mirror of
https://github.com/k2-fsa/icefall.git
synced 2025-12-10 22:45:27 +00:00
88 lines
2.5 KiB
Bash
Executable File
88 lines
2.5 KiB
Bash
Executable File
#!/usr/bin/env bash
|
||
|
||
# Prepare script for MLS English ASR recipe in icefall
|
||
export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
|
||
set -eou pipefail
|
||
|
||
stage=-1
|
||
stop_stage=100
|
||
|
||
# Configuration for BPE tokenizer
|
||
vocab_sizes=(500)
|
||
|
||
# Directory where dataset will be downloaded
|
||
dl_dir=$PWD/download
|
||
|
||
. shared/parse_options.sh || exit 1
|
||
|
||
# All files generated by this script are saved in "data/".
|
||
mkdir -p data/manifests data/fbank data/audio data/lang
|
||
|
||
log() {
|
||
local fname=${BASH_SOURCE[1]##*/}
|
||
echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${LINENO}:${FUNCNAME[1]}) $*"
|
||
}
|
||
|
||
log "Starting MLS English data preparation"
|
||
|
||
# Stage 0: Download corpus
|
||
if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then
|
||
log "Stage 0: Download MLS English dataset"
|
||
if [ ! -d $dl_dir/mls_english ]; then
|
||
git clone https://huggingface.co/datasets/parler-tts/mls_eng \
|
||
$dl_dir/mls_english || {
|
||
log "Failed to download MLS English dataset"; exit 1; }
|
||
fi
|
||
fi
|
||
|
||
# Stage 1: Compute fbank & emit manifests
|
||
if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
|
||
log "Stage 1: Compute & validate MLS English fbank"
|
||
# we already did `mkdir -p data/manifests data/fbank data/audio` above
|
||
|
||
if [ ! -e data/fbank/.mls_eng-fbank.done ]; then
|
||
python local/compute_fbank_mls_english.py \
|
||
--manifest-dir data/manifests \
|
||
--audio-dir data/audio \
|
||
--dl-dir $dl_dir/mls_english \
|
||
--fbank-dir data/fbank
|
||
|
||
# Validate each split’s manifest
|
||
for split in train dev test; do
|
||
python local/validate_manifest.py \
|
||
--manifest data/manifests/mls_eng_cuts_${split}.jsonl.gz
|
||
done
|
||
|
||
touch data/fbank/.mls_eng-fbank.done
|
||
log "fbank + manifest generation complete."
|
||
else
|
||
log "Skipping: fbank already done (data/fbank/.mls_eng-fbank.done exists)."
|
||
fi
|
||
fi
|
||
|
||
|
||
# Stage 2: Prepare transcript for BPE
|
||
if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
|
||
log "Stage 2: Generate transcript for BPE"
|
||
if [ ! -f data/lang/transcript.txt ]; then
|
||
./local/utils/generate_transcript.py --lang-dir data/lang
|
||
fi
|
||
fi
|
||
|
||
# Stage 3: Train BPE models
|
||
if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
|
||
log "Stage 3: Train BPE models"
|
||
for vocab_size in "${vocab_sizes[@]}"; do
|
||
bpe_dir=data/lang_bpe_${vocab_size}
|
||
mkdir -p $bpe_dir
|
||
if [ ! -f $bpe_dir/bpe.model ]; then
|
||
./local/train_bpe_model.py \
|
||
--lang-dir $bpe_dir \
|
||
--vocab-size $vocab_size \
|
||
--transcript data/lang/transcript.txt
|
||
fi
|
||
done
|
||
fi
|
||
|
||
log "MLS English data preparation completed successfully"
|