icefall/egs/libritts/ASR/prepare.sh

#!/usr/bin/env bash

# fix segmentation fault reported in https://github.com/k2-fsa/icefall/issues/674
export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python

set -eou pipefail

stage=0
stop_stage=100
sampling_rate=16000
nj=32
perturb_speed=true
vocab_sizes=(
  # 5000
  # 2000
  # 1000
  500
)

dl_dir=$PWD/download

. shared/parse_options.sh || exit 1

# All files generated by this script are saved in "data".
# You can safely remove "data" and rerun this script to regenerate it.
mkdir -p data

log() {
  # This function is from espnet
  local fname=${BASH_SOURCE[1]##*/}
  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
}

log "dl_dir: $dl_dir"

if [ $stage -le -1 ] && [ $stop_stage -ge -1 ]; then
  log "Stage -1: Download LM" # we directly use the librispeech lm here
  mkdir -p $dl_dir/lm
  if [ ! -e $dl_dir/lm/.done ]; then
    ./local/download_lm.py --out-dir=$dl_dir/lm
    touch $dl_dir/lm/.done
  fi
fi

if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then
  log "Stage 0: Download data"

  # If you have pre-downloaded it to /path/to/LibriTTS,
  # you can create a symlink
  #
  #   ln -sfv /path/to/LibriTTS $dl_dir/LibriTTS
  #
  if [ ! -d $dl_dir/LibriTTS ]; then
    lhotse download libritts $dl_dir
  fi

  # If you have pre-downloaded it to /path/to/musan,
  # you can create a symlink
  #
  #   ln -sfv /path/to/musan $dl_dir/musan
  #
  if [ ! -d $dl_dir/musan ]; then
    lhotse download musan $dl_dir
  fi
fi

if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
  log "Stage 1: Prepare LibriTTS manifest"
  # We assume that you have downloaded the LibriTTS corpus
  # to $dl_dir/LibriTTS
  mkdir -p data/manifests
  if [ ! -e data/manifests/.libritts.done ]; then
    lhotse prepare libritts --num-jobs 32 $dl_dir/LibriTTS data/manifests
    touch data/manifests/.libritts.done
  fi
fi

if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
  log "Stage 2: Prepare musan manifest"
  # We assume that you have downloaded the musan corpus
  # to data/musan
  if [ ! -f data/manifests/.musan_manifests.done ]; then
    log "It may take 6 minutes"
    mkdir -p data/manifests
    lhotse prepare musan $dl_dir/musan data/manifests
    touch data/manifests/.musan_manifests.done
  fi
fi

if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
  log "Stage 3: Compute Fbank for LibriTTS"
  mkdir -p data/fbank
  if [ ! -e data/fbank/.libritts.done ]; then
    ./local/compute_fbank_libritts.py \
      --sampling-rate $sampling_rate \
      --perturb-speed $perturb_speed
    touch data/fbank/.libritts.done
  fi

  # Here we shuffle and combine the train-clean-100, train-clean-360 and 
  # train-other-500 together to form the training set.
  if [ ! -f data/fbank/libritts_cuts_train-all-shuf.jsonl.gz ]; then
    cat <(gunzip -c data/fbank/libritts_cuts_train-clean-100.jsonl.gz) \
      <(gunzip -c data/fbank/libritts_cuts_train-clean-360.jsonl.gz) \
      <(gunzip -c data/fbank/libritts_cuts_train-other-500.jsonl.gz) | \
      shuf | gzip -c > data/fbank/libritts_cuts_train-all-shuf.jsonl.gz
  fi

  if [ ! -e data/fbank/.libritts-validated.done ]; then
    log "Validating data/fbank for LibriTTS"
    ./local/validate_manifest.py \
      data/fbank/libritts_cuts_train-all-shuf.jsonl.gz
    touch data/fbank/.libritts-validated.done
  fi
fi

if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
  log "Stage 4: Compute fbank for musan"
  if [ ! -f data/fbank/.msuan.done ]; then
    mkdir -p data/fbank
    ./local/compute_fbank_musan.py
    touch data/fbank/.msuan.done
  fi
fi

if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
  log "Stage 5: Train BPE model for normalized text"

  if [ ! -f data/text ]; then
    gunzip -c data/manifests/libritts_supervisions_train-clean-100.jsonl.gz \
      | jq ".text" | sed 's/"//g' \
      | ./local/norm_text.py > data/text

    gunzip -c data/manifests/libritts_supervisions_train-clean-360.jsonl.gz \
      | jq ".text" | sed 's/"//g' \
      | ./local/norm_text.py >> data/text

    gunzip -c data/manifests/libritts_supervisions_train-other-500.jsonl.gz \
      | jq ".text" | sed 's/"//g' \
      | ./local/norm_text.py >> data/text
  fi

  for vocab_size in ${vocab_sizes[@]}; do
    lang_dir=data/lang_bpe_${vocab_size}
    mkdir -p $lang_dir

    cp data/text $lang_dir/text

    if [ ! -f $lang_dir/bpe.model ]; then
      ./local/train_bpe_model.py \
        --lang-dir $lang_dir \
        --vocab-size $vocab_size \
        --transcript $lang_dir/text
    fi
  done
fi

if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
  log "Stage 6: Prepare phone based lang"
  lang_dir=data/lang_phone
  mkdir -p $lang_dir

  if [ ! -f $dl_dir/lm/librispeech-lexicon.txt ]; then
    log "No lexicon file in $dl_dir/lm, please run :"
    log "prepare.sh --stage -1 --stop-stage -1"
    exit -1
  fi

  if [ ! -f $lang_dir/lexicon.txt ]; then
    (echo '!SIL SIL'; echo '<SPOKEN_NOISE> SPN'; echo '<UNK> SPN'; ) |
      cat - $dl_dir/lm/librispeech-lexicon.txt |
      sort | uniq > $lang_dir/lexicon.txt
  fi

  if [ ! -f $lang_dir/L_disambig.pt ]; then
    ./local/prepare_lang.py --lang-dir $lang_dir
  fi

  if [ ! -f $lang_dir/L.fst ]; then
    log "Converting L.pt to L.fst"
    ./shared/convert-k2-to-openfst.py \
      --olabels aux_labels \
      $lang_dir/L.pt \
      $lang_dir/L.fst
  fi

  if [ ! -f $lang_dir/L_disambig.fst ]; then
    log "Converting L_disambig.pt to L_disambig.fst"
    ./shared/convert-k2-to-openfst.py \
      --olabels aux_labels \
      $lang_dir/L_disambig.pt \
      $lang_dir/L_disambig.fst
  fi
fi