InverseContextFst Class Reference

#include <context-fst.h>

Inheritance diagram for InverseContextFst:

Collaboration diagram for InverseContextFst:

Public Types
typedef StdArc	Arc
typedef StdArc::StateId	StateId
typedef StdArc::Weight	Weight
typedef StdArc::Label	Label
Public Types inherited from DeterministicOnDemandFst< StdArc >
typedef StdArc ::StateId	StateId
typedef StdArc ::Weight	Weight
typedef StdArc ::Label	Label

Public Member Functions
	InverseContextFst (Label subsequential_symbol, const std::vector< int32 > &phones, const std::vector< int32 > &disambig_syms, int32 context_width, int32 central_position)
	Constructor. More...
virtual StateId	Start ()
virtual Weight	Final (StateId s)
virtual bool	GetArc (StateId s, Label ilabel, Arc *arc)
	Note: ilabel must not be epsilon. More...
	~InverseContextFst ()
const std::vector< std::vector< int32 > > &	IlabelInfo () const
void	SwapIlabelInfo (std::vector< std::vector< int32 > > *vec)
Public Member Functions inherited from DeterministicOnDemandFst< StdArc >
virtual Weight	Final (StateId s)=0
virtual bool	GetArc (StateId s, Label ilabel, StdArc *oarc)=0
	Note: ilabel must not be epsilon. More...
virtual	~DeterministicOnDemandFst ()

Private Types
typedef unordered_map< std::vector< int32 >, StateId, kaldi::VectorHasher< int32 > >	VectorToStateMap
typedef unordered_map< std::vector< int32 >, Label, kaldi::VectorHasher< int32 > >	VectorToLabelMap

Private Member Functions
StateId	FindState (const std::vector< int32 > &seq)
	Returns the state-id corresponding to this vector of phones; creates the state it if necessary. More...
Label	FindLabel (const std::vector< int32 > &label_info)
	Finds the label index corresponding to this context-window of phones (likely of width context_width_). More...
bool	IsDisambigSymbol (Label lab)
bool	IsPhoneSymbol (Label lab)
void	CreateDisambigArc (StateId s, Label ilabel, Arc *arc)
	Create disambiguation-symbol self-loop arc; where 'ilabel' must correspond to a disambiguation symbol. More...
void	CreatePhoneOrEpsArc (StateId src, StateId dst, Label ilabel, const std::vector< int32 > &phone_seq, Arc *arc)
	Creates an arc, this function is to be called only when 'ilabel' corresponds to a phone. More...
void	ShiftSequenceLeft (Label label, std::vector< int32 > *phone_seq)
	If phone_seq is nonempty then this function it left by one and appends 'label' to it, otherwise it does nothing. More...
void	GetFullPhoneSequence (const std::vector< int32 > &seq, Label label, std::vector< int32 > *full_phone_sequence)
	This utility function does something equivalent to the following 3 steps: *full_phone_sequence = seq; full_phone_sequence->append(label) Replace any values equal to 'subsequential_symbol_' in full_phone_sequence with zero (this is to avoid having to keep track of the value of 'subsequential_symbol_' outside of this program). More...

Private Attributes
int32	context_width_
int32	central_position_
kaldi::ConstIntegerSet< Label >	phone_syms_
kaldi::ConstIntegerSet< Label >	disambig_syms_
Label	subsequential_symbol_
int32	pseudo_eps_symbol_
VectorToStateMap	state_map_
std::vector< std::vector< int32 > >	state_seqs_
VectorToLabelMap	ilabel_map_
std::vector< std::vector< int32 > >	ilabel_info_

Detailed Description

Definition at line 152 of file context-fst.h.

Member Typedef Documentation

Arc

typedef StdArc Arc

Definition at line 154 of file context-fst.h.

Label

typedef StdArc::Label Label

Definition at line 157 of file context-fst.h.

StateId

typedef StdArc::StateId StateId

Definition at line 155 of file context-fst.h.

VectorToLabelMap

typedef unordered_map<std::vector<int32>, Label, kaldi::VectorHasher<int32> > VectorToLabelMap

private

Definition at line 256 of file context-fst.h.

VectorToStateMap

typedef unordered_map<std::vector<int32>, StateId, kaldi::VectorHasher<int32> > VectorToStateMap

private

Definition at line 250 of file context-fst.h.

Weight

typedef StdArc::Weight Weight

Definition at line 156 of file context-fst.h.

Constructor & Destructor Documentation

InverseContextFst()

InverseContextFst	(	Label	subsequential_symbol,
		const std::vector< int32 > &	phones,
		const std::vector< int32 > &	disambig_syms,
		int32	context_width,
		int32	central_position
	)

Constructor.

Parameters

[in]	subsequential_symbol	The integer id of the 'subsequential symbol' (usually represented as '$') that terminates sequences on the output of C.fst (input of InverseContextFst). Search for "quential" in https://cs.nyu.edu/~mohri/pub/hbka.pdf. This may just be the first unused integer id. Must be nonzer.
[in]	phones	List of integer ids of phones, as you would see in phones.txt
[in]	disambig_syms	List of integer ids of disambiguation symbols, e.g. the ids of #0, #1, #2 in phones.txt
[in]	context_width	Size of context window, e.g. 3 for triphone.
[in]	central_position	Central position in context window (zero-based), e.g. 1 for triphone. See The ContextFst object for more details.

Definition at line 27 of file context-fst.cc.

References InverseContextFst::central_position_, InverseContextFst::context_width_, ConstIntegerSet< I >::count(), InverseContextFst::disambig_syms_, ConstIntegerSet< I >::empty(), InverseContextFst::FindLabel(), InverseContextFst::FindState(), rnnlm::i, KALDI_ASSERT, KALDI_WARN, InverseContextFst::phone_syms_, and InverseContextFst::pseudo_eps_symbol_.

                           :
    context_width_(context_width),
    central_position_(central_position),
    phone_syms_(phones),
    disambig_syms_(disambig_syms),
    subsequential_symbol_(subsequential_symbol) {

  {  // This block checks the inputs.
    KALDI_ASSERT(subsequential_symbol != 0
                 && disambig_syms_.count(subsequential_symbol) == 0
                 && phone_syms_.count(subsequential_symbol) == 0);
    if (phone_syms_.empty())
      KALDI_WARN << "Context FST created but there are no phone symbols: probably "
          "input FST was empty.";
    KALDI_ASSERT(phone_syms_.count(0) == 0 && disambig_syms_.count(0) == 0 &&
                 central_position_ >= 0 && central_position_ < context_width_);
    for (size_t i = 0; i < phones.size(); i++) {
      KALDI_ASSERT(disambig_syms_.count(phones[i]) == 0);
    }
  }

  // empty vector, will be the ilabel_info vector that corresponds to epsilon,
  // in case our FST needs to output epsilons.
  vector<int32> empty_vec;
  Label epsilon_label = FindLabel(empty_vec);

  // epsilon_vec is the phonetic context window we have at the very start of a
  // sequence, meaning "no real phones have been seen yet".
  vector<int32> epsilon_vec(context_width_ - 1, 0);
  StateId start_state = FindState(epsilon_vec);

  KALDI_ASSERT(epsilon_label == 0 && start_state == 0);

  if (context_width_ > central_position_ + 1 && !disambig_syms_.empty()) {
    // We add a symbol whose sequence representation is [ 0 ], and whose
    // symbol-id is 1.  This is treated as a disambiguation symbol, we call it
    // #-1 in printed form.  It is necessary to ensure that all determinizable
    // LG's will have determinizable CLG's.  The problem it fixes is quite
    // subtle-- it relates to reordering of disambiguation symbols (they appear
    // earlier in CLG than in LG, relative to phones), and the fact that if a
    // disambig symbol appears at the very start of a sequence in CLG, it's not
    // clear exatly where it appeared on the corresponding sequence at the input
    // of LG.
    vector<int32> pseudo_eps_vec;
    pseudo_eps_vec.push_back(0);
    pseudo_eps_symbol_= FindLabel(pseudo_eps_vec);
    KALDI_ASSERT(pseudo_eps_symbol_ == 1);
  } else {
    pseudo_eps_symbol_ = 0;  // use actual epsilon.
  }
}

~InverseContextFst()

~InverseContextFst ( )

inline

Definition at line 188 of file context-fst.h.

{ }

Member Function Documentation

CreateDisambigArc()

void CreateDisambigArc ( StateId  s, Label  ilabel, Arc *  arc  )

inlineprivate

Create disambiguation-symbol self-loop arc; where 'ilabel' must correspond to a disambiguation symbol.

Called from CreateArc().

Definition at line 184 of file context-fst.cc.

References InverseContextFst::FindLabel().

Referenced by InverseContextFst::GetArc(), and InverseContextFst::IsPhoneSymbol().

                                                                           {
  // Creates a self-loop arc corresponding to the disambiguation symbol.
  vector<int32> label_info;       // This will be a vector containing just [ -olabel ].
  label_info.push_back(-ilabel);  // olabel is a disambiguation symbol.  Use its negative
                                  // so we can more easily distinguish them from phones.
  Label olabel = FindLabel(label_info);
  arc->ilabel = ilabel;
  arc->olabel = olabel;
  arc->weight = Weight::One();
  arc->nextstate = s;  // self-loop.
}

CreatePhoneOrEpsArc()

void CreatePhoneOrEpsArc ( StateId  src, StateId  dst, Label  ilabel, const std::vector< int32 > &  phone_seq, Arc *  arc  )

inlineprivate

Creates an arc, this function is to be called only when 'ilabel' corresponds to a phone.

Called from CreateArc(). The olabel may end be epsilon, instead of a phone-in-context, if the system has right context and we are very near the beginning of the phone sequence.

Definition at line 196 of file context-fst.cc.

References InverseContextFst::central_position_, InverseContextFst::FindLabel(), KALDI_PARANOID_ASSERT, InverseContextFst::pseudo_eps_symbol_, and InverseContextFst::subsequential_symbol_.

Referenced by InverseContextFst::GetArc(), and InverseContextFst::IsPhoneSymbol().

                                                      {
  KALDI_PARANOID_ASSERT(phone_seq[central_position_] != subsequential_symbol_);

  arc->ilabel = ilabel;
  arc->weight = Weight::One();
  arc->nextstate = dest;
  if (phone_seq[central_position_] == 0) {
    // This can happen at the beginning of the graph.  In this case we don't
    // output a real phone, we createdt an epsilon arc (but sometimes we need to
    // use a special disambiguation symbol instead of epsilon).
    arc->olabel = pseudo_eps_symbol_;
  } else {
    // We have a phone in the central position.
    arc->olabel = FindLabel(phone_seq);
  }
}

Final()

InverseContextFst::Weight Final ( StateId  s )

virtual

Definition at line 109 of file context-fst.cc.

References InverseContextFst::central_position_, InverseContextFst::context_width_, KALDI_ASSERT, InverseContextFst::state_seqs_, and InverseContextFst::subsequential_symbol_.

Referenced by InverseContextFst::Start().

                                                          {
  KALDI_ASSERT(static_cast<size_t>(s) < state_seqs_.size());

  const vector<int32> &phone_context = state_seqs_[s];

  KALDI_ASSERT(phone_context.size() == context_width_ - 1);

  bool has_final_prob;

  if (central_position_ < context_width_ - 1) {
    has_final_prob = (phone_context[central_position_] == subsequential_symbol_);
    // if phone_context[central_position_] != subsequential_symbol_ then we have
    // pending phones-in-context that we still need to output, so we need to
    // consume more subsequential symbols before we can terminate.
  } else {
    has_final_prob = true;
  }
  return has_final_prob ? Weight::One() : Weight::Zero();
}

FindLabel()

StdArc::Label FindLabel ( const std::vector< int32 > &  label_info )

private

Finds the label index corresponding to this context-window of phones (likely of width context_width_).

Inserts it into the ilabel_info_/ilabel_map_ tables if necessary.

Definition at line 231 of file context-fst.cc.

References InverseContextFst::ilabel_info_, and InverseContextFst::ilabel_map_.

Referenced by InverseContextFst::CreateDisambigArc(), InverseContextFst::CreatePhoneOrEpsArc(), InverseContextFst::InverseContextFst(), and InverseContextFst::SwapIlabelInfo().

                                                                       {
  // Finds the ilabel corresponding to this vector (creates a new ilabel if
  // necessary).
  VectorToLabelMap::const_iterator iter = ilabel_map_.find(label_vec);
  if (iter == ilabel_map_.end()) {  // Not already in map.
    Label this_label = ilabel_info_.size();
    ilabel_info_.push_back(label_vec);
    ilabel_map_[label_vec] = this_label;
    return this_label;
  } else {
    return iter->second;
  }
}

FindState()

StdArc::StateId FindState ( const std::vector< int32 > &  seq )

private

Returns the state-id corresponding to this vector of phones; creates the state it if necessary.

Requires seq.size() == context_width_ - 1.

Definition at line 216 of file context-fst.cc.

References InverseContextFst::context_width_, KALDI_ASSERT, InverseContextFst::state_map_, and InverseContextFst::state_seqs_.

Referenced by InverseContextFst::GetArc(), InverseContextFst::InverseContextFst(), and InverseContextFst::SwapIlabelInfo().

                                                                   {
  // Finds state-id corresponding to this vector of phones.  Inserts it if
  // necessary.
  KALDI_ASSERT(static_cast<int32>(seq.size()) == context_width_ - 1);
  VectorToStateMap::const_iterator iter = state_map_.find(seq);
  if (iter == state_map_.end()) {  // Not already in map.
    StateId this_state_id = (StateId)state_seqs_.size();
    state_seqs_.push_back(seq);
    state_map_[seq] = this_state_id;
    return this_state_id;
  } else {
    return iter->second;
  }
}

GetArc()

bool GetArc ( StateId  s, Label  ilabel, Arc *  arc  )

virtual

Note: ilabel must not be epsilon.

Definition at line 129 of file context-fst.cc.

References InverseContextFst::central_position_, InverseContextFst::context_width_, InverseContextFst::CreateDisambigArc(), InverseContextFst::CreatePhoneOrEpsArc(), InverseContextFst::FindState(), InverseContextFst::GetFullPhoneSequence(), InverseContextFst::IsDisambigSymbol(), InverseContextFst::IsPhoneSymbol(), KALDI_ASSERT, KALDI_ERR, InverseContextFst::ShiftSequenceLeft(), InverseContextFst::state_seqs_, and InverseContextFst::subsequential_symbol_.

Referenced by InverseContextFst::Start().

                                                                {
  KALDI_ASSERT(ilabel != 0 && static_cast<size_t>(s) < state_seqs_.size() &&
               state_seqs_[s].size() == context_width_ - 1);

  if (IsDisambigSymbol(ilabel)) {
    // A disambiguation-symbol self-loop arc.
    CreateDisambigArc(s, ilabel, arc);
    return true;
  } else if (IsPhoneSymbol(ilabel)) {
    const vector<int32> &seq = state_seqs_[s];
    if (!seq.empty() && seq.back() == subsequential_symbol_) {
      return false;  // A real phone is not allowed to follow the subsequential
                     // symbol.
    }

    // next_seq will be 'seq' shifted left by 1, with 'ilabel' appended.
    vector<int32> next_seq(seq);
    ShiftSequenceLeft(ilabel, &next_seq);

    // full-seq will be the full context window of size context_width_.
    vector<int32> full_seq;
    GetFullPhoneSequence(seq, ilabel, &full_seq);

    StateId next_s = FindState(next_seq);

    CreatePhoneOrEpsArc(s, next_s, ilabel, full_seq, arc);
    return true;
  } else if (ilabel == subsequential_symbol_) {
    const vector<int32> &seq = state_seqs_[s];

    if (central_position_ + 1 == context_width_ ||
        seq[central_position_] == subsequential_symbol_) {
      // We already had "enough" subsequential symbols in a row and don't want to
      // accept any more, or we'd be making the subsequential symbol the central phone.
      return false;
    }

    // full-seq will be the full context window of size context_width_.
    vector<int32> full_seq;
    GetFullPhoneSequence(seq, ilabel, &full_seq);

    vector<int32> next_seq(seq);
    ShiftSequenceLeft(ilabel, &next_seq);
    StateId next_s = FindState(next_seq);

    CreatePhoneOrEpsArc(s, next_s, ilabel, full_seq, arc);
    return true;
  } else {
    KALDI_ERR << "ContextFst: CreateArc, invalid ilabel supplied [confusion "
              << "about phone list or disambig symbols?]: " << ilabel;
  }
  return false;  // won't get here.  suppress compiler error.
}

GetFullPhoneSequence()

void GetFullPhoneSequence ( const std::vector< int32 > &  seq, Label  label, std::vector< int32 > *  full_phone_sequence  )

inlineprivate

This utility function does something equivalent to the following 3 steps: *full_phone_sequence = seq; full_phone_sequence->append(label) Replace any values equal to 'subsequential_symbol_' in full_phone_sequence with zero (this is to avoid having to keep track of the value of 'subsequential_symbol_' outside of this program).

This function assumes that seq.size() == context_width_ - 1, and also that 'subsequential_symbol_' does not appear in positions 0 through central_position_ of 'seq'.

Definition at line 93 of file context-fst.cc.

References InverseContextFst::central_position_, InverseContextFst::context_width_, rnnlm::i, and InverseContextFst::subsequential_symbol_.

Referenced by InverseContextFst::GetArc(), and InverseContextFst::IsPhoneSymbol().

                                           {
  int32 context_width = context_width_;
  full_phone_sequence->reserve(context_width);
  full_phone_sequence->insert(full_phone_sequence->end(),
                              seq.begin(), seq.end());
  full_phone_sequence->push_back(label);
  for (int32 i = central_position_ + 1; i < context_width; i++) {
    if ((*full_phone_sequence)[i] == subsequential_symbol_) {
      (*full_phone_sequence)[i] = 0;
    }
  }
}

IlabelInfo()

const std::vector<std::vector<int32> >& IlabelInfo ( ) const

inline

Definition at line 194 of file context-fst.h.

References InverseContextFst::ilabel_info_.

Referenced by fst::TestContextFst().

                                                         {
    return ilabel_info_;
  }

IsDisambigSymbol()

bool IsDisambigSymbol ( Label  lab )

inlineprivate

Definition at line 213 of file context-fst.h.

References ConstIntegerSet< I >::count(), and InverseContextFst::disambig_syms_.

Referenced by InverseContextFst::GetArc().

{ return (disambig_syms_.count(lab) != 0); }

IsPhoneSymbol()

bool IsPhoneSymbol ( Label  lab )

inlineprivate

Definition at line 215 of file context-fst.h.

References ConstIntegerSet< I >::count(), InverseContextFst::CreateDisambigArc(), InverseContextFst::CreatePhoneOrEpsArc(), InverseContextFst::GetFullPhoneSequence(), InverseContextFst::phone_syms_, and InverseContextFst::ShiftSequenceLeft().

Referenced by InverseContextFst::GetArc().

{ return (phone_syms_.count(lab) != 0); }

ShiftSequenceLeft()

void ShiftSequenceLeft ( Label  label, std::vector< int32 > *  phone_seq  )

inlineprivate

If phone_seq is nonempty then this function it left by one and appends 'label' to it, otherwise it does nothing.

We expect (but do not check) that phone_seq->size() == context_width_ - 1.

Definition at line 85 of file context-fst.cc.

Referenced by InverseContextFst::GetArc(), and InverseContextFst::IsPhoneSymbol().

                                                                       {
  if (!phone_seq->empty()) {
    phone_seq->erase(phone_seq->begin());
    phone_seq->push_back(label);
  }
}

Start()

virtual StateId Start ( )

inlinevirtual

Implements DeterministicOnDemandFst< StdArc >.

Definition at line 181 of file context-fst.h.

References InverseContextFst::Final(), and InverseContextFst::GetArc().

{ return 0; }

SwapIlabelInfo()

void SwapIlabelInfo ( std::vector< std::vector< int32 > > *  vec )

inline

Definition at line 200 of file context-fst.h.

References InverseContextFst::FindLabel(), InverseContextFst::FindState(), and InverseContextFst::ilabel_info_.

Referenced by fst::ComposeContext().

{ ilabel_info_.swap(*vec); }

Member Data Documentation

central_position_

int32 central_position_

private

Definition at line 271 of file context-fst.h.

Referenced by InverseContextFst::CreatePhoneOrEpsArc(), InverseContextFst::Final(), InverseContextFst::GetArc(), InverseContextFst::GetFullPhoneSequence(), and InverseContextFst::InverseContextFst().

context_width_

int32 context_width_

private

Definition at line 262 of file context-fst.h.

Referenced by InverseContextFst::Final(), InverseContextFst::FindState(), InverseContextFst::GetArc(), InverseContextFst::GetFullPhoneSequence(), and InverseContextFst::InverseContextFst().

disambig_syms_

kaldi::ConstIntegerSet<Label> disambig_syms_

private

Definition at line 285 of file context-fst.h.

Referenced by InverseContextFst::InverseContextFst(), and InverseContextFst::IsDisambigSymbol().

ilabel_info_

std::vector<std::vector<int32> > ilabel_info_

private

Definition at line 333 of file context-fst.h.

Referenced by InverseContextFst::FindLabel(), InverseContextFst::IlabelInfo(), and InverseContextFst::SwapIlabelInfo().

ilabel_map_

VectorToLabelMap ilabel_map_

private

Definition at line 325 of file context-fst.h.

Referenced by InverseContextFst::FindLabel().

phone_syms_

kaldi::ConstIntegerSet<Label> phone_syms_

private

Definition at line 276 of file context-fst.h.

Referenced by InverseContextFst::InverseContextFst(), and InverseContextFst::IsPhoneSymbol().

pseudo_eps_symbol_

int32 pseudo_eps_symbol_

private

Definition at line 309 of file context-fst.h.

Referenced by InverseContextFst::CreatePhoneOrEpsArc(), and InverseContextFst::InverseContextFst().

state_map_

VectorToStateMap state_map_

private

Definition at line 314 of file context-fst.h.

Referenced by InverseContextFst::FindState().

state_seqs_

std::vector<std::vector<int32> > state_seqs_

private

Definition at line 318 of file context-fst.h.

Referenced by InverseContextFst::Final(), InverseContextFst::FindState(), and InverseContextFst::GetArc().

subsequential_symbol_

Label subsequential_symbol_

private

Definition at line 292 of file context-fst.h.

Referenced by InverseContextFst::CreatePhoneOrEpsArc(), InverseContextFst::Final(), InverseContextFst::GetArc(), and InverseContextFst::GetFullPhoneSequence().

---

The documentation for this class was generated from the following files:

• fstext/context-fst.h
• fstext/context-fst.cc