mirror of
https://github.com/modelscope/FunASR
synced 2025-09-15 14:48:36 +08:00
update repo
This commit is contained in:
嘉渊
parent
17a08d878b
commit
32f1051c3d
@ -34,14 +34,14 @@ fi
|
||||
# validate utt-key list, IC0803W0380 is a bad utterance
|
||||
awk '{print $1}' $corpus/wav.scp | grep -v 'IC0803W0380' > $tmp/wav_utt.list
|
||||
awk '{print $1}' $corpus/trans.txt > $tmp/trans_utt.list
|
||||
tools/filter_scp.pl -f 1 $tmp/wav_utt.list $tmp/trans_utt.list > $tmp/utt.list
|
||||
utils/filter_scp.pl -f 1 $tmp/wav_utt.list $tmp/trans_utt.list > $tmp/utt.list
|
||||
|
||||
# wav.scp
|
||||
awk -F'\t' -v path_prefix=$corpus '{printf("%s\t%s/%s\n",$1,path_prefix,$2)}' $corpus/wav.scp > $tmp/tmp_wav.scp
|
||||
tools/filter_scp.pl -f 1 $tmp/utt.list $tmp/tmp_wav.scp | sort -k 1 | uniq > $tmp/wav.scp
|
||||
utils/filter_scp.pl -f 1 $tmp/utt.list $tmp/tmp_wav.scp | sort -k 1 | uniq > $tmp/wav.scp
|
||||
|
||||
# text
|
||||
tools/filter_scp.pl -f 1 $tmp/utt.list $corpus/trans.txt | sort -k 1 | uniq > $tmp/text
|
||||
utils/filter_scp.pl -f 1 $tmp/utt.list $corpus/trans.txt | sort -k 1 | uniq > $tmp/text
|
||||
|
||||
# copy prepared resources from tmp_dir to target dir
|
||||
mkdir -p $dir
|
||||
|
||||
@ -34,14 +34,14 @@ fi
|
||||
# validate utt-key list, IC0803W0380 is a bad utterance
|
||||
awk '{print $1}' $corpus/wav.scp | grep -v 'IC0803W0380' > $tmp/wav_utt.list
|
||||
awk '{print $1}' $corpus/trans.txt > $tmp/trans_utt.list
|
||||
tools/filter_scp.pl -f 1 $tmp/wav_utt.list $tmp/trans_utt.list > $tmp/utt.list
|
||||
utils/filter_scp.pl -f 1 $tmp/wav_utt.list $tmp/trans_utt.list > $tmp/utt.list
|
||||
|
||||
# wav.scp
|
||||
awk -F'\t' -v path_prefix=$corpus '{printf("%s\t%s/%s\n",$1,path_prefix,$2)}' $corpus/wav.scp > $tmp/tmp_wav.scp
|
||||
tools/filter_scp.pl -f 1 $tmp/utt.list $tmp/tmp_wav.scp | sort -k 1 | uniq > $tmp/wav.scp
|
||||
utils/filter_scp.pl -f 1 $tmp/utt.list $tmp/tmp_wav.scp | sort -k 1 | uniq > $tmp/wav.scp
|
||||
|
||||
# text
|
||||
tools/filter_scp.pl -f 1 $tmp/utt.list $corpus/trans.txt | sort -k 1 | uniq > $tmp/text
|
||||
utils/filter_scp.pl -f 1 $tmp/utt.list $corpus/trans.txt | sort -k 1 | uniq > $tmp/text
|
||||
|
||||
# copy prepared resources from tmp_dir to target dir
|
||||
mkdir -p $dir
|
||||
|
||||
@ -9,31 +9,28 @@ count=1
|
||||
gpu_inference=true # Whether to perform gpu decoding, set false for cpu decoding
|
||||
# for gpu decoding, inference_nj=ngpu*njob; for cpu decoding, inference_nj=njob
|
||||
njob=5
|
||||
train_cmd=tools/run.pl
|
||||
train_cmd=utils/run.pl
|
||||
infer_cmd=utils/run.pl
|
||||
|
||||
# general configuration
|
||||
feats_dir="../DATA" #feature output dictionary
|
||||
exp_dir="."
|
||||
lang=zh
|
||||
dumpdir=dump/fbank
|
||||
feats_type=fbank
|
||||
token_type=char
|
||||
type=sound
|
||||
scp=wav.scp
|
||||
speed_perturb="0.9 1.0 1.1"
|
||||
dataset_type=large
|
||||
scp=feats.scp
|
||||
type=kaldi_ark
|
||||
stage=0
|
||||
stage=3
|
||||
stop_stage=4
|
||||
|
||||
# feature configuration
|
||||
feats_dim=80
|
||||
sample_frequency=16000
|
||||
nj=100
|
||||
speed_perturb="0.9,1.0,1.1"
|
||||
nj=64
|
||||
|
||||
# data
|
||||
tr_dir=
|
||||
dev_tst_dir=
|
||||
tr_dir=/nfs/wangjiaming.wjm/asr_data/aishell2/AISHELL-2/iOS/data
|
||||
dev_tst_dir=/nfs/wangjiaming.wjm/asr_data/aishell2/AISHELL-DEV-TEST-SET
|
||||
|
||||
# exp tag
|
||||
tag="exp1"
|
||||
@ -51,13 +48,13 @@ valid_set=dev_ios
|
||||
test_sets="dev_ios test_ios"
|
||||
|
||||
asr_config=conf/train_asr_transformer.yaml
|
||||
model_dir="baseline_$(basename "${asr_config}" .yaml)_${feats_type}_${lang}_${token_type}_${tag}"
|
||||
model_dir="baseline_$(basename "${asr_config}" .yaml)_${lang}_${token_type}_${tag}"
|
||||
|
||||
inference_config=conf/decode_asr_transformer.yaml
|
||||
inference_asr_model=valid.acc.ave_10best.pb
|
||||
|
||||
# you can set gpu num for decoding here
|
||||
gpuid_list=$CUDA_VISIBLE_DEVICES # set gpus for decoding, e.g., gpuid_list=2,3, the same as training stage by default
|
||||
gpuid_list=$CUDA_VISIBLE_DEVICES # set gpus for decoding, the same as training stage by default
|
||||
ngpu=$(echo $gpuid_list | awk -F "," '{print NF}')
|
||||
|
||||
if ${gpu_inference}; then
|
||||
@ -73,61 +70,24 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
|
||||
# For training set
|
||||
local/prepare_data.sh ${tr_dir} ${feats_dir}/data/local/train ${feats_dir}/data/train || exit 1;
|
||||
# # For dev and test set
|
||||
for x in Android iOS Mic; do
|
||||
for x in iOS; do
|
||||
local/prepare_data.sh ${dev_tst_dir}/${x}/dev ${feats_dir}/data/local/dev_${x,,} ${feats_dir}/data/dev_${x,,} || exit 1;
|
||||
local/prepare_data.sh ${dev_tst_dir}/${x}/test ${feats_dir}/data/local/test_${x,,} ${feats_dir}/data/test_${x,,} || exit 1;
|
||||
done
|
||||
done
|
||||
# Normalize text to capital letters
|
||||
for x in train dev_ios test_ios; do
|
||||
mv ${feats_dir}/data/${x}/text ${feats_dir}/data/${x}/text.org
|
||||
paste -d " " <(cut -f 1 ${feats_dir}/data/${x}/text.org) <(cut -f 2- ${feats_dir}/data/${x}/text.org \
|
||||
| tr 'A-Z' 'a-z' | tr -d " ") \
|
||||
> ${feats_dir}/data/${x}/text
|
||||
tools/text2token.py -n 1 -s 1 ${feats_dir}/data/${x}/text > ${feats_dir}/data/${x}/text.org
|
||||
utils/text2token.py -n 1 -s 1 ${feats_dir}/data/${x}/text > ${feats_dir}/data/${x}/text.org
|
||||
mv ${feats_dir}/data/${x}/text.org ${feats_dir}/data/${x}/text
|
||||
done
|
||||
fi
|
||||
|
||||
feat_train_dir=${feats_dir}/${dumpdir}/${train_set}; mkdir -p ${feat_train_dir}
|
||||
feat_dev_dir=${feats_dir}/${dumpdir}/${valid_set}; mkdir -p ${feat_dev_dir}
|
||||
if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
|
||||
echo "stage 1: Feature Generation"
|
||||
# compute fbank features
|
||||
fbankdir=${feats_dir}/fbank
|
||||
steps/compute_fbank.sh --cmd "$train_cmd" --nj $nj --speed_perturb ${speed_perturb} \
|
||||
${feats_dir}/data/train ${exp_dir}/exp/make_fbank/train ${fbankdir}/train
|
||||
tools/fix_data_feat.sh ${fbankdir}/train
|
||||
for x in android ios mic; do
|
||||
steps/compute_fbank.sh --cmd "$train_cmd" --nj $nj \
|
||||
${feats_dir}/data/dev_${x} ${exp_dir}/exp/make_fbank/dev_${x} ${fbankdir}/dev_${x}
|
||||
tools/fix_data_feat.sh ${fbankdir}/dev_${x}
|
||||
steps/compute_fbank.sh --cmd "$train_cmd" --nj $nj \
|
||||
${feats_dir}/data/test_${x} ${exp_dir}/exp/make_fbank/test_${x} ${fbankdir}/test_${x}
|
||||
tools/fix_data_feat.sh ${fbankdir}/test_${x}
|
||||
done
|
||||
|
||||
# compute global cmvn
|
||||
steps/compute_cmvn.sh --cmd "$train_cmd" --nj $nj \
|
||||
${fbankdir}/train ${exp_dir}/exp/make_fbank/train
|
||||
|
||||
# apply cmvn
|
||||
steps/apply_cmvn.sh --cmd "$train_cmd" --nj $nj \
|
||||
${fbankdir}/${train_set} ${fbankdir}/train/cmvn.json ${exp_dir}/exp/make_fbank/${train_set} ${feat_train_dir}
|
||||
steps/apply_cmvn.sh --cmd "$train_cmd" --nj $nj \
|
||||
${fbankdir}/${valid_set} ${fbankdir}/train/cmvn.json ${exp_dir}/exp/make_fbank/${valid_set} ${feat_dev_dir}
|
||||
for x in android ios mic; do
|
||||
steps/apply_cmvn.sh --cmd "$train_cmd" --nj $nj \
|
||||
${fbankdir}/test_${x} ${fbankdir}/train/cmvn.json ${exp_dir}/exp/make_fbank/test_${x} ${feats_dir}/${dumpdir}/test_${x}
|
||||
done
|
||||
|
||||
cp ${fbankdir}/${train_set}/text ${fbankdir}/${train_set}/speech_shape ${fbankdir}/${train_set}/text_shape ${feat_train_dir}
|
||||
tools/fix_data_feat.sh ${feat_train_dir}
|
||||
cp ${fbankdir}/${valid_set}/text ${fbankdir}/${valid_set}/speech_shape ${fbankdir}/${valid_set}/text_shape ${feat_dev_dir}
|
||||
tools/fix_data_feat.sh ${feat_dev_dir}
|
||||
for x in android ios mic; do
|
||||
cp ${fbankdir}/test_${x}/text ${fbankdir}/test_${x}/speech_shape ${fbankdir}/test_${x}/text_shape ${feats_dir}/${dumpdir}/test_${x}
|
||||
tools/fix_data_feat.sh ${feats_dir}/${dumpdir}/test_${x}
|
||||
done
|
||||
echo "stage 1: Feature and CMVN Generation"
|
||||
utils/compute_cmvn.sh --cmd "$train_cmd" --nj $nj --feats_dim ${feats_dim} ${feats_dir}/data/${train_set}
|
||||
fi
|
||||
|
||||
token_list=${feats_dir}/data/${lang}_token_list/char/tokens.txt
|
||||
@ -135,23 +95,17 @@ echo "dictionary: ${token_list}"
|
||||
if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
|
||||
echo "stage 2: Dictionary Preparation"
|
||||
mkdir -p ${feats_dir}/data/${lang}_token_list/char/
|
||||
|
||||
|
||||
echo "make a dictionary"
|
||||
echo "<blank>" > ${token_list}
|
||||
echo "<s>" >> ${token_list}
|
||||
echo "</s>" >> ${token_list}
|
||||
tools/text2token.py -s 1 -n 1 --space "" ${feats_dir}/data/${train_set}/text | cut -f 2- -d" " | tr " " "\n" \
|
||||
utils/text2token.py -s 1 -n 1 --space "" ${feats_dir}/data/${train_set}/text | cut -f 2- -d" " | tr " " "\n" \
|
||||
| sort | uniq | grep -a -v -e '^\s*$' | awk '{print $0}' >> ${token_list}
|
||||
num_token=$(cat ${token_list} | wc -l)
|
||||
echo "<unk>" >> ${token_list}
|
||||
vocab_size=$(cat ${token_list} | wc -l)
|
||||
awk -v v=,${vocab_size} '{print $0v}' ${feat_train_dir}/text_shape > ${feat_train_dir}/text_shape.char
|
||||
awk -v v=,${vocab_size} '{print $0v}' ${feat_dev_dir}/text_shape > ${feat_dev_dir}/text_shape.char
|
||||
mkdir -p ${feats_dir}/asr_stats_fbank_zh_char/${train_set}
|
||||
mkdir -p ${feats_dir}/asr_stats_fbank_zh_char/${valid_set}
|
||||
cp ${feat_train_dir}/speech_shape ${feat_train_dir}/text_shape ${feat_train_dir}/text_shape.char ${feats_dir}/asr_stats_fbank_zh_char/${train_set}
|
||||
cp ${feat_dev_dir}/speech_shape ${feat_dev_dir}/text_shape ${feat_dev_dir}/text_shape.char ${feats_dir}/asr_stats_fbank_zh_char/${valid_set}
|
||||
fi
|
||||
fi
|
||||
|
||||
# Training Stage
|
||||
world_size=$gpu_num # run on one machine
|
||||
@ -170,21 +124,23 @@ if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
|
||||
rank=$i
|
||||
local_rank=$i
|
||||
gpu_id=$(echo $CUDA_VISIBLE_DEVICES | cut -d',' -f$[$i+1])
|
||||
asr_train.py \
|
||||
train.py \
|
||||
--task_name asr \
|
||||
--gpu_id $gpu_id \
|
||||
--use_preprocessor true \
|
||||
--dataset_type $dataset_type \
|
||||
--token_type char \
|
||||
--token_list $token_list \
|
||||
--train_data_file $feats_dir/$dumpdir/${train_set}/data.list \
|
||||
--valid_data_file $feats_dir/$dumpdir/${valid_set}/data.list \
|
||||
--data_dir ${feats_dir}/data \
|
||||
--train_set ${train_set} \
|
||||
--valid_set ${valid_set} \
|
||||
--cmvn_file ${feats_dir}/data/${train_set}/cmvn/cmvn.mvn \
|
||||
--speed_perturb ${speed_perturb} \
|
||||
--dataset_type $dataset_type \
|
||||
--resume true \
|
||||
--output_dir ${exp_dir}/exp/${model_dir} \
|
||||
--config $asr_config \
|
||||
--input_size $feats_dim \
|
||||
--ngpu $gpu_num \
|
||||
--num_worker_count $count \
|
||||
--multiprocessing_distributed true \
|
||||
--dist_init_method $init_method \
|
||||
--dist_world_size $world_size \
|
||||
--dist_rank $rank \
|
||||
@ -207,7 +163,7 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
|
||||
exit 0
|
||||
fi
|
||||
mkdir -p "${_logdir}"
|
||||
_data="${feats_dir}/${dumpdir}/${dset}"
|
||||
_data="${feats_dir}/data/${dset}"
|
||||
key_file=${_data}/${scp}
|
||||
num_scp_file="$(<${key_file} wc -l)"
|
||||
_nj=$([ $inference_nj -le $num_scp_file ] && echo "$inference_nj" || echo "$num_scp_file")
|
||||
@ -228,6 +184,7 @@ if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
|
||||
--njob ${njob} \
|
||||
--gpuid_list ${gpuid_list} \
|
||||
--data_path_and_name_and_type "${_data}/${scp},speech,${type}" \
|
||||
--cmvn_file ${feats_dir}/data/${train_set}/cmvn/cmvn.mvn \
|
||||
--key_file "${_logdir}"/keys.JOB.scp \
|
||||
--asr_train_config "${asr_exp}"/config.yaml \
|
||||
--asr_model_file "${asr_exp}"/"${inference_asr_model}" \
|
||||
|
||||
Loading…
Reference in New Issue
Block a user