subsample-feats.cc File Reference

#include <sstream>
#include <algorithm>
#include <iterator>
#include <utility>
#include "base/kaldi-common.h"
#include "util/common-utils.h"
#include "matrix/kaldi-matrix.h"

Include dependency graph for subsample-feats.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 30 of file subsample-feats.cc.

References VectorBase< Real >::CopyFromVec(), SequentialTableReader< Holder >::Done(), ParseOptions::GetArg(), rnnlm::i, KALDI_ASSERT, KALDI_LOG, KALDI_WARN, SequentialTableReader< Holder >::Key(), rnnlm::n, SequentialTableReader< Holder >::Next(), ParseOptions::NumArgs(), MatrixBase< Real >::NumRows(), ParseOptions::PrintUsage(), ParseOptions::Read(), ParseOptions::Register(), MatrixBase< Real >::Row(), SequentialTableReader< Holder >::Value(), and TableWriter< Holder >::Write().

                                 {
  try {
    using namespace kaldi;
    using namespace std;

    const char *usage =
        "Sub-samples features by taking every n'th frame.\n"
        "With negative values of n, will repeat each frame n times\n"
        "(e.g. --n=-2 will repeat each frame twice)\n"
        "\n"
        "Usage: subsample-feats [options] <in-rspecifier> <out-wspecifier>\n"
        "  e.g. subsample-feats --n=2 ark:- ark:-\n";

    ParseOptions po(usage);

    int32 n = 1, offset = 0;

    po.Register("n", &n, "Take every n'th feature, for this value of n"
                "(with negative value, repeats each feature n times)");
    po.Register("offset", &offset, "Start with the feature with this offset, "
                "then take every n'th feature.");

    KALDI_ASSERT(n != 0);
    if (n < 0)
      KALDI_ASSERT(offset == 0 &&
                   "--offset option cannot be used with negative n.");

    po.Read(argc, argv);

    if (po.NumArgs() != 2) {
      po.PrintUsage();
      exit(1);
    }

    string rspecifier = po.GetArg(1);
    string wspecifier = po.GetArg(2);

    SequentialBaseFloatMatrixReader feat_reader(rspecifier);
    BaseFloatMatrixWriter feat_writer(wspecifier);

    int32 num_done = 0, num_err = 0;
    int64 frames_in = 0, frames_out = 0;

    // process all keys
    for (; !feat_reader.Done(); feat_reader.Next()) {
      std::string utt = feat_reader.Key();
      const Matrix<BaseFloat> feats(feat_reader.Value());


      if (n > 0) {
        // This code could, of course, be much more efficient; I'm just
        // keeping it simple.
        int32 num_indexes = 0;
        for (int32 k = offset; k < feats.NumRows(); k += n)
          num_indexes++; // k is the index.

        frames_in += feats.NumRows();
        frames_out += num_indexes;

        if (num_indexes == 0) {
          KALDI_WARN << "For utterance " << utt << ", output would have no rows, "
                     << "producing no output.";
          num_err++;
          continue;
        }
        Matrix<BaseFloat> output(num_indexes, feats.NumCols());
        int32 i = 0;
        for (int32 k = offset; k < feats.NumRows(); k += n, i++) {
          SubVector<BaseFloat> src(feats, k), dest(output, i);
          dest.CopyFromVec(src);
        }
        KALDI_ASSERT(i == num_indexes);
        feat_writer.Write(utt, output);
        num_done++;
      } else {
        int32 repeat = -n;
        Matrix<BaseFloat> output(feats.NumRows() * repeat, feats.NumCols());
        for (int32 i = 0; i < output.NumRows(); i++)
          output.Row(i).CopyFromVec(feats.Row(i / repeat));
        frames_in += feats.NumRows();
        frames_out += feats.NumRows() * repeat;
        feat_writer.Write(utt, output);
        num_done++;
      }
    }
    KALDI_LOG << "Processed " << num_done << " feature matrices; " << num_err
              << " with errors.";
    KALDI_LOG << "Processed " << frames_in << " input frames and "
              << frames_out << " output frames.";
    return (num_done != 0 ? 0 : 1);
  } catch(const std::exception &e) {
    std::cerr << e.what();
    return -1;
  }
}