process-pitch-feats.cc File Reference

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

Include dependency graph for process-pitch-feats.cc:

Go to the source code of this file.

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

Functions
void	ProcessPovFeatures (Matrix< BaseFloat > *mat)
void	TakeLogOfPitch (Matrix< BaseFloat > *mat)
void	SubtractMovingAverage (int32 normalization_window_size, Matrix< BaseFloat > *mat)
void	SetToMovingAverage (int32 average_window_size, Matrix< BaseFloat > *mat)
int	main (int argc, char *argv[])

Function Documentation

main()

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

Definition at line 105 of file process-pitch-feats.cc.

References SequentialTableReader< Holder >::Done(), ParseOptions::GetArg(), KALDI_ASSERT, KALDI_ERR, KALDI_LOG, KALDI_VLOG, KALDI_WARN, SequentialTableReader< Holder >::Key(), SequentialTableReader< Holder >::Next(), ParseOptions::NumArgs(), MatrixBase< Real >::NumCols(), MatrixBase< Real >::NumRows(), TableWriter< Holder >::Open(), ParseOptions::PrintUsage(), kaldi::ProcessPovFeatures(), ParseOptions::Read(), ParseOptions::Register(), kaldi::SetToMovingAverage(), kaldi::SubtractMovingAverage(), kaldi::TakeLogOfPitch(), 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) into something suitable\n"
        "to put into a speech recognizer.  First use interpolate-feats\n"
        "Usage:  process-pitch-feats [options...] <feats-rspecifier> <feats-wspecifier>\n";

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

    int32 normalization_window_size = 151; // should be odd number
    int32 average_window_size = 5;
    
    // Register the options
    po.Register("normalization-window-size",
                &normalization_window_size, "Size of window used for "
                "moving window nomalization (must be odd).");
    po.Register("average-window-size",
                &average_window_size,
                "Size of moving average window (must be odd).");
    
    // parse options (+filling the registered variables)
    po.Read(argc, argv);

    if (po.NumArgs() != 2) {
      po.PrintUsage();
      exit(1);
    }
    
    KALDI_ASSERT(average_window_size > 0 && average_window_size % 2 == 1 &&
                 "--average-window-size option must be an odd positive number.");
    KALDI_ASSERT(normalization_window_size > 0 && normalization_window_size % 2 == 1 &&
                 "--normalization-window-size option must be an odd positive number.");
    
    std::string input_rspecifier = po.GetArg(1);
    std::string output_wspecifier = po.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;

    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;
      }
      
      ProcessPovFeatures(&features);
      TakeLogOfPitch(&features);
      SubtractMovingAverage(normalization_window_size, &features);
      SetToMovingAverage(average_window_size, &features);
      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;
    }
    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;
  }
}