update

examples/vm_sound_classification/run.sh

@@ -20,7 +20,7 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml"
 
 END
 
@@ -37,6 +37,7 @@ final_model_name=final_model_name
 filename_patterns="/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav"
 label_plan=4
 config_file="yaml/conv2d-classifier-2-ch4.yaml"
+pretrained_model=null
 nohup_name=nohup.out
 
 country=en-US
@@ -130,6 +131,7 @@ if [ ${stage} -le 2 ] && [ ${stop_stage} -ge 2 ]; then
   --valid_dataset "${valid_dataset}" \
   --serialization_dir "${file_dir}" \
   --config_file "${config_file}" \
+  --pretrained_model "${pretrained_model}" \
 
 fi
 

examples/vm_sound_classification/run_batch.sh

@@ -3,80 +3,80 @@
 
 # sound ch4
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch4 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 2 \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch4 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 3 \
---config_file "yaml/conv2d-classifier-3-ch4.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch4 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 4 \
---config_file "yaml/conv2d-classifier-4-ch4.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch4 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 8 \
---config_file "yaml/conv2d-classifier-8-ch4.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch4 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2 \
+#--config_file "yaml/conv2d-classifier-2-ch4.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch4 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 3 \
+#--config_file "yaml/conv2d-classifier-3-ch4.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch4 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 4 \
+#--config_file "yaml/conv2d-classifier-4-ch4.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch4 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 8 \
+#--config_file "yaml/conv2d-classifier-8-ch4.yaml"
 
 
 # sound ch8
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch8 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 2 \
---config_file "yaml/conv2d-classifier-2-ch8.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch8 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 3 \
---config_file "yaml/conv2d-classifier-3-ch8.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch8 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 4 \
---config_file "yaml/conv2d-classifier-4-ch8.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch8 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 8 \
---config_file "yaml/conv2d-classifier-8-ch8.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch8 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2 \
+#--config_file "yaml/conv2d-classifier-2-ch8.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch8 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 3 \
+#--config_file "yaml/conv2d-classifier-3-ch8.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch8 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 4 \
+#--config_file "yaml/conv2d-classifier-4-ch8.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch8 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 8 \
+#--config_file "yaml/conv2d-classifier-8-ch8.yaml"
 
 
 # sound ch16
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch16 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 2 \
---config_file "yaml/conv2d-classifier-2-ch16.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch16 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 3 \
---config_file "yaml/conv2d-classifier-3-ch16.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch16 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 4 \
---config_file "yaml/conv2d-classifier-4-ch16.yaml"
-
-
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch16 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 8 \
---config_file "yaml/conv2d-classifier-8-ch16.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-ch16 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2 \
+#--config_file "yaml/conv2d-classifier-2-ch16.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch16 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 3 \
+#--config_file "yaml/conv2d-classifier-3-ch16.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-ch16 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 4 \
+#--config_file "yaml/conv2d-classifier-4-ch16.yaml"
+#
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch16 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 8 \
+#--config_file "yaml/conv2d-classifier-8-ch16.yaml"
 
 
 # sound ch32
@@ -99,10 +99,23 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 #--config_file "yaml/conv2d-classifier-4-ch32.yaml"
 
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch32 \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 8 \
---config_file "yaml/conv2d-classifier-8-ch32.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-ch32 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 8 \
+#--config_file "yaml/conv2d-classifier-8-ch32.yaml"
+
+
+# pretrained voicemail
+
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-ch4 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2-voicemail \
+#--config_file "yaml/conv2d-classifier-2-ch4.yaml"
+#
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-ch32 \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2-voicemail \
+#--config_file "yaml/conv2d-classifier-2-ch32.yaml"
 
 
 # voicemail ch4
@@ -110,57 +123,68 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-sg-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-SG/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-us-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-US/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-PE/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-id-id-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/id-ID/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-JP-2-ch4 \
+sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-jp-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ja-JP/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ko-kr-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ko-KR/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch4 \
-#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
-#--label_plan 2-voicemail \
-#--config_file "yaml/conv2d-classifier-2-ch4.yaml"
+sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch4 \
+--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
+--label_plan 2-voicemail \
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-pt-br-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/pt-BR/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-th-th-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/th-TH/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-zh-tw-2-ch4 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/zh-TW/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch4.yaml"
+--config_file "yaml/conv2d-classifier-2-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch4.zip"
 
 
 # voicemail ch32
@@ -168,54 +192,65 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-sg-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-SG/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-us-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-US/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-PE/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-id-id-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/id-ID/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-JP-2-ch32 \
+sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-jp-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ja-JP/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ko-kr-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ko-KR/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-pt-br-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/pt-BR/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-th-th-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/th-TH/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"
 
 sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-zh-tw-2-ch32 \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/zh-TW/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-ch32.yaml"
+--config_file "yaml/conv2d-classifier-2-ch32.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/vm_sound_classification/trained_models/voicemail-2-ch32.zip"

examples/vm_sound_classification/step_1_prepare_data.py

@@ -19,7 +19,6 @@ from tqdm import tqdm
 def get_args():
     parser = argparse.ArgumentParser()
     parser.add_argument("--file_dir", default="./", type=str)
-    parser.add_argument("--task", default="default", type=str)
     parser.add_argument("--filename_patterns", type=str)
 
     parser.add_argument("--train_dataset", default="train.xlsx", type=str)

examples/vm_sound_classification/step_3_train_model.py

@@ -11,7 +11,9 @@ from pathlib import Path
 import random
 import sys
 import shutil
+import tempfile
 from typing import List
+import zipfile
 
 pwd = os.path.abspath(os.path.dirname(__file__))
 sys.path.append(os.path.join(pwd, "../../"))
@@ -21,6 +23,7 @@ import torch
 from torch.utils.data.dataloader import DataLoader
 from tqdm import tqdm
 
+from project_settings import project_path
 from toolbox.torch.modules.loss import FocalLoss, HingeLoss, HingeLinear
 from toolbox.torch.training.metrics.categorical_accuracy import CategoricalAccuracy
 from toolbox.torch.utils.data.vocabulary import Vocabulary
@@ -46,6 +49,12 @@ def get_args():
     parser.add_argument("--seed", default=0, type=int)
 
     parser.add_argument("--config_file", default="conv2d_classifier.yaml", type=str)
+    parser.add_argument(
+        "--pretrained_model",
+        # default=(project_path / "trained_models/voicemail-en-sg-2-ch4.zip").as_posix(),
+        default="null",
+        type=str
+    )
 
     args = parser.parse_args()
     return args
@@ -175,6 +184,23 @@ def main():
     model = WaveClassifierPretrainedModel(
         config=config,
     )
+
+    if args.pretrained_model is not None and os.path.exists(args.pretrained_model):
+        logger.info(f"load pretrained model state dict from: {args.pretrained_model}")
+        with zipfile.ZipFile(args.pretrained_model, "r") as f_zip:
+            out_root = Path(tempfile.gettempdir()) / "vm_sound_classification"
+            # print(out_root.as_posix())
+            if out_root.exists():
+                shutil.rmtree(out_root.as_posix())
+            out_root.mkdir(parents=True, exist_ok=True)
+            f_zip.extractall(path=out_root)
+
+        tgt_path = out_root / os.path.basename(args.pretrained_model)
+        model_pt_file = tgt_path / "model.pt"
+        with open(model_pt_file, "r") as f:
+            state_dict = torch.load(f, map_location="cpu")
+        model.load_state_dict(state_dict=state_dict)
+
     model.to(device)
     model.train()