interpolate-pitch.cc File Reference

#include "base/kaldi-common.h"
#include "util/common-utils.h"

Include dependency graph for interpolate-pitch.cc:

Go to the source code of this file.

Classes
struct	PitchInterpolatorOptions
struct	PitchInterpolatorStats
class	PitchInterpolator

Namespaces
	kaldi
	This code computes Goodness of Pronunciation (GOP) and extracts phone-level pronunciation feature for mispronunciations detection tasks, the reference:

Functions
void	LinearlyInterpolatePitch (Matrix< BaseFloat > *mat)
int	main (int argc, char *argv[])

Function Documentation

main()

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

Definition at line 292 of file interpolate-pitch.cc.

References SequentialTableReader< Holder >::Done(), ParseOptions::GetArg(), KALDI_ERR, KALDI_LOG, KALDI_VLOG, KALDI_WARN, SequentialTableReader< Holder >::Key(), kaldi::LinearlyInterpolatePitch(), SequentialTableReader< Holder >::Next(), ParseOptions::NumArgs(), MatrixBase< Real >::NumCols(), MatrixBase< Real >::NumRows(), TableWriter< Holder >::Open(), PitchInterpolatorStats::Print(), ParseOptions::PrintUsage(), ParseOptions::Read(), PitchInterpolatorOptions::Register(), ParseOptions::Register(), SequentialTableReader< Holder >::Value(), and TableWriter< Holder >::Write().

                                 {
  try {
    using namespace kaldi;
    const char *usage =
        "This is a rather special-purpose program which processes 2-dimensional\n"
        "features consisting of (prob-of-voicing, pitch).  By default we do model-based\n"
        "pitch smoothing and interpolation (see code), or if --linear-interpolation=true,\n"
        "just linear interpolation across gaps where pitch == 0 (not predicted).\n"
        "Usage:  interpolate-pitch [options...] <feats-rspecifier> <feats-wspecifier>\n";

    
    // construct all the global objects
    ParseOptions opts(usage);

    bool linear_interpolation = false;
    PitchInterpolatorOptions interpolate_opts;
    
    opts.Register("linear-interpolation",
                &linear_interpolation, "If true, just do simple linear "
                "interpolation across gaps (else, model-based)");
    interpolate_opts.Register(&opts);
    
    // parse options (+filling the registered variables)
    opts.Read(argc, argv);

    if (opts.NumArgs() != 2) {
      opts.PrintUsage();
      exit(1);
    }
    
    std::string input_rspecifier = opts.GetArg(1);
    std::string output_wspecifier = opts.GetArg(2);

    SequentialBaseFloatMatrixReader reader(input_rspecifier);
    BaseFloatMatrixWriter kaldi_writer;  // typedef to TableWriter<something>.

    if (!kaldi_writer.Open(output_wspecifier))
       KALDI_ERR << "Could not initialize output with wspecifier "
                << output_wspecifier;

    int32 num_done = 0, num_err = 0;
    PitchInterpolatorStats stats;
    
    for (; !reader.Done(); reader.Next()) {
      std::string utt = reader.Key();   
      Matrix<BaseFloat> features = reader.Value();
      int num_frames = features.NumRows();

      if (num_frames == 0 && features.NumCols() != 2) {
        KALDI_WARN << "Feature file has bad size "
                   << features.NumRows() << " by " << features.NumCols();
        num_err++;
        continue;
      }
      
      if (linear_interpolation) LinearlyInterpolatePitch(&features);
      else {
        // work happens in constructor of this class.
        PitchInterpolator pi(interpolate_opts, &features, &stats);
      }
      kaldi_writer.Write(utt, features);
      num_done++;
        
      if (num_done % 10 == 0)
        KALDI_LOG << "Processed " << num_done << " utterances";
      KALDI_VLOG(2) << "Processed features for key " << utt;
    }
    if (!linear_interpolation) stats.Print();
    KALDI_LOG << "Done " << num_done << " utterances, " << num_err
              << " with errors.";
    return (num_done != 0 ? 0 : 1);
  } catch(const std::exception &e) {
    std::cerr << e.what();
    return -1;
  }
}