lattice-to-mpe-post.cc File Reference

#include "base/kaldi-common.h"
#include "util/common-utils.h"
#include "fstext/fstext-lib.h"
#include "lat/kaldi-lattice.h"
#include "lat/lattice-functions.h"
#include "gmm/am-diag-gmm.h"
#include "hmm/transition-model.h"

Include dependency graph for lattice-to-mpe-post.cc:

Go to the source code of this file.

Functions
int	main (int argc, char *argv[])

Function Documentation

main()

int main	(	int	argc,
		char *	argv[]
	)

Definition at line 29 of file lattice-to-mpe-post.cc.

References SequentialTableReader< Holder >::Done(), SequentialTableReader< Holder >::FreeCurrent(), ParseOptions::GetArg(), RandomAccessTableReader< Holder >::HasKey(), KALDI_ERR, KALDI_LOG, KALDI_VLOG, KALDI_WARN, SequentialTableReader< Holder >::Key(), kaldi::LatticeForwardBackwardMpeVariants(), fst::LatticeScale(), SequentialTableReader< Holder >::Next(), fst::NumArcs(), ParseOptions::NumArgs(), ParseOptions::PrintUsage(), ParseOptions::Read(), TransitionModel::Read(), ParseOptions::Register(), fst::ScaleLattice(), kaldi::SortAndUniq(), kaldi::SplitStringToIntegers(), Input::Stream(), RandomAccessTableReader< Holder >::Value(), SequentialTableReader< Holder >::Value(), and TableWriter< Holder >::Write().

                                 {
  try {
    typedef kaldi::int32 int32;
    using fst::SymbolTable;
    using fst::VectorFst;
    using fst::StdArc;
    using namespace kaldi;
  
    const char *usage =
        "Do forward-backward and collect frame level MPE posteriors over\n" 
        "lattices, which can be fed into gmm-acc-stats2 to do MPE traning.\n"
        "Caution: this is not really MPE, this is MPFE (minimum phone frame\n"
        "error).  The posteriors may be positive or negative.\n"
        "Usage: lattice-to-mpe-post [options] <model> <num-posteriors-rspecifier>\n"
        " <lats-rspecifier> <posteriors-wspecifier> \n"
        "e.g.: lattice-to-mpe-post --acoustic-scale=0.1 1.mdl ark:num.post\n"
        " ark:1.lats ark:1.post\n";

    kaldi::BaseFloat acoustic_scale = 1.0, lm_scale = 1.0;
    bool one_silence_class = false;
    std::string silence_phones_str;
    kaldi::ParseOptions po(usage);
    po.Register("acoustic-scale", &acoustic_scale,
                "Scaling factor for acoustic likelihoods");
    po.Register("lm-scale", &lm_scale,
                "Scaling factor for \"graph costs\" (including LM costs)");
    po.Register("one-silence-class", &one_silence_class, "If true, newer "
                 "behavior which will tend to reduce insertions.");
    po.Register("silence-phones", &silence_phones_str,
                "Colon-separated list of integer id's of silence phones, e.g. 46:47");
    po.Read(argc, argv);

    if (po.NumArgs() != 4) {
      po.PrintUsage();
      exit(1);
    }
 
    std::vector<int32> silence_phones;
    if (!kaldi::SplitStringToIntegers(silence_phones_str, ":", false, &silence_phones))
      KALDI_ERR << "Invalid silence-phones string " << silence_phones_str;
    kaldi::SortAndUniq(&silence_phones);
    if (silence_phones.empty())
      KALDI_WARN << "No silence phones specified, make sure this is what you intended.";

    if (acoustic_scale == 0.0)
      KALDI_ERR << "Do not use a zero acoustic scale (cannot be inverted)";

    std::string model_filename = po.GetArg(1), 
        alignments_rspecifier = po.GetArg(2),  
        lats_rspecifier = po.GetArg(3),
        posteriors_wspecifier = po.GetArg(4);

    SequentialLatticeReader lattice_reader(lats_rspecifier);
    PosteriorWriter posterior_writer(posteriors_wspecifier);
    if (alignments_rspecifier.find("ali-to-post") != std::string::npos) {
      KALDI_WARN << "Warning, this program has been changed to read alignments "
                 << "not posteriors.  Remove ali-to-post from your scripts.";
    }
    RandomAccessInt32VectorReader alignments_reader(alignments_rspecifier);
    
    TransitionModel trans_model;
    {
      bool binary;
      Input ki(model_filename, &binary);
      trans_model.Read(ki.Stream(), binary);
    }

    int32 num_done = 0, num_err = 0;
    double total_lat_frame_acc = 0.0, lat_frame_acc;
    double total_time = 0, lat_time;


    for (; !lattice_reader.Done(); lattice_reader.Next()) {
      std::string key = lattice_reader.Key();
      kaldi::Lattice lat = lattice_reader.Value();
      lattice_reader.FreeCurrent();
      if (acoustic_scale != 1.0 || lm_scale != 1.0)
        fst::ScaleLattice(fst::LatticeScale(lm_scale, acoustic_scale), &lat);
      
      kaldi::uint64 props = lat.Properties(fst::kFstProperties, false);
      if (!(props & fst::kTopSorted)) {
        if (fst::TopSort(&lat) == false)
          KALDI_ERR << "Cycles detected in lattice.";
      }
      
      if (!alignments_reader.HasKey(key)) {
        KALDI_WARN << "No alignment for utterance " << key;
        num_err++;
      } else {
        const std::vector<int32> &alignment = alignments_reader.Value(key);
        Posterior post;
        lat_frame_acc = LatticeForwardBackwardMpeVariants(
            trans_model, silence_phones, lat, alignment,
            "mpfe", one_silence_class, &post);
        total_lat_frame_acc += lat_frame_acc;
        lat_time = post.size();
        total_time += lat_time;
        KALDI_VLOG(2) << "Processed lattice for utterance: " << key << "; found "
                      << lat.NumStates() << " states and " << fst::NumArcs(lat)
                      << " arcs. Average frame accuracies = " << (lat_frame_acc/lat_time)
                      << " over " << lat_time << " frames.";
        posterior_writer.Write(key, post);
        num_done++; 
      }
    }

    KALDI_LOG << "Overall average frame-accuracy is "
              << (total_lat_frame_acc/total_time) << " over " << total_time
              << " frames.";
    KALDI_LOG << "Done " << num_done << " lattices.";
    return (num_done != 0 ? 0 : 1);
  } catch(const std::exception &e) {
    std::cerr << e.what();
    return -1;
  }
}