#!/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=-1 stop_stage=100 dl_dir=$PWD/download mkdir -p $dl_dir . shared/parse_options.sh || exit 1 # All files generated by this script are saved in "data". # You can safely remove "data" and rerun this script to regenerate it. mkdir -p data log() { # This function is from espnet local fname=${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: build monotonic_align lib (used by ./matcha)" for recipe in matcha; do if [ ! -d $recipe/monotonic_align/build ]; then cd $recipe/monotonic_align python3 setup.py build_ext --inplace cd ../../ else log "monotonic_align lib for $recipe already built" fi done fi if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then log "Stage 0: Download data" # The directory $dl_dir/BANSYP contains the following 3 directories # ls -lh $dl_dir/BZNSYP/ # total 0 # drwxr-xr-x 10002 kuangfangjun root 0 Jan 4 2019 PhoneLabeling # drwxr-xr-x 3 kuangfangjun root 0 Jan 31 2019 ProsodyLabeling # drwxr-xr-x 10003 kuangfangjun root 0 Aug 26 17:45 Wave # If you have trouble accessing huggingface.co, please use # # cd $dl_dir # wget https://huggingface.co/openspeech/BZNSYP/resolve/main/BZNSYP.tar.bz2 # tar xf BZNSYP.tar.bz2 # cd .. # If you have pre-downloaded it to /path/to/BZNSYP, you can create a symlink # # ln -sfv /path/to/BZNSYP $dl_dir/BZNSYP # if [ ! -d $dl_dir/BZNSYP/Wave ]; then lhotse download baker-zh $dl_dir fi fi if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then log "Stage 1: Prepare baker-zh manifest" # We assume that you have downloaded the baker corpus # to $dl_dir/BZNSYP mkdir -p data/manifests if [ ! -e data/manifests/.baker-zh.done ]; then lhotse prepare baker-zh $dl_dir/BZNSYP data/manifests touch data/manifests/.baker-zh.done fi fi if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then log "Stage 2: Generate tokens.txt" if [ ! -e data/tokens.txt ]; then python3 ./local/generate_tokens.py --tokens data/tokens.txt fi fi if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then log "Stage 3: Generate raw cutset" if [ ! -e data/manifests/baker_zh_cuts_raw.jsonl.gz ]; then lhotse cut simple \ -r ./data/manifests/baker_zh_recordings_all.jsonl.gz \ -s ./data/manifests/baker_zh_supervisions_all.jsonl.gz \ ./data/manifests/baker_zh_cuts_raw.jsonl.gz fi fi if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then log "Stage 4: Convert text to tokens" if [ ! -e data/manifests/baker_zh_cuts.jsonl.gz ]; then python3 ./local/convert_text_to_tokens.py \ --in-file ./data/manifests/baker_zh_cuts_raw.jsonl.gz \ --out-file ./data/manifests/baker_zh_cuts.jsonl.gz fi fi if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then log "Stage 5: Generate fbank (used by ./matcha)" mkdir -p data/fbank if [ ! -e data/fbank/.baker-zh.done ]; then ./local/compute_fbank_baker_zh.py touch data/fbank/.baker-zh.done fi if [ ! -e data/fbank/.baker-zh-validated.done ]; then log "Validating data/fbank for baker-zh (used by ./matcha)" python3 ./local/validate_manifest.py \ data/fbank/baker_zh_cuts.jsonl.gz touch data/fbank/.baker-zh-validated.done fi fi if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then log "Stage 6: Split the baker-zh cuts into train, valid and test sets (used by ./matcha)" if [ ! -e data/fbank/.baker_zh_split.done ]; then lhotse subset --last 600 \ data/fbank/baker_zh_cuts.jsonl.gz \ data/fbank/baker_zh_cuts_validtest.jsonl.gz lhotse subset --first 100 \ data/fbank/baker_zh_cuts_validtest.jsonl.gz \ data/fbank/baker_zh_cuts_valid.jsonl.gz lhotse subset --last 500 \ data/fbank/baker_zh_cuts_validtest.jsonl.gz \ data/fbank/baker_zh_cuts_test.jsonl.gz rm data/fbank/baker_zh_cuts_validtest.jsonl.gz n=$(( $(gunzip -c data/fbank/baker_zh_cuts.jsonl.gz | wc -l) - 600 )) lhotse subset --first $n \ data/fbank/baker_zh_cuts.jsonl.gz \ data/fbank/baker_zh_cuts_train.jsonl.gz touch data/fbank/.baker_zh_split.done fi fi if [ $stage -le 7 ] && [ $stop_stage -ge 7 ]; then log "Stage 6: Compute fbank mean and std (used by ./matcha)" if [ ! -f ./data/fbank/cmvn.json ]; then ./local/compute_fbank_statistics.py ./data/fbank/baker_zh_cuts_train.jsonl.gz ./data/fbank/cmvn.json fi fi