From e79c9a801e7e7458ce6894fa85178fa974dfd18b Mon Sep 17 00:00:00 2001
From: 嘉渊 <wangjiaming.wjm@alibaba-inc.com>
Date: 星期五, 21 七月 2023 11:28:18 +0800
Subject: [PATCH] update
---
egs/callhome/eend_ola/run_test.sh | 60 ++++++++++++++++++++++++++++++++++++++++++++++++++----------
1 files changed, 50 insertions(+), 10 deletions(-)
diff --git a/egs/callhome/eend_ola/run_test.sh b/egs/callhome/eend_ola/run_test.sh
index c198e73..57d6418 100644
--- a/egs/callhome/eend_ola/run_test.sh
+++ b/egs/callhome/eend_ola/run_test.sh
@@ -8,6 +8,11 @@
count=1
# general configuration
+dump_cmd=utils/run.pl
+nj=64
+
+# feature configuration
+data_dir="/nfs/wangjiaming.wjm/EEND_DATA_sad30_snr10n15n20/convert_test/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"
callhome_feats_dir_chunk2000="/nfs/wangjiaming.wjm/EEND_ARK_DATA/dump/callhome_chunk2000/data"
@@ -27,8 +32,8 @@
exp_dir="."
input_size=345
-stage=5
-stop_stage=5
+stage=0
+stop_stage=0
# exp tag
tag="exp1"
@@ -62,13 +67,45 @@
local/run_prepare_shared_eda.sh
fi
-## Prepare data for training and inference
-#if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
-# echo "stage 0: Prepare data for training and inference"
-# echo "The detail can be found in https://github.com/hitachi-speech/EEND"
-# . ./local/
-#fi
-#
+# Prepare data for training and inference
+if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
+ echo "stage 0: Prepare data for training and inference"
+ simu_opts_num_speaker_array=(1 2 3 4)
+ simu_opts_sil_scale_array=(2 2 5 9)
+ simu_opts_num_train=100000
+
+ # for simulated data of chunk500 and chunk2000
+ for dset in swb_sre_tr swb_sre_cv; do
+ if [ "$dset" == "swb_sre_tr" ]; then
+ n_mixtures=${simu_opts_num_train}
+ dataset=train
+ else
+ n_mixtures=500
+ dataset=dev
+ fi
+ simu_data_dir=${dset}_ns"$(IFS="n"; echo "${simu_opts_num_speaker_array[*]}")"_beta"$(IFS="n"; echo "${simu_opts_sil_scale_array[*]}")"_${n_mixtures}
+# mkdir -p ${data_dir}/simu/data/${simu_data_dir}/.work
+# split_scps=
+# for n in $(seq $nj); do
+# split_scps="$split_scps ${data_dir}/simu/data/${simu_data_dir}/.work/wav.$n.scp"
+# done
+# utils/split_scp.pl "${data_dir}/simu/data/${simu_data_dir}/wav.scp" $split_scps || exit 1
+# python local/split.py ${data_dir}/simu/data/${simu_data_dir}
+# # for chunk_size=500
+# output_dir=${data_dir}/ark_data/dump/simu_data/$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 ${data_dir}/ark_data/dump/simu_data/$dataset \
+# --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 \
+ --out_path ${data_dir}/ark_data/dump/simu_data/data/$dataset \
+ --split_num $nj
+ done
+fi
# Training on simulated two-speaker data
world_size=$gpu_num
@@ -245,7 +282,7 @@
python local/model_averaging.py ${exp_dir}/exp/${callhome_model_dir}/$callhome_ave_id.pb $models
fi
-# inference
+# inference and compute DER
if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
echo "Inference"
mkdir -p ${exp_dir}/exp/${callhome_model_dir}/inference/log
@@ -255,4 +292,7 @@
--output_rttm_file ${exp_dir}/exp/${callhome_model_dir}/inference/rttm \
--wav_scp_file ${callhome_feats_dir_chunk2000}/${callhome_valid_dataset}/${callhome2_wav_scp_file} \
1> ${exp_dir}/exp/${callhome_model_dir}/inference/log/infer.log 2>&1
+ md-eval.pl -c 0.25 \
+ -r ${callhome_feats_dir_chunk2000}/${callhome_valid_dataset}/rttm \
+ -s ${exp_dir}/exp/${callhome_model_dir}/inference/rttm > ${exp_dir}/exp/${callhome_model_dir}/inference/result_med11_collar0.25 2>/dev/null || exit
fi
\ No newline at end of file
--
Gitblit v1.9.1