speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

update

Browse Source
This commit is contained in:
嘉渊 2023-07-23 00:12:57 +08:00
parent c8680dcb7b
commit 3aad0e15ec
2 changed files with 63 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
egs/callhome/eend_ola/local/dump_feature.py
View File

@ -34,7 +34,7 @@ class KaldiData:
utt2spk_file = os.path.join(self.data_dir, 'utt2spk.{}'.format(idx)) utt2spk_file = os.path.join(self.data_dir, 'utt2spk.{}'.format(idx))
self.utt2spk = load_utt2spk(utt2spk_file) self.utt2spk = load_utt2spk(utt2spk_file)
wav_file = os.path.join(self.data_dir, 'wav.{}.scp'.format(idx)) wav_file = os.path.join(self.data_dir, 'wav.scp.{}'.format(idx))
self.wavs = load_wav_scp(wav_file) self.wavs = load_wav_scp(wav_file)
reco2dur_file = os.path.join(self.data_dir, 'reco2dur.{}'.format(idx)) reco2dur_file = os.path.join(self.data_dir, 'reco2dur.{}'.format(idx))

92
egs/callhome/eend_ola/run_test.sh
View File

@ -3,7 +3,7 @@
. ./path.sh || exit 1; . ./path.sh || exit 1;
# machines configuration # machines configuration
CUDA_VISIBLE_DEVICES="7" CUDA_VISIBLE_DEVICES="0"
gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}') gpu_num=$(echo $CUDA_VISIBLE_DEVICES | awk -F "," '{print NF}')
count=1 count=1
@ -12,7 +12,7 @@ dump_cmd=utils/run.pl
nj=64 nj=64
# feature configuration # feature configuration
data_dir="/nfs/wangjiaming.wjm/EEND_DATA_sad30_snr10n15n20/convert_test/data" data_dir="./data"
simu_feats_dir="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/simu_data/data" simu_feats_dir="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/simu_data/data"
simu_feats_dir_chunk2000="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/simu_data_chunk2000/data" simu_feats_dir_chunk2000="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/simu_data_chunk2000/data"
callhome_feats_dir_chunk2000="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/callhome_chunk2000/data" callhome_feats_dir_chunk2000="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/callhome_chunk2000/data"
@ -74,36 +74,68 @@ if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
simu_opts_sil_scale_array=(2 2 5 9) simu_opts_sil_scale_array=(2 2 5 9)
simu_opts_num_train=100000 simu_opts_num_train=100000
# for simulated data of chunk500 and chunk2000 # # for simulated data of chunk500 and chunk2000
for dset in swb_sre_cv swb_sre_tr; do # for dset in swb_sre_cv swb_sre_tr; do
if [ "$dset" == "swb_sre_tr" ]; then # if [ "$dset" == "swb_sre_tr" ]; then
n_mixtures=${simu_opts_num_train} # n_mixtures=${simu_opts_num_train}
dataset=train # dataset=train
else # else
n_mixtures=500 # n_mixtures=500
dataset=dev # dataset=dev
fi # fi
simu_data_dir=${dset}_ns"$(IFS="n"; echo "${simu_opts_num_speaker_array[*]}")"_beta"$(IFS="n"; echo "${simu_opts_sil_scale_array[*]}")"_${n_mixtures} # simu_data_dir=${dset}_ns"$(IFS="n"; echo "${simu_opts_num_speaker_array[*]}")"_beta"$(IFS="n"; echo "${simu_opts_sil_scale_array[*]}")"_${n_mixtures}
mkdir -p ${data_dir}/simu/data/${simu_data_dir}/.work # mkdir -p ${data_dir}/simu/data/${simu_data_dir}/.work
split_scps= # split_scps=
for n in $(seq $nj); do # for n in $(seq $nj); do
split_scps="$split_scps ${data_dir}/simu/data/${simu_data_dir}/.work/wav.$n.scp" # split_scps="$split_scps ${data_dir}/simu/data/${simu_data_dir}/.work/wav.scp.$n"
done # done
utils/split_scp.pl "${data_dir}/simu/data/${simu_data_dir}/wav.scp" $split_scps || exit 1 # utils/split_scp.pl "${data_dir}/simu/data/${simu_data_dir}/wav.scp" $split_scps || exit 1
python local/split.py ${data_dir}/simu/data/${simu_data_dir} # python local/split.py ${data_dir}/simu/data/${simu_data_dir}
# for chunk_size=500 # # for chunk_size=500
output_dir=${data_dir}/ark_data/dump/simu_data/$dataset # output_dir=${data_dir}/ark_data/dump/simu_data/$dataset
mkdir -p $output_dir/.logs # mkdir -p $output_dir/.logs
$dump_cmd --max-jobs-run $nj JOB=1:$nj $output_dir/.logs/dump.JOB.log \ # $dump_cmd --max-jobs-run $nj JOB=1:$nj $output_dir/.logs/dump.JOB.log \
# python local/dump_feature.py \
# --data_dir ${data_dir}/simu/data/${simu_data_dir}/.work \
# --output_dir $output_dir \
# --index JOB
# mkdir -p ${data_dir}/ark_data/dump/simu_data/data/$dataset
# python local/gen_feats_scp.py \
# --root_path ${data_dir}/ark_data/dump/simu_data/$dataset \
# --out_path ${data_dir}/ark_data/dump/simu_data/data/$dataset \
# --split_num $nj
# grep "ns2" ${data_dir}/ark_data/dump/simu_data/data/$dataset/feats.scp > ${data_dir}/ark_data/dump/simu_data/data/$dataset/feats_2spkr.scp
# # for chunk_size=2000
# output_dir=${data_dir}/ark_data/dump/simu_data_chunk2000/$dataset
# mkdir -p $output_dir/.logs
# $dump_cmd --max-jobs-run $nj JOB=1:$nj $output_dir/.logs/dump.JOB.log \
# python local/dump_feature.py \
# --data_dir ${data_dir}/simu/data/${simu_data_dir}/.work \
# --output_dir $output_dir \
# --index JOB \
# --num_frames 2000
# mkdir -p ${data_dir}/ark_data/dump/simu_data_chunk2000/data/$dataset
# python local/gen_feats_scp.py \
# --root_path ${data_dir}/ark_data/dump/simu_data_chunk2000/$dataset \
# --out_path ${data_dir}/ark_data/dump/simu_data_chunk2000/data/$dataset \
# --split_num $nj
# grep "ns2" ${data_dir}/ark_data/dump/simu_data_chunk2000/data/$dataset/feats.scp > ${data_dir}/ark_data/dump/simu_data_chunk2000/data/$dataset/feats_2spkr.scp
# done
# for callhome data
for dset in callhome1_spkall callhome2_spkall; do
find $data_dir/eval/$dset -maxdepth 1 -type f -exec cp {} {}.1 \;
output_dir=${data_dir}/ark_data/dump/callhome/$dset
python local/dump_feature.py \ python local/dump_feature.py \
--data_dir ${data_dir}/simu/data/${simu_data_dir}/.work \ --data_dir $data_dir/eval/$dset \
--output_dir ${data_dir}/ark_data/dump/simu_data/$dataset \ --output_dir $output_dir \
--index JOB --index 1 \
mkdir -p ${data_dir}/ark_data/dump/simu_data/data/$dataset --num_frames 2000
mkdir -p ${data_dir}/ark_data/dump/callhome/data/$dset
python local/gen_feats_scp.py \ python local/gen_feats_scp.py \
--root_path ${data_dir}/ark_data/dump/simu_data/$dataset \ --root_path ${data_dir}/ark_data/dump/callhome/$dset \
--out_path ${data_dir}/ark_data/dump/simu_data/data/$dataset \ --out_path ${data_dir}/ark_data/dump/callhome/data/$dset \
--split_num $nj --split_num 1
done done
fi fi