📄 hiddenmarkovmodel.h

📁 这是一个从音频信号里提取特征参量的程序
💻 H
📖 第 1 页 / 共 2 页
字号:
12 下一页
// file: $isip/class/pr/HiddenMarkovModel/HiddenMarkovModel.h// version: $Id: HiddenMarkovModel.h,v 1.70 2003/04/16 13:23:33 alphonso Exp $//// make sure definitions are only made once//#ifndef ISIP_HIDDEN_MARKOV_MODEL#define ISIP_HIDDEN_MARKOV_MODEL#ifndef ISIP_DOUBLE#include <Double.h>#endif#ifndef ISIP_ULONG#include <Ulong.h>#endif#ifndef ISIP_VECTOR_BYTE#include <VectorByte.h>#endif#ifndef ISIP_MATRIX_FLOAT#include <MatrixFloat.h>#endif#ifndef ISIP_FRONT_END#include <FrontEnd.h>#endif#ifndef ISIP_FRONT_END#include <FrontEnd.h>#endif#ifndef ISIP_TRAIN_NODE#include <TrainNode.h>#endif#ifndef ISIP_WRAPPER#include <Wrapper.h>#endif#ifndef ISIP_HASH_KEY#include <HashKey.h>#endif#ifndef ISIP_LANGUAGE_MODEL#include <LanguageModel.h>#endif#ifndef ISIP_HIERARCHICAL_SEARCH#include <HierarchicalSearch.h>#endif#ifndef ISIP_STACK_SEARCH#include <StackSearch.h>#endif#ifndef ISIP_TRANSCRIPTION_DATABASE#include <TranscriptionDatabase.h>#endif#ifndef ISIP_AUDIO_DATABASE#include <AudioDatabase.h>#endif#ifndef ISIP_ANNOTATION_GRAPH#include <AnnotationGraph.h>#endif#ifndef ISIP_ANNOTATION#include <Annotation.h>#endif#ifndef ISIP_SDB#include <Sdb.h>#endif#ifndef ISIP_PHONETIC_DECISION_TREE#include <PhoneticDecisionTree.h>#endif#ifndef ISIP_ANNOTATION_GRAPH#include <AnnotationGraph.h>#endif// HiddenMarkovModel: an intermediate class providing base//                    for Decode and Train algorithm using//                    Viterbi or Baum-Welch implementation//class HiddenMarkovModel {    //--------------------------------------------------------------------------  //  // public constants  //  //--------------------------------------------------------------------------public:    // define the class name  //  static const String CLASS_NAME;  //----------------------------------------  //  // other important constants  //  //----------------------------------------  // define the algorithm choices  //  enum ALGORITHM { DECODE = 0, TRAIN, MODEL_CREATION, INITIALIZE,		   MIXTURE_SPLITTING,		   CONTEXT_GENERATION, TRAIN_PARAMETER_TYING,		   TEST_PARAMETER_TYING, FORCED_ALIGNMENT,		   ADAPT_DECODE, DEF_ALGORITHM = DECODE };    // define the implementation choices  //  enum IMPLEMENTATION { VITERBI = 0, STACK, BAUM_WELCH, GLOBAL, 			VARIANCE_SPLITTING, SYMBOL_GENERATION, ML,			MODEL_GENERATION, DEF_IMPLEMENTATION = VITERBI };  // define the adaptation supervision choices  //  enum ADAPT_SUPERVISION_MODE { SUPERVISED = 0, UNSUPERVISED,                                DEF_ADAPT_SUPERVISION_MODE = UNSUPERVISED };  // define the adaptation sequence choices  //  enum ADAPT_SEQUENCE_MODE { INCREMENTAL = 0, BATCH,                             DEF_ADAPT_SEQUENCE_MODE = INCREMENTAL };  // define the context choices  //  enum CONTEXT_MODE { SYMBOL_ONLY = 0, SYMBOL_INTERNAL, CROSS_SYMBOL, 		      DEF_CONTEXT_MODE = SYMBOL_ONLY };  // define the cross-word choices  //  enum FUNCTION_MODE { NONE = 0, GRAMMAR_DECODING, NETWORK_DECODING,		       NETWORK_RESCORING, NETWORK_GENERATION,		       NETWORK_VERIFICATION, ACCUMULATE, UPDATE,		       DEF_FUNCTION_MODE = NONE };      // define output modes  //  enum OUTPUT_MODE { FILE = 0, TRANSFORM, LIST, DATABASE,		     DEF_OUTPUT_MODE = FILE };  // define update modes  //  enum UPDATE_MODE { TRANSITIONS = 0, OBSERVATIONS, ALL, DEF_UPDATE_MODE = ALL };    // define output formats  //  enum OUTPUT_FORMAT { NIST_SCORE = 0, NATIVE, JSGF, DEF_OUTPUT_FORMAT = NATIVE };  // define output types  //  enum OUTPUT_TYPE { TEXT = 0, BINARY, DEF_OUTPUT_TYPE = BINARY };    // define internal on/off states  //    static const byte OFF = 0;  static const byte ON = 1;  // define internal default values  //  static const String DEF_TRANSCRIPTION_LEVEL;  static const String DEF_FORCE_ALIGNMENT_LEVEL;    // define internal constants  //  static const float DEF_WEIGTH_ONE = 0;  static const float DEF_WEIGTH_HALF = -0.693147;  static const float DEF_BETA_PRUNING_THRESHOLD = 3.40282347e+38F;    static const float DEF_FLT_MIN = 1.17549435e-38F;  static const float DEF_FLT_MAX = 3.40282347e+38F;  static const float DEF_DBL_MIN = 2.2250738585072014e-308;  static const float DEF_DBL_MAX = 1.7976931348623157e+308;    // define the static NameMap objects  //  static const NameMap IMPL_MAP;  static const NameMap ALGO_MAP;  static const NameMap ADAPT_SUP_MODE_MAP;  static const NameMap ADAPT_SEQ_MODE_MAP;  static const NameMap CONTEXT_MODE_MAP;  static const NameMap FUNCTION_MODE_MAP;      static const NameMap OUTPUT_MODE_MAP;  static const NameMap OUTPUT_FORMAT_MAP;  static const NameMap OUTPUT_TYPE_MAP;      static const NameMap UPDATE_MODE_MAP;     //----------------------------------------  //  // i/o related constants  //  //----------------------------------------    static const String DEF_PARAM;  static const String PARAM_UPDATE_LEVELS;  static const String PARAM_NUM_LEVELS;  static const String PARAM_NUM_MIXTURES;    static const String PARAM_NUM_ITERATIONS;  static const String PARAM_VAR_FLOOR;  static const String PARAM_VAR_FLOOR_FILE;  static const String PARAM_BETA_THRESHOLD;  static const String PARAM_TRANSCRIPTION_LEVEL;  static const String PARAM_FORCE_ALIGNMENT_LEVEL;  static const String PARAM_TRANS_DB_FILE;  static const String PARAM_AUDIO_DB_FILE;  static const String PARAM_MIN_MPD;  static const String PARAM_MIN_OCCUPANCY;  static const String PARAM_MIN_MODEL_COUNT;      static const String PARAM_ALGORITHM;  static const String PARAM_IMPLEMENTATION;  static const String PARAM_ADAPT_SUPERVISION_MODE;  static const String PARAM_ADAPT_SEQUENCE_MODE;    static const String PARAM_CONFIGURATION_FILE;  static const String PARAM_FRONT_END_FILE;    static const String PARAM_LANGUAGE_MODEL_FILE;  static const String PARAM_ACOUSTIC_MODEL_FILE;      static const String PARAM_UPDATE_LANGUAGE_MODEL_FILE;  static const String PARAM_UPDATE_ACOUSTIC_MODEL_FILE;  static const String PARAM_ACCUMULATOR_FILE;  static const String PARAM_ACCUMULATOR_LIST;    static const String PARAM_OUTPUT_MODE;  static const String PARAM_TRANSCRIPTION_BOUNDS;  static const String PARAM_OUTPUT_FORMAT;  static const String PARAM_OUTPUT_TYPE;    static const String PARAM_OUTPUT_LEVELS;  static const String PARAM_UPDATE_MODE;  static const String PARAM_FUNCTION_MODE;  static const String PARAM_CONTEXT_MODE;  static const String PARAM_CONTEXT_LEVEL;  static const String PARAM_CONTEXT_ORDER;      static const String PARAM_OUTPUT_FILE;  static const String PARAM_OUTPUT_LIST;  static const String PARAM_CONTEXT_LIST;  static const String PARAM_QUES_ANS_FILE;  static const String PARAM_PHONETIC_DT_FILE;  static const String PARAM_PHONETIC_DT_SPLIT_THRESHOLD;  static const String PARAM_PHONETIC_DT_MERGE_THRESHOLD;  static const String PARAM_PHONETIC_DT_NUM_OCC_THRESHOLD;  static const String OPTION_TEXT;  static const String OPTION_BINARY;    static const String OPTION_OUT_MODE_FILE;  static const String OPTION_OUT_MODE_TRANSFORM;  static const String OPTION_OUT_MODE_LIST;  static const String OPTION_OUT_FORM_NIST_SCORE;  static const String OPTION_OUT_FORM_NATIVE;  static const String OPTION_OUT_FORM_JSGF;  static const String OPTION_TRANSCRIPTION_LEVEL;  static const String OPTION_UPDATE_TRANSITIONS;  static const String OPTION_UPDATE_OBSERVATIONS;  static const String OPTION_UPDATE_ALL;  static const String OPTION_FUNCTION_NONE;  static const String OPTION_FUNCTION_GRAMMAR_DECODING;  static const String OPTION_FUNCTION_NETWORK_DECODING;  static const String OPTION_FUNCTION_NETWORK_RESCORING;  static const String OPTION_FUNCTION_NETWORK_GENERATION;  static const String OPTION_FUNCTION_NETWORK_VERIFICATION;  static const String OPTION_FUNCTION_ACCUMULATE;  static const String OPTION_FUNCTION_UPDATE;      //----------------------------------------  //  // default values and arguments  //  //----------------------------------------    // default values  //  static const boolean DEF_VERIFY = false;  static const boolean DEF_STREAM = false;  static const boolean DEF_TRANSCRIPTION_BOUNDS = false;    static const long DEF_NUM_LEVELS = 3;  static const long DEF_NUM_MIXTURES = 1;    static const long DEF_NUM_ITERATIONS = 1;  static const long DEF_ALIGNMENT_LEVEL = -1;    static const long DEF_INITIAL_LEVEL = 0;    static const long DEF_CONTEXT_LEVEL = 0;  static const long DEF_CONTEXT_ORDER = 3;      static const long DEF_CAPACITY = 12000;    static const float DEF_MIN_MPD = 1000;  static const float DEF_MIN_OCCUPANCY = 1.0e-10;  static const long DEF_MIN_MODEL_COUNT = 1;  static const float DEF_VAR_FLOOR = 2e-4;    //----------------------------------------  //  // error codes  //  //----------------------------------------    static const long ERR = (long)100000;  static const long ERR_ADAPT_NO_GAUSSIAN = (long)100050;    //---------------------------------------------------------------------------  //  // protected data  //  //---------------------------------------------------------------------------protected:    // algorithm name  //  ALGORITHM algorithm_d;    // implementation name  //  IMPLEMENTATION implementation_d;  // adaptation supervision mode  //  ADAPT_SUPERVISION_MODE adapt_supervision_mode_d;  // adaptation sequence mode  //  ADAPT_SEQUENCE_MODE adapt_sequence_mode_d;  // context mode  //  CONTEXT_MODE context_mode_d;  // cross-word mode  //  FUNCTION_MODE function_mode_d;      // output mode  //  OUTPUT_MODE output_mode_d;  // update mode  //  UPDATE_MODE update_mode_d;      // output format  //  OUTPUT_FORMAT output_format_d;  // output type  //  OUTPUT_TYPE output_type_d;    // update levels  //  String update_levels_str_d;    // vector that contains the indices of re-estimation update levels  //  VectorByte update_levels_d;    // output levels  //  String output_levels_str_d;    // vector that contains the indices of hypothesis output levels  //  VectorByte output_levels_d;    // parameter file  //  Filename param_file_d;  // language model file  //  Filename lm_model_file_d;  // statistical model file  //  Filename ac_model_file_d;  // configuration file  //  Filename cnfg_file_d;  // fornt-end file  //  Filename fend_file_d;    // updated model file  //  Filename update_lm_model_file_d;  // updated statistics file  //  Filename update_ac_model_file_d;      // transcription file  //  Filename transcription_db_file_d;    // audio database file  //  Filename audio_db_file_d;  // output file  //  Filename output_file_d;  // accumulator file  //  Filename accum_file_d;  // accumulator list  //  Filename accum_list_d;    // output list file  //  Filename output_list_d;  // context generation file  //  Filename context_list_d;    // variance floor file  //  Filename variance_floor_file_d;    // phonetic question-answers file  //  Filename ques_ans_file_d;    // phonetic decision-tree file  //  Filename phonetic_dt_file_d;    // split-threshold for the phonetic-decision-tree  //  Float phonetic_dt_split_threshold_d;  // merge-threshold for the phonetic-decision-tree  //  Float phonetic_dt_merge_threshold_d;  // num_occ-threshold for the phonetic-decision-tree  //  Float phonetic_dt_num_occ_threshold_d;  // variance floor  //  Float variance_floor_d;  // beta pruning threshold  //  Float beta_threshold_d;    // transcription level  //  String transcription_level_d;  // alignment level  //  String force_alignment_level_d;    // force alignment level  //  Long alignment_level_d;  // initial level  //  Long initial_level_d;    // context level  //  Long context_level_d;  // context order  //  Long context_order_d;      // streaming input flag  //  boolean stream_d;    // verification mode flag  //  boolean verify_d;    // front end  //  FrontEnd fe_d;    // utterance trellis  //  BiGraph<TrainNode>* trellis_d;  // hierarchical search  //  HierarchicalSearch search_engine_d;  // stack search  //  StackSearch stack_engine_d;  // number of search levels  //  Long num_levels_d;  // number of mixtures  //  Long num_mixtures_d;    // number of iterations of training  //  Long num_iterations_d;    // minimum model probability deviance  //  Float min_mpd_d;  // floor on the occupancy probability  //  Float min_occupancy_d;  // minimum number of times a model must occur before update  //  Long min_model_count_d;    // transcription database  //  TranscriptionDatabase transcription_db_d;  // audio database  //  AudioDatabase audio_db_d;  // flag indicates if non-speech symbols are forced at transcription ends  //  Boolean transcription_bounds_d;    // debug level  //  static Integral::DEBUG debug_level_d;    // verbosity  //  static Integral::DEBUG verbosity_d;  // static memory manager  //  static MemoryManager mgr_d;    //---------------------------------------------------------------------------  //  // required public methods  //  //---------------------------------------------------------------------------public:    // method: name  //  static const String& name() {    return CLASS_NAME;  }  // other static methods  //     static boolean diagnose(Integral::DEBUG debug_level);  // method: setDebug  //  boolean setDebug(Integral::DEBUG debug_level) {    debug_level_d = debug_level;    HierarchicalSearch::setDebug(debug_level);    StackSearch::setDebug(debug_level);    SearchNode::setDebug(debug_level);    Trace::setDebug(debug_level);    Instance::setDebug(debug_level);    History::setDebug(debug_level);    Context::setDebug(debug_level);        return true;  }    // constructor(s)/destructor(s)  //  ~HiddenMarkovModel();  HiddenMarkovModel();  HiddenMarkovModel(const HiddenMarkovModel& arg);  // method: assign  //  boolean assign(const HiddenMarkovModel& arg) {    return Error::handle(name(), L"assign", Error::ARG, __FILE__, __LINE__);  }    // i/o methods  //  long sofSize() const;  boolean read(Sof& sof, long tag, const String& name = CLASS_NAME);  boolean write(Sof& sof, long tag, const String& name = CLASS_NAME) const;    boolean readData(Sof& sof, const String& pname = DEF_PARAM,		   long size = SofParser::FULL_OBJECT,		   boolean param = true,		   boolean nested = false);  boolean writeData(Sof& sof, const String& param = DEF_PARAM) const;    // method: new  //  static void* operator new(size_t size) {    return mgr_d.get();  }  // method: new[]  //  static void* operator new[](size_t size) {    return mgr_d.getBlock(size);  }  // method: delete  //  static void operator delete(void* ptr) {    mgr_d.release(ptr);  }  // method: delete[]  //  static void operator delete[](void* ptr) {    mgr_d.releaseBlock(ptr);  }  // method: setGrowSize  //  static boolean setGrowSize(long grow_size) {    return mgr_d.setGrow(grow_size);  }
12 下一页
⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -