Kaldiに関する処理を日本語のドキュメントでまとめてみた（特徴量抽出編）3 ref: http://qiita.com/GushiSnow/items/e099baf9d1c2e72cb3d1

file0.txt

for x in train test; do                                                     
  steps/make_mfcc.sh --cmd "$train_cmd" --nj $njobs \                        
  data/$x exp/make_mfcc/$x $mfccdir || exit 1;                                
  steps/compute_cmvn_stats.sh data/$x exp/make_mfcc/$x $mfccdir || exit 1;   
done

file1.txt

check_sorted_and_uniq $data/utt2spk

file10.txt

#セグメントなし
else
  echo "$0: [info]: no segments file exists: assuming wav.scp indexed by utterance."
  split_scps=""
  for n in $(seq $nj); do
    split_scps="$split_scps $logdir/wav_${name}.$n.scp"
  done

  utils/split_scp.pl $scp $split_scps || exit 1;


  # add ,p to the input rspecifier so that we can just skip over
  # utterances that have bad wave data.

  $cmd JOB=1:$nj $logdir/make_mfcc_${name}.JOB.log \
    compute-mfcc-feats  $vtln_opts --verbose=2 --config=$mfcc_config \
     scp,p:$logdir/wav_${name}.JOB.scp ark:- \| \
      copy-feats --compress=$compress ark:- \
      ark,scp:$mfccdir/raw_mfcc_$name.JOB.ark,$mfccdir/raw_mfcc_$name.JOB.scp \
      || exit 1;
fi

file2.txt

check_sorted_and_uniq $data/spk2utt  

file3.txt

num_utts=`cat $tmpdir/utts | wc -l`
if [ -f $data/text ]; then
  check_sorted_and_uniq $data/text
  text_len=`cat $data/text | wc -l`
  illegal_sym_list="<s> </s> #0"
  for x in $illegal_sym_list; do
    if grep -w "$x" $data/text > /dev/null; then
      echo "$0: Error: in $data, text contains illegal symbol $x"
      exit 1;
    fi
  done
  awk '{print $1}' < $data/text > $tmpdir/utts.txt
  if ! cmp -s $tmpdir/utts{,.txt}; then
    echo "$0: Error: in $data, utterance lists extracted from utt2spk and text"
    echo "$0: differ, partial diff is:"
    partial_diff $tmpdir/utts{,.txt}
    exit 1;
  fi
fi

file4.txt

if [ -f $data/wav.scp ]; then
  check_sorted_and_uniq $data/wav.scp

  if [ -f $data/segments ]; then

    check_sorted_and_uniq $data/segments
    # We have a segments file -> interpret wav file as "recording-ids" not utter
ance-ids.
    ! cat $data/segments | \
      awk '{if (NF != 4 || ($4 <= $3 && $4 != -1)) { print "Bad line in segments
 file", $0; exit(1); }}' && \
      echo "$0: badly formatted segments file" && exit 1;

    segments_len=`cat $data/segments | wc -l`
    if [ -f $data/text ]; then
      ! cmp -s $tmpdir/utts <(awk '{print $1}' <$data/text) && \
        echo "$0: Utterance list differs between $data/text and $data/segments "
 && \
        echo "$0: Lengths are $segments_len vs $num_utts";
    fi

    cat $data/segments | awk '{print $2}' | sort | uniq > $tmpdir/recordings
    awk '{print $1}' $data/wav.scp > $tmpdir/recordings.wav
    if ! cmp -s $tmpdir/recordings{,.wav}; then
      echo "$0: Error: in $data, recording-ids extracted from segments and wav.scp"
      echo "$0: differ, partial diff is:"
      partial_diff $tmpdir/recordings{,.wav}
      exit 1;
    fi

file5.txt

    if [ -f $data/reco2file_and_channel ]; then
      # this file is needed only for ctm scoring; it's indexed by recording-id.
      check_sorted_and_uniq $data/reco2file_and_channel
      ! cat $data/reco2file_and_channel | \
        awk '{if (NF != 3 || ($3 != "A" && $3 != "B" )) { 
                if ( NF == 3 && $3 == "1" ) {
                  warning_issued = 1;
                } else {
                  print "Bad line ", $0; exit 1; 
                }
              }
            } 
            END {
              if (warning_issued == 1) {
                print "The channel should be marked as A or B, not 1! You should change it ASAP! "
              }
            }' && echo "$0: badly formatted reco2file_and_channel file" && exit 1;
      cat $data/reco2file_and_channel | awk '{print $1}' > $tmpdir/recordings.r2fc
      if ! cmp -s $tmpdir/recordings{,.r2fc}; then
        echo "$0: Error: in $data, recording-ids extracted from segments and reco2file_and_channel"
        echo "$0: differ, partial diff is:"
        partial_diff $tmpdir/recordings{,.r2fc}
        exit 1;
      fi
    fi
  else
    # No segments file -> assume wav.scp indexed by utterance.
    cat $data/wav.scp | awk '{print $1}' > $tmpdir/utts.wav
    if ! cmp -s $tmpdir/utts{,.wav}; then
      echo "$0: Error: in $data, utterance lists extracted from utt2spk and wav.scp"
      echo "$0: differ, partial diff is:"
      partial_diff $tmpdir/utts{,.wav}
      exit 1;
    fi

    if [ -f $data/reco2file_and_channel ]; then
      # this file is needed only for ctm scoring; it's indexed by recording-id.
      check_sorted_and_uniq $data/reco2file_and_channel
      ! cat $data/reco2file_and_channel | \
        awk '{if (NF != 3 || ($3 != "A" && $3 != "B")) { print "Bad line ", $0; exit 1; }}' && \
        echo "$0: badly formatted reco2file_and_channel file" && exit 1;
      cat $data/reco2file_and_channel | awk '{print $1}' > $tmpdir/utts.r2fc
      if ! cmp -s $tmpdir/utts{,.r2fc}; then
        echo "$0: Error: in $data, utterance-ids extracted from segments and reco2file_and_channel"
        echo "$0: differ, partial diff is:"
        partial_diff $tmpdir/utts{,.r2fc}
        exit 1;
      fi
    fi
  fi
fi

file6.txt

-f $data/feats.scp #特徴量
-f $data/cmvn.scp #cmvn
-f $data/spk2gender #話者の性別
-f $data/spk2warp #不明
-f $data/utt2warp #不明
-f $data/$f         #その他データ

file7.txt

for n in $(seq $nj); 
do                   
    # the next command does nothing unless $mfccdir/storage/ exists, see      
    # utils/create_data_link.pl for more info.                                
    utils/create_data_link.pl $mfccdir/raw_mfcc_$name.$n.ark                  
 done      

file8.txt

for n in $(seq $nj); do
  # the next command does nothing unless $mfccdir/storage/ exists, see
  # utils/create_data_link.pl for more info.
  utils/create_data_link.pl $mfccdir/raw_mfcc_$name.$n.ark
done

file9.txt

#セグメントあり
if [ -f $data/segments ]; then
  echo "$0 [info]: segments file exists: using that."

  split_segments=""
  for n in $(seq $nj); do
    split_segments="$split_segments $logdir/segments.$n"
  done

#scpファイルを分割 
  utils/split_scp.pl $data/segments $split_segments || exit 1;
  rm $logdir/.error 2>/dev/null

  $cmd JOB=1:$nj $logdir/make_mfcc_${name}.JOB.log \
    extract-segments scp,p:$scp $logdir/segments.JOB ark:- \| \
    compute-mfcc-feats $vtln_opts --verbose=2 --config=$mfcc_config ark:- ark:- \| \
    copy-feats --compress=$compress ark:- \
      ark,scp:$mfccdir/raw_mfcc_$name.JOB.ark,$mfccdir/raw_mfcc_$name.JOB.scp \
     || exit 1;