File size: 5,092 Bytes
69ad385
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
#!/usr/bin/env bash

: <<'END'

sh run.sh --stage 3 --stop_stage 3 --system_version windows --file_folder_name file_dir --final_model_name vm_sound_classification8 \
--filename_patterns "E:/programmer/asr_datasets/voicemail/wav_finished/en-US/wav_finished/*/*.wav \
E:/programmer/asr_datasets/voicemail/wav_finished/id-ID/wav_finished/*/*.wav" \


sh run.sh --stage 3 --stop_stage 3 --system_version windows --file_folder_name file_dir --final_model_name vm_sound_classification8 \
--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav"

sh run.sh --stage 4 --stop_stage 4 --system_version windows --file_folder_name file_dir --final_model_name vm_sound_classification8 \
--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav"

sh run.sh --stage 4 --stop_stage 4 --system_version centos --file_folder_name file_dir --final_model_name vm_sound_classification8 \
--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav"


"

END


# sh run.sh --stage -1 --stop_stage 9
# sh run.sh --stage -1 --stop_stage 5 --system_version centos --file_folder_name task_cnn_voicemail_id_id --final_model_name cnn_voicemail_id_id
# sh run.sh --stage 3 --stop_stage 4
# sh run.sh --stage 4 --stop_stage 4
# sh run.sh --stage 3 --stop_stage 3 --system_version centos --file_folder_name task_cnn_voicemail_id_id

# params
system_version="windows";
verbose=true;
stage=0 # start from 0 if you need to start from data preparation
stop_stage=9

work_dir="$(pwd)"
file_folder_name=file_folder_name
final_model_name=final_model_name
filename_patterns="/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav"
nohup_name=nohup.out

country=en-US

# model params
batch_size=64
max_epochs=200
save_top_k=10
patience=5


# parse options
while true; do
  [ -z "${1:-}" ] && break;  # break if there are no arguments
  case "$1" in
    --*) name=$(echo "$1" | sed s/^--// | sed s/-/_/g);
      eval '[ -z "${'"$name"'+xxx}" ]' && echo "$0: invalid option $1" 1>&2 && exit 1;
      old_value="(eval echo \\$$name)";
      if [ "${old_value}" == "true" ] || [ "${old_value}" == "false" ]; then
        was_bool=true;
      else
        was_bool=false;
      fi

      # Set the variable to the right value-- the escaped quotes make it work if
      # the option had spaces, like --cmd "queue.pl -sync y"
      eval "${name}=\"$2\"";

      # Check that Boolean-valued arguments are really Boolean.
      if $was_bool && [[ "$2" != "true" && "$2" != "false" ]]; then
        echo "$0: expected \"true\" or \"false\": $1 $2" 1>&2
        exit 1;
      fi
      shift 2;
      ;;

    *) break;
  esac
done

file_dir="${work_dir}/${file_folder_name}"
final_model_dir="${work_dir}/../../trained_models/${final_model_name}";

train_dataset="${file_dir}/train.xlsx"
valid_dataset="${file_dir}/valid.xlsx"
vocabulary_dir="${file_dir}/vocabulary"


$verbose && echo "system_version: ${system_version}"
$verbose && echo "file_folder_name: ${file_folder_name}"

if [ $system_version == "windows" ]; then
  alias python3='D:/Users/tianx/PycharmProjects/virtualenv/vm_sound_classification/Scripts/python.exe'
elif [ $system_version == "centos" ] || [ $system_version == "ubuntu" ]; then
  #source /data/local/bin/vm_sound_classification/bin/activate
  alias python3='/data/local/bin/vm_sound_classification/bin/python3'
fi


if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
  $verbose && echo "stage 0: prepare data"
  cd "${work_dir}" || exit 1
  python3 step_1_prepare_data.py \
  --file_dir "${file_dir}" \
  --filename_patterns "${filename_patterns}" \
  --train_dataset "${train_dataset}" \
  --valid_dataset "${valid_dataset}" \

fi


if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then
  $verbose && echo "stage 1: make vocabulary"
  cd "${work_dir}" || exit 1
  python3 step_2_make_vocabulary.py \
  --vocabulary_dir "${vocabulary_dir}" \
  --train_dataset "${train_dataset}" \
  --valid_dataset "${valid_dataset}" \

fi


if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
  $verbose && echo "stage 2: train global model"
  cd "${work_dir}" || exit 1
  python3 step_3_train_global_model.py \
  --vocabulary_dir "${vocabulary_dir}" \
  --train_dataset "${train_dataset}" \
  --valid_dataset "${valid_dataset}" \
  --serialization_dir "${file_dir}/global_model" \

fi


if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
  $verbose && echo "stage 3: train country model"
  cd "${work_dir}" || exit 1
  python3 step_4_train_country_model.py \
  --vocabulary_dir "${vocabulary_dir}" \
  --train_dataset "${train_dataset}" \
  --valid_dataset "${valid_dataset}" \
  --country "${country}" \
  --serialization_dir "${file_dir}/country_model" \

fi


if [ ${stage} -le 4 ] && [ ${stop_stage} -ge 4 ]; then
  $verbose && echo "stage 4: train union model"
  cd "${work_dir}" || exit 1
  python3 step_5_train_union.py \
  --vocabulary_dir "${vocabulary_dir}" \
  --train_dataset "${train_dataset}" \
  --valid_dataset "${valid_dataset}" \
  --serialization_dir "${file_dir}/union" \

fi