0",
:info => "Initial values (comma-separated) for cepstral mean when 'prior' is used"
},
[ 15] {
:name => "compallsen",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Compute all senone scores in every frame (can be faster when there are many senones)"
},
[ 16] {
:name => "debug",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Verbosity level for debugging messages"
},
[ 17] {
:name => "dict",
:type => :string,
:default => nil,
:required => true,
:value => "/usr/local/Cellar/cmu-pocketsphinx/HEAD/share/pocketsphinx/model/lm/en_US/cmu07a.dic",
:info => "Main pronunciation dictionary (lexicon) input file"
},
[ 18] {
:name => "dictcase",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Dictionary is case sensitive (NOTE: case insensitivity applies to ASCII characters only)"
},
[ 19] {
:name => "dither",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Add 1/2-bit noise"
},
[ 20] {
:name => "doublebw",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Use double bandwidth filters (same center freq)"
},
[ 21] {
:name => "ds",
:type => :integer,
:default => 1,
:required => false,
:value => 1,
:info => "Frame GMM computation downsampling ratio"
},
[ 22] {
:name => "fdict",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Noise word pronunciation dictionary input file"
},
[ 23] {
:name => "feat",
:type => :string,
:default => "1s_c_d_dd",
:required => false,
:value => "1s_c_d_dd",
:info => "Feature stream type, depends on the acoustic model"
},
[ 24] {
:name => "featparams",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "File containing feature extraction parameters."
},
[ 25] {
:name => "fillprob",
:type => :float,
:default => 1.0e-08,
:required => false,
:value => 1.0e-08,
:info => "Filler word transition probability"
},
[ 26] {
:name => "frate",
:type => :integer,
:default => 100,
:required => false,
:value => 100,
:info => "Frame rate"
},
[ 27] {
:name => "fsg",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Sphinx format finite state grammar file"
},
[ 28] {
:name => "fsgusealtpron",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Add alternate pronunciations to FSG"
},
[ 29] {
:name => "fsgusefiller",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Insert filler words at each state."
},
[ 30] {
:name => "fwdflat",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Run forward flat-lexicon search over word lattice (2nd pass)"
},
[ 31] {
:name => "fwdflatbeam",
:type => :float,
:default => 1.0e-64,
:required => false,
:value => 1.0e-64,
:info => "Beam width applied to every frame in second-pass flat search"
},
[ 32] {
:name => "fwdflatefwid",
:type => :integer,
:default => 4,
:required => false,
:value => 4,
:info => "Minimum number of end frames for a word to be searched in fwdflat search"
},
[ 33] {
:name => "fwdflatlw",
:type => :float,
:default => 8.5,
:required => false,
:value => 8.5,
:info => "Language model probability weight for flat lexicon (2nd pass) decoding"
},
[ 34] {
:name => "fwdflatsfwin",
:type => :integer,
:default => 25,
:required => false,
:value => 25,
:info => "Window of frames in lattice to search for successor words in fwdflat search "
},
[ 35] {
:name => "fwdflatwbeam",
:type => :float,
:default => 7.0e-29,
:required => false,
:value => 7.0e-29,
:info => "Beam width applied to word exits in second-pass flat search"
},
[ 36] {
:name => "fwdtree",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Run forward lexicon-tree search (1st pass)"
},
[ 37] {
:name => "hmm",
:type => :string,
:default => nil,
:required => false,
:value => "/usr/local/Cellar/cmu-pocketsphinx/HEAD/share/pocketsphinx/model/hmm/en_US/hub4wsj_sc_8k",
:info => "Directory containing acoustic model files."
},
[ 38] {
:name => "input_endian",
:type => :string,
:default => "little",
:required => false,
:value => "little",
:info => "Endianness of input data, big or little, ignored if NIST or MS Wav"
},
[ 39] {
:name => "jsgf",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "JSGF grammar file"
},
[ 40] {
:name => "kdmaxbbi",
:type => :integer,
:default => -1,
:required => false,
:value => -1,
:info => "Maximum number of Gaussians per leaf node in kd-Trees"
},
[ 41] {
:name => "kdmaxdepth",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Maximum depth of kd-Trees to use"
},
[ 42] {
:name => "kdtree",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "kd-Tree file for Gaussian selection"
},
[ 43] {
:name => "keyphrase",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Keyphrase to spot"
},
[ 44] {
:name => "kws",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "A file with keyphrases to spot, one per line"
},
[ 45] {
:name => "kws_plp",
:type => :float,
:default => 0.1,
:required => false,
:value => 0.1,
:info => "Phone loop probability for keyword spotting"
},
[ 46] {
:name => "kws_threshold",
:type => :float,
:default => 1.0,
:required => false,
:value => 1.0,
:info => "Threshold for p(hyp)/p(alternatives) ratio"
},
[ 47] {
:name => "latsize",
:type => :integer,
:default => 5000,
:required => false,
:value => 5000,
:info => "Initial backpointer table size"
},
[ 48] {
:name => "lda",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "File containing transformation matrix to be applied to features (single-stream features only)"
},
[ 49] {
:name => "ldadim",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Dimensionality of output of feature transformation (0 to use entire matrix)"
},
[ 50] {
:name => "lextreedump",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Whether to dump the lextree structure to stderr (for debugging), 1 for Ravi's format, 2 for Dot format, Larger than 2 will be treated as Ravi's format"
},
[ 51] {
:name => "lifter",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Length of sin-curve for liftering, or 0 for no liftering."
},
[ 52] {
:name => "lm",
:type => :string,
:default => nil,
:required => false,
:value => "/usr/local/Cellar/cmu-pocketsphinx/HEAD/share/pocketsphinx/model/lm/en_US/hub4.5000.DMP",
:info => "Word trigram language model input file"
},
[ 53] {
:name => "lmctl",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Specify a set of language model\n"
},
[ 54] {
:name => "lmname",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Which language model in -lmctl to use by default"
},
[ 55] {
:name => "logbase",
:type => :float,
:default => 1.0001,
:required => false,
:value => 1.0001,
:info => "Base in which all log-likelihoods calculated"
},
[ 56] {
:name => "logfn",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "File to write log messages in"
},
[ 57] {
:name => "logspec",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Write out logspectral files instead of cepstra"
},
[ 58] {
:name => "lowerf",
:type => :float,
:default => 133.33334,
:required => false,
:value => 133.33334,
:info => "Lower edge of filters"
},
[ 59] {
:name => "lpbeam",
:type => :float,
:default => 1.0e-40,
:required => false,
:value => 1.0e-40,
:info => "Beam width applied to last phone in words"
},
[ 60] {
:name => "lponlybeam",
:type => :float,
:default => 7.0e-29,
:required => false,
:value => 7.0e-29,
:info => "Beam width applied to last phone in single-phone words"
},
[ 61] {
:name => "lw",
:type => :float,
:default => 6.5,
:required => false,
:value => 6.5,
:info => "Language model probability weight"
},
[ 62] {
:name => "maxhmmpf",
:type => :integer,
:default => 10000,
:required => false,
:value => 10000,
:info => "Maximum number of active HMMs to maintain at each frame (or -1 for no pruning)"
},
[ 63] {
:name => "maxnewoov",
:type => :integer,
:default => 20,
:required => false,
:value => 20,
:info => "Maximum new OOVs that can be added at run time"
},
[ 64] {
:name => "maxwpf",
:type => :integer,
:default => -1,
:required => false,
:value => -1,
:info => "Maximum number of distinct word exits at each frame (or -1 for no pruning)"
},
[ 65] {
:name => "mdef",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Model definition input file"
},
[ 66] {
:name => "mean",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Mixture gaussian means input file"
},
[ 67] {
:name => "mfclogdir",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Directory to log feature files to"
},
[ 68] {
:name => "min_endfr",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Nodes ignored in lattice construction if they persist for fewer than N frames"
},
[ 69] {
:name => "mixw",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Senone mixture weights input file (uncompressed)"
},
[ 70] {
:name => "mixwfloor",
:type => :float,
:default => 1.0e-07,
:required => false,
:value => 1.0e-07,
:info => "Senone mixture weights floor (applied to data from -mixw file)"
},
[ 71] {
:name => "mllr",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "MLLR transformation to apply to means and variances"
},
[ 72] {
:name => "mmap",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Use memory-mapped I/O (if possible) for model files"
},
[ 73] {
:name => "ncep",
:type => :integer,
:default => 13,
:required => false,
:value => 13,
:info => "Number of cep coefficients"
},
[ 74] {
:name => "nfft",
:type => :integer,
:default => 512,
:required => false,
:value => 512,
:info => "Size of FFT"
},
[ 75] {
:name => "nfilt",
:type => :integer,
:default => 40,
:required => false,
:value => 40,
:info => "Number of filter banks"
},
[ 76] {
:name => "nwpen",
:type => :float,
:default => 1.0,
:required => false,
:value => 1.0,
:info => "New word transition penalty"
},
[ 77] {
:name => "pbeam",
:type => :float,
:default => 1.0e-48,
:required => false,
:value => 1.0e-48,
:info => "Beam width applied to phone transitions"
},
[ 78] {
:name => "pip",
:type => :float,
:default => 1.0,
:required => false,
:value => 1.0,
:info => "Phone insertion penalty"
},
[ 79] {
:name => "pl_beam",
:type => :float,
:default => 1.0e-10,
:required => false,
:value => 1.0e-10,
:info => "Beam width applied to phone loop search for lookahead"
},
[ 80] {
:name => "pl_pbeam",
:type => :float,
:default => 1.0e-05,
:required => false,
:value => 1.0e-05,
:info => "Beam width applied to phone loop transitions for lookahead"
},
[ 81] {
:name => "pl_window",
:type => :integer,
:default => 0,
:required => false,
:value => 0,
:info => "Phoneme lookahead window size, in frames"
},
[ 82] {
:name => "rawlogdir",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Directory to log raw audio files to"
},
[ 83] {
:name => "remove_dc",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Remove DC offset from each frame"
},
[ 84] {
:name => "remove_noise",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Remove noise with spectral subtraction in mel-energies"
},
[ 85] {
:name => "remove_silence",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Enables VAD, removes silence frames from processing"
},
[ 86] {
:name => "round_filters",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Round mel filter frequencies to DFT points"
},
[ 87] {
:name => "samprate",
:type => :float,
:default => 16000.0,
:required => false,
:value => 16000.0,
:info => "Sampling rate"
},
[ 88] {
:name => "seed",
:type => :integer,
:default => -1,
:required => false,
:value => -1,
:info => "Seed for random number generator; if less than zero, pick our own"
},
[ 89] {
:name => "sendump",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Senone dump (compressed mixture weights) input file"
},
[ 90] {
:name => "senlogdir",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Directory to log senone score files to"
},
[ 91] {
:name => "senmgau",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Senone to codebook mapping input file (usually not needed)"
},
[ 92] {
:name => "silprob",
:type => :float,
:default => 0.005,
:required => false,
:value => 0.005,
:info => "Silence word transition probability"
},
[ 93] {
:name => "smoothspec",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Write out cepstral-smoothed logspectral files"
},
[ 94] {
:name => "svspec",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Subvector specification (e.g., 24,0-11/25,12-23/26-38 or 0-12/13-25/26-38)"
},
[ 95] {
:name => "tmat",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "HMM state transition matrix input file"
},
[ 96] {
:name => "tmatfloor",
:type => :float,
:default => 0.0001,
:required => false,
:value => 0.0001,
:info => "HMM state transition probability floor (applied to -tmat file)"
},
[ 97] {
:name => "topn",
:type => :integer,
:default => 4,
:required => false,
:value => 4,
:info => "Maximum number of top Gaussians to use in scoring."
},
[ 98] {
:name => "topn_beam",
:type => :string,
:default => "0",
:required => false,
:value => "0",
:info => "Beam width used to determine top-N Gaussians (or a list, per-feature)"
},
[ 99] {
:name => "toprule",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Start rule for JSGF (first public rule is default)"
},
[100] {
:name => "transform",
:type => :string,
:default => "legacy",
:required => false,
:value => "legacy",
:info => "Which type of transform to use to calculate cepstra (legacy, dct, or htk)"
},
[101] {
:name => "unit_area",
:type => :boolean,
:default => true,
:required => false,
:value => true,
:info => "Normalize mel filters to unit area"
},
[102] {
:name => "upperf",
:type => :float,
:default => 6855.4976,
:required => false,
:value => 6855.4976,
:info => "Upper edge of filters"
},
[103] {
:name => "usewdphones",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Use within-word phones only"
},
[104] {
:name => "uw",
:type => :float,
:default => 1.0,
:required => false,
:value => 1.0,
:info => "Unigram weight"
},
[105] {
:name => "vad_postspeech",
:type => :integer,
:default => 50,
:required => false,
:value => 50,
:info => "Num of silence frames to trigger vad from speech to silence."
},
[106] {
:name => "vad_prespeech",
:type => :integer,
:default => 10,
:required => false,
:value => 10,
:info => "Num of speech frames to trigger vad from silence to speech."
},
[107] {
:name => "vad_threshold",
:type => :float,
:default => 2.0,
:required => false,
:value => 2.0,
:info => "Threshold for decision between noise and silence frames. Log-ratio between signal level and noise level."
},
[108] {
:name => "var",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Mixture gaussian variances input file"
},
[109] {
:name => "varfloor",
:type => :float,
:default => 0.0001,
:required => false,
:value => 0.0001,
:info => "Mixture gaussian variance floor (applied to data from -var file)"
},
[110] {
:name => "varnorm",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Variance normalize each utterance (only if CMN == current)"
},
[111] {
:name => "verbose",
:type => :boolean,
:default => false,
:required => false,
:value => false,
:info => "Show input filenames"
},
[112] {
:name => "warp_params",
:type => :string,
:default => nil,
:required => false,
:value => nil,
:info => "Parameters defining the warping function"
},
[113] {
:name => "warp_type",
:type => :string,
:default => "inverse_linear",
:required => false,
:value => "inverse_linear",
:info => "Warping function type (or shape)"
},
[114] {
:name => "wbeam",
:type => :float,
:default => 7.0e-29,
:required => false,
:value => 7.0e-29,
:info => "Beam width applied to word exits"
},
[115] {
:name => "wip",
:type => :float,
:default => 0.65,
:required => false,
:value => 0.65,
:info => "Word insertion penalty"
},
[116] {
:name => "wlen",
:type => :float,
:default => 0.025625,
:required => false,
:value => 0.025625,
:info => "Hamming window length"
}
]