Blame view
src/online/online-faster-decoder.cc
9.55 KB
8dcb6dfcb first commit |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 |
// online/online-faster-decoder.cc // Copyright 2012 Cisco Systems (author: Matthias Paulik) // Modifications to the original contribution by Cisco Systems made by: // Vassil Panayotov // See ../../COPYING for clarification regarding multiple authors // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // You may obtain a copy of the License at // // http://www.apache.org/licenses/LICENSE-2.0 // // THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY // KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED // WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE, // MERCHANTABLITY OR NON-INFRINGEMENT. // See the Apache 2 License for the specific language governing permissions and // limitations under the License. #include "base/timer.h" #include "online-faster-decoder.h" #include "fstext/fstext-utils.h" #include "hmm/hmm-utils.h" namespace kaldi { void OnlineFasterDecoder::ResetDecoder(bool full) { ClearToks(toks_.Clear()); StateId start_state = fst_.Start(); KALDI_ASSERT(start_state != fst::kNoStateId); Arc dummy_arc(0, 0, Weight::One(), start_state); Token *dummy_token = new Token(dummy_arc, NULL); toks_.Insert(start_state, dummy_token); prev_immortal_tok_ = immortal_tok_ = dummy_token; utt_frames_ = 0; if (full) frame_ = 0; } void OnlineFasterDecoder::MakeLattice(const Token *start, const Token *end, fst::MutableFst<LatticeArc> *out_fst) const { out_fst->DeleteStates(); if (start == NULL) return; bool is_final = false; double this_cost = start->cost_ + fst_.Final(start->arc_.nextstate).Value(); if (this_cost != std::numeric_limits<double>::infinity()) is_final = true; std::vector<LatticeArc> arcs_reverse; // arcs in reverse order. for (const Token *tok = start; tok != end; tok = tok->prev_) { BaseFloat tot_cost = tok->cost_ - (tok->prev_ ? tok->prev_->cost_ : 0.0), graph_cost = tok->arc_.weight.Value(), ac_cost = tot_cost - graph_cost; LatticeArc l_arc(tok->arc_.ilabel, tok->arc_.olabel, LatticeWeight(graph_cost, ac_cost), tok->arc_.nextstate); arcs_reverse.push_back(l_arc); } if(arcs_reverse.back().nextstate == fst_.Start()) { arcs_reverse.pop_back(); // that was a "fake" token... gives no info. } StateId cur_state = out_fst->AddState(); out_fst->SetStart(cur_state); for (ssize_t i = static_cast<ssize_t>(arcs_reverse.size())-1; i >= 0; i--) { LatticeArc arc = arcs_reverse[i]; arc.nextstate = out_fst->AddState(); out_fst->AddArc(cur_state, arc); cur_state = arc.nextstate; } if (is_final) { Weight final_weight = fst_.Final(start->arc_.nextstate); out_fst->SetFinal(cur_state, LatticeWeight(final_weight.Value(), 0.0)); } else { out_fst->SetFinal(cur_state, LatticeWeight::One()); } RemoveEpsLocal(out_fst); } void OnlineFasterDecoder::UpdateImmortalToken() { unordered_set<Token*> emitting; for (const Elem *e = toks_.GetList(); e != NULL; e = e->tail) { Token* tok = e->val; while (tok->arc_.ilabel == 0) //deal with non-emitting ones ... tok = tok->prev_; if (tok != NULL) emitting.insert(tok); } Token* the_one = NULL; while (1) { if (emitting.size() == 1) { the_one = *(emitting.begin()); break; } if (emitting.size() == 0) break; unordered_set<Token*> prev_emitting; unordered_set<Token*>::iterator it; for (it = emitting.begin(); it != emitting.end(); ++it) { Token* tok = *it; Token* prev_token = tok->prev_; while ((prev_token != NULL) && (prev_token->arc_.ilabel == 0)) prev_token = prev_token->prev_; //deal with non-emitting ones if (prev_token == NULL) continue; prev_emitting.insert(prev_token); } // for emitting = prev_emitting; } // while if (the_one != NULL) { prev_immortal_tok_ = immortal_tok_; immortal_tok_ = the_one; return; } } bool OnlineFasterDecoder::PartialTraceback(fst::MutableFst<LatticeArc> *out_fst) { UpdateImmortalToken(); if(immortal_tok_ == prev_immortal_tok_) return false; //no partial traceback at that point of time MakeLattice(immortal_tok_, prev_immortal_tok_, out_fst); return true; } void OnlineFasterDecoder::FinishTraceBack(fst::MutableFst<LatticeArc> *out_fst) { Token *best_tok = NULL; bool is_final = ReachedFinal(); if (!is_final) { for (const Elem *e = toks_.GetList(); e != NULL; e = e->tail) if (best_tok == NULL || *best_tok < *(e->val) ) best_tok = e->val; } else { double best_cost = std::numeric_limits<double>::infinity(); for (const Elem *e = toks_.GetList(); e != NULL; e = e->tail) { double this_cost = e->val->cost_ + fst_.Final(e->key).Value(); if (this_cost != std::numeric_limits<double>::infinity() && this_cost < best_cost) { best_cost = this_cost; best_tok = e->val; } } } MakeLattice(best_tok, immortal_tok_, out_fst); } void OnlineFasterDecoder::TracebackNFrames(int32 nframes, fst::MutableFst<LatticeArc> *out_fst) { Token *best_tok = NULL; for (const Elem *e = toks_.GetList(); e != NULL; e = e->tail) if (best_tok == NULL || *best_tok < *(e->val) ) best_tok = e->val; if (best_tok == NULL) { out_fst->DeleteStates(); return; } bool is_final = false; double this_cost = best_tok->cost_ + fst_.Final(best_tok->arc_.nextstate).Value(); if (this_cost != std::numeric_limits<double>::infinity()) is_final = true; std::vector<LatticeArc> arcs_reverse; // arcs in reverse order. for (Token *tok = best_tok; (tok != NULL) && (nframes > 0); tok = tok->prev_) { if (tok->arc_.ilabel != 0) // count only the non-epsilon arcs --nframes; BaseFloat tot_cost = tok->cost_ - (tok->prev_ ? tok->prev_->cost_ : 0.0); BaseFloat graph_cost = tok->arc_.weight.Value(); BaseFloat ac_cost = tot_cost - graph_cost; LatticeArc larc(tok->arc_.ilabel, tok->arc_.olabel, LatticeWeight(graph_cost, ac_cost), tok->arc_.nextstate); arcs_reverse.push_back(larc); } if(arcs_reverse.back().nextstate == fst_.Start()) arcs_reverse.pop_back(); // that was a "fake" token... gives no info. StateId cur_state = out_fst->AddState(); out_fst->SetStart(cur_state); for (ssize_t i = static_cast<ssize_t>(arcs_reverse.size())-1; i >= 0; i--) { LatticeArc arc = arcs_reverse[i]; arc.nextstate = out_fst->AddState(); out_fst->AddArc(cur_state, arc); cur_state = arc.nextstate; } if (is_final) { Weight final_weight = fst_.Final(best_tok->arc_.nextstate); out_fst->SetFinal(cur_state, LatticeWeight(final_weight.Value(), 0.0)); } else { out_fst->SetFinal(cur_state, LatticeWeight::One()); } RemoveEpsLocal(out_fst); } bool OnlineFasterDecoder::EndOfUtterance() { fst::VectorFst<LatticeArc> trace; int32 sil_frm = opts_.inter_utt_sil / (1 + utt_frames_ / opts_.max_utt_len_); TracebackNFrames(sil_frm, &trace); std::vector<int32> isymbols; fst::GetLinearSymbolSequence(trace, &isymbols, static_cast<std::vector<int32>* >(0), static_cast<LatticeArc::Weight*>(0)); std::vector<std::vector<int32> > split; SplitToPhones(trans_model_, isymbols, &split); for (size_t i = 0; i < split.size(); i++) { int32 tid = split[i][0]; int32 phone = trans_model_.TransitionIdToPhone(tid); if (silence_set_.count(phone) == 0) return false; } return true; } OnlineFasterDecoder::DecodeState OnlineFasterDecoder::Decode(DecodableInterface *decodable) { if (state_ == kEndFeats || state_ == kEndUtt) // new utterance ResetDecoder(state_ == kEndFeats); ProcessNonemitting(std::numeric_limits<float>::max()); int32 batch_frame = 0; Timer timer; double64 tstart = timer.Elapsed(), tstart_batch = tstart; BaseFloat factor = -1; for (; !decodable->IsLastFrame(frame_ - 1) && batch_frame < opts_.batch_size; ++frame_, ++utt_frames_, ++batch_frame) { if (batch_frame != 0 && (batch_frame % opts_.update_interval) == 0) { // adjust the beam if needed BaseFloat tend = timer.Elapsed(); BaseFloat elapsed = (tend - tstart) * 1000; // warning: hardcoded 10ms frames assumption! factor = elapsed / (opts_.rt_max * opts_.update_interval * 10); BaseFloat min_factor = (opts_.rt_min / opts_.rt_max); if (factor > 1 || factor < min_factor) { BaseFloat update_factor = (factor > 1)? -std::min(opts_.beam_update * factor, opts_.max_beam_update): std::min(opts_.beam_update / factor, opts_.max_beam_update); effective_beam_ += effective_beam_ * update_factor; effective_beam_ = std::min(effective_beam_, max_beam_); } tstart = tend; } if (batch_frame != 0 && (frame_ % 200) == 0) // one log message at every 2 seconds assuming 10ms frames KALDI_VLOG(3) << "Beam: " << effective_beam_ << "; Speed: " << ((timer.Elapsed() - tstart_batch) * 1000) / (batch_frame*10) << " xRT"; BaseFloat weight_cutoff = ProcessEmitting(decodable); ProcessNonemitting(weight_cutoff); } if (batch_frame == opts_.batch_size && !decodable->IsLastFrame(frame_ - 1)) { if (EndOfUtterance()) state_ = kEndUtt; else state_ = kEndBatch; } else { state_ = kEndFeats; } return state_; } } // namespace kaldi |