diff --git a/egs/tedlium3/ASR/.prepare.sh.swp b/egs/tedlium3/ASR/.prepare.sh.swp index c55fde1b5..a88e7099c 100644 Binary files a/egs/tedlium3/ASR/.prepare.sh.swp and b/egs/tedlium3/ASR/.prepare.sh.swp differ diff --git a/egs/tedlium3/ASR/prepare.sh b/egs/tedlium3/ASR/prepare.sh index 69a435356..9aa4d347a 100755 --- a/egs/tedlium3/ASR/prepare.sh +++ b/egs/tedlium3/ASR/prepare.sh @@ -213,10 +213,16 @@ fi if [ $stage -le 9 ] && [ $stop_stage -ge 9 ]; then log "Stage 9: Split cuts by speaker id" + gzip -d data/fbank/tedlium_cuts_test.json.gz + + i=0 for spk in $dl_dir/tedlium3/legacy/test/sph/*; do spk_id=${spk#*sph\/} spk_id=${spk_id%.sph} echo $spk_id + cat data/fbank/tedlium_cuts_test.jsonl | grep speaker\":\ \"$spk_id\" > tedlium_cuts_test_$i.jsonl + gzip tedlium_cuts_test_$i.jsonl + i=`expr $i+1` done #cat data/fbank/tedlium_cuts_test.jsonl.gz | grep