mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-26 18:24:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix index error

Browse Source
This commit is contained in:
Yuekai Zhang 2024-02-20 10:20:00 +08:00
parent 6fd14d202b
commit be001a896c
3 changed files with 25 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
egs/multi_zh-hans/ASR/local/compute_fbank_kespeech_splits.py
View File

@ -137,8 +137,8 @@ def compute_fbank_kespeech_splits(args):
set_audio_duration_mismatch_tolerance(0.01) # 10ms tolerance
set_caching_enabled(False)
for i in range(start, stop):
idx = f"{i + 1}".zfill(num_digits)
logging.info(f"Processing {idx}/{num_splits}")
idx = f"{i}".zfill(num_digits)
logging.info(f"Processing {i+1}/{num_splits}")
cuts_path = output_dir / f"kespeech-asr_cuts_{subset}.{idx}.jsonl.gz"
if cuts_path.is_file():

27
egs/multi_zh-hans/ASR/prepare.sh
View File

@ -5,8 +5,8 @@ export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
set -eou pipefail
stage=120
stop_stage=120
stage=121
stop_stage=121
num_splits=100
dl_dir=$PWD/download
@ -274,7 +274,7 @@ if [ $stage -le 12 ] && [ $stop_stage -ge 12 ]; then
touch data/fbank/.kespeech_preprocess_complete
fi
if [ -f data/fbank/.kespeech.train_phase1.split.${num_splits}.done ]; then
if [ ! -f data/fbank/.kespeech.train_phase1.split.${num_splits}.done ]; then
log "Spliting KeSpeech train_phase1"
lhotse split ${num_splits} \
data/fbank/kespeech/kespeech-asr_cuts_train_phase1_raw.jsonl.gz \
@ -282,7 +282,7 @@ if [ $stage -le 12 ] && [ $stop_stage -ge 12 ]; then
touch data/fbank/.kespeech.train_phase1.split.${num_splits}.done
fi
if [ -f data/fbank/.kespeech.train_phase2.split.${num_splits}.done ]; then
if [ ! -f data/fbank/.kespeech.train_phase2.split.${num_splits}.done ]; then
log "Spliting KeSpeech train_phase2"
lhotse split ${num_splits} \
data/fbank/kespeech/kespeech-asr_cuts_train_phase2_raw.jsonl.gz \
@ -327,7 +327,7 @@ if [ $stage -le 120 ] && [ $stop_stage -ge 120 ]; then
touch data/fbank/.kespeech_preprocess_complete
fi
if [ -f data/fbank/.kespeech.train_phase1.split.${num_splits}.done ]; then
if [ ! -f data/fbank/.kespeech.train_phase1.split.${num_splits}.done ]; then
log "Spliting KeSpeech train_phase1"
lhotse split ${num_splits} \
data/fbank/kespeech/kespeech-asr_cuts_train_phase1_raw.jsonl.gz \
@ -335,7 +335,7 @@ if [ $stage -le 120 ] && [ $stop_stage -ge 120 ]; then
touch data/fbank/.kespeech.train_phase1.split.${num_splits}.done
fi
if [ -f data/fbank/.kespeech.train_phase2.split.${num_splits}.done ]; then
if [ ! -f data/fbank/.kespeech.train_phase2.split.${num_splits}.done ]; then
log "Spliting KeSpeech train_phase2"
lhotse split ${num_splits} \
data/fbank/kespeech/kespeech-asr_cuts_train_phase2_raw.jsonl.gz \
@ -356,6 +356,21 @@ if [ $stage -le 120 ] && [ $stop_stage -ge 120 ]; then
fi
fi
if [ $stage -le 121 ] && [ $stop_stage -ge 121 ]; then
log "Stage 121: tmp"
log "Compute KeSpeech fbank for train_phase1"
./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --stop 1 --training-subset train_phase1 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
log "Compute KeSpeech fbank for train_phase2"
./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase2 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
log "Compute KeSpeech fbank for test/dev"
./local/compute_fbank_kespeech_dev_test.py --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
touch data/fbank/.kespeech.done
fi
fi
if [ $stage -le 13 ] && [ $stop_stage -ge 13 ]; then
log "Stage 13: BPE model training (note that we use transcripts of wenetspeech only for BPE training)"

4
egs/wenetspeech/ASR/local/compute_fbank_wenetspeech_splits.py
View File

@ -150,8 +150,8 @@ def compute_fbank_wenetspeech_splits(args):
set_caching_enabled(False)
#with get_executor() as ex: # Initialize the executor only once.
for i in range(start, stop):
idx = f"{i + 1}".zfill(num_digits)
logging.info(f"Processing {idx}/{num_splits}")
idx = f"{i}".zfill(num_digits)
logging.info(f"Processing {i+1}/{num_splits}")
cuts_path = output_dir / f"cuts_{subset}.{idx}.jsonl.gz"
if cuts_path.is_file():