From 3d43832c4fbaab9c58dcc93cafd19ce857a79a4a Mon Sep 17 00:00:00 2001 From: dohe0342 Date: Tue, 21 Feb 2023 14:35:18 +0900 Subject: [PATCH] from local --- egs/tedlium2/ASR/.prepare.sh.swp | Bin 0 -> 4096 bytes egs/tedlium2/ASR/prepare_ted3.sh | 155 +++++++++++++------------------ 2 files changed, 64 insertions(+), 91 deletions(-) create mode 100644 egs/tedlium2/ASR/.prepare.sh.swp diff --git a/egs/tedlium2/ASR/.prepare.sh.swp b/egs/tedlium2/ASR/.prepare.sh.swp new file mode 100644 index 0000000000000000000000000000000000000000..1b5b08a238b312664e6e3d2c805f711814c6413c GIT binary patch literal 4096 zcmYc?2=nw+u+%eP00IF9hUwv7l4tQbvutu;ysim2^zKO|B z`T5xaMfrL;`N@en=z8j42I`mR7iAY0BqpcoXC|knCFbPlr=}O{m!ziTWR~U{={p7o z=@%5G79>J?|88#Bru4S~@RAS(oT8H|k#!C75dNl`&qD3mNyMm3Fwz-S1JhQMeD XjE2By2#kinXb6mkz-R~z%@6 data/fbank/tedlium_cuts_train-shuf.jsonl.gz + mv data/fbank/tedlium_cuts_train-shuf.jsonl.gz \ + data/fbank/tedlium_cuts_train.jsonl.gz + + touch data/fbank/.tedlium3.done fi fi @@ -115,28 +128,24 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then fi if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then - log "Stage 5: Prepare phone based lang" - lang_dir=data/lang_phone + log "Stage 5: Prepare BPE train data and set of words" + lang_dir=data/lang mkdir -p $lang_dir - if [ ! -f $lang_dir/train.text ]; then + if [ ! -f $lang_dir/train.txt ]; then + gunzip -c $dl_dir/tedlium3/LM/*.en.gz | sed 's: <\/s>::g' > $lang_dir/train_orig.txt + ./local/prepare_transcripts.py \ - --lang-dir $lang_dir \ - --manifests-dir data/manifests + --input-text-path $lang_dir/train_orig.txt \ + --output-text-path $lang_dir/train.txt fi - if [ ! -f $lang_dir/lexicon_words.txt ]; then - ./local/prepare_lexicon.py \ - --lang-dir $lang_dir \ - --manifests-dir data/manifests - fi + if [ ! -f $lang_dir/words.txt ]; then - (echo '!SIL SIL'; echo ' '; ) | - cat - $lang_dir/lexicon_words.txt | - sort | uniq > $lang_dir/lexicon.txt + awk '{print $1}' $dl_dir/tedlium3/TEDLIUM.152k.dic | + sed 's:([0-9])::g' | sort | uniq > $lang_dir/words_orig.txt - if [ ! -f $lang_dir/L_disambig.pt ]; then - ./local/prepare_lang.py --lang-dir $lang_dir + ./local/prepare_words.py --lang-dir $lang_dir fi fi @@ -148,92 +157,56 @@ if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then mkdir -p $lang_dir # We reuse words.txt from phone based lexicon # so that the two can share G.pt later. - cp data/lang_phone/words.txt $lang_dir - - if [ ! -f $lang_dir/transcript_words.txt ]; then - log "Generate data for BPE training" - cat data/lang_phone/train.text | - cut -d " " -f 2- > $lang_dir/transcript_words.txt - # remove the for transcript_words.txt - sed -i 's/ //g' $lang_dir/transcript_words.txt - sed -i 's/ //g' $lang_dir/transcript_words.txt - sed -i 's///g' $lang_dir/transcript_words.txt - fi + cp data/lang/words.txt $lang_dir ./local/train_bpe_model.py \ --lang-dir $lang_dir \ --vocab-size $vocab_size \ - --transcript $lang_dir/transcript_words.txt + --transcript data/lang/train.txt if [ ! -f $lang_dir/L_disambig.pt ]; then - ./local/prepare_lang_bpe.py --lang-dir $lang_dir + ./local/prepare_lang_bpe.py --lang-dir $lang_dir --oov "" fi done fi if [ $stage -le 7 ] && [ $stop_stage -ge 7 ]; then - log "Stage 7: Prepare bigram P" - - for vocab_size in ${vocab_sizes[@]}; do - lang_dir=data/lang_bpe_${vocab_size} - - if [ ! -f $lang_dir/transcript_tokens.txt ]; then - ./local/convert_transcript_words_to_tokens.py \ - --lexicon $lang_dir/lexicon.txt \ - --transcript $lang_dir/transcript_words.txt \ - --oov "" \ - > $lang_dir/transcript_tokens.txt - fi - - if [ ! -f $lang_dir/P.arpa ]; then - ./shared/make_kn_lm.py \ - -ngram-order 2 \ - -text $lang_dir/transcript_tokens.txt \ - -lm $lang_dir/P.arpa - fi - - if [ ! -f $lang_dir/P.fst.txt ]; then - python3 -m kaldilm \ - --read-symbol-table="$lang_dir/tokens.txt" \ - --disambig-symbol='#0' \ - --max-order=2 \ - $lang_dir/P.arpa > $lang_dir/P.fst.txt - fi - done -fi - -if [ $stage -le 8 ] && [ $stop_stage -ge 8 ]; then - log "Stage 8: Prepare G" + log "Stage 7: Prepare G" # We assume you have install kaldilm, if not, please install # it using: pip install kaldilm mkdir -p data/lm - if [ ! -f data/lm/G_3_gram.fst.txt ]; then + if [ ! -f data/lm/G_4_gram_small.fst.txt ]; then # It is used in building HLG python3 -m kaldilm \ - --read-symbol-table="data/lang_phone/words.txt" \ - --disambig-symbol='#0' \ - --max-order=3 \ - $dl_dir/lm/3-gram.pruned.1e-7.arpa > data/lm/G_3_gram.fst.txt - fi - - if [ ! -f data/lm/G_4_gram.fst.txt ]; then - # It is used for LM rescoring - python3 -m kaldilm \ - --read-symbol-table="data/lang_phone/words.txt" \ + --read-symbol-table="data/lang/words.txt" \ --disambig-symbol='#0' \ --max-order=4 \ - $dl_dir/lm/4-gram.arpa > data/lm/G_4_gram.fst.txt + --max-arpa-warnings=-1 \ + $dl_dir/lm/4gram_small.arpa > data/lm/G_4_gram_small.fst.txt + fi + + if [ ! -f data/lm/G_4_gram_big.fst.txt ]; then + # It is used for LM rescoring + python3 -m kaldilm \ + --read-symbol-table="data/lang/words.txt" \ + --disambig-symbol='#0' \ + --max-order=4 \ + --max-arpa-warnings=-1 \ + $dl_dir/lm/4gram_big.arpa > data/lm/G_4_gram_big.fst.txt fi fi -if [ $stage -le 9 ] && [ $stop_stage -ge 9 ]; then - log "Stage 9: Compile HLG" - ./local/compile_hlg.py --lang-dir data/lang_phone +if [ $stage -le 8 ] && [ $stop_stage -ge 8 ]; then + log "Stage 8: Compile HLG" for vocab_size in ${vocab_sizes[@]}; do lang_dir=data/lang_bpe_${vocab_size} - ./local/compile_hlg.py --lang-dir $lang_dir + + if [ ! -f $lang_dir/HLG.pt ]; then + ./local/compile_hlg.py \ + --lang-dir $lang_dir \ + --lm G_4_gram_small + fi done fi -