update

examples/sound_classification_by_cnn/run_batch_l3.sh

@@ -9,10 +9,10 @@
 #--config_file "yaml/conv2d-classifier-2-l3-ch4.yaml"
 
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch4-cnn \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 3 \
---config_file "yaml/conv2d-classifier-3-l3-ch4.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch4-cnn \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 3 \
+#--config_file "yaml/conv2d-classifier-3-l3-ch4.yaml"
 
 
 #sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-l3-ch4-cnn \
@@ -87,10 +87,10 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 #--config_file "yaml/conv2d-classifier-2-l3-ch32.yaml"
 
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch32-cnn \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 3 \
---config_file "yaml/conv2d-classifier-3-l3-ch32.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch32-cnn \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 3 \
+#--config_file "yaml/conv2d-classifier-3-l3-ch32.yaml"
 
 
 #sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-l3-ch32-cnn \
@@ -122,20 +122,20 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 #--label_plan 2-voicemail \
 #--config_file "yaml/conv2d-classifier-2-l3-ch16.yaml"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-l3-ch32-cnn \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
---label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-l3-ch32.yaml"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-l3-ch32-cnn \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
+#--label_plan 2-voicemail \
+#--config_file "yaml/conv2d-classifier-2-l3-ch32.yaml"
 
 
 # voicemail ch4
 
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-ph-2-l3-ch4-cnn \
-#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-PH/wav_finished/*/*.wav" \
-#--label_plan 2-voicemail \
-#--config_file "yaml/conv2d-classifier-2-l3-ch4.yaml" \
-#--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch4-cnn.zip"
-#
+sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-ph-2-l3-ch4-cnn \
+--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-PH/wav_finished/*/*.wav" \
+--label_plan 2-voicemail \
+--config_file "yaml/conv2d-classifier-2-l3-ch4.yaml" \
+--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch4-cnn.zip"
+
 #sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-sg-2-l3-ch4-cnn \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-SG/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
@@ -148,11 +148,11 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 #--config_file "yaml/conv2d-classifier-2-l3-ch4.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch4-cnn.zip"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch4-cnn \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
---label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-l3-ch4.yaml" \
---pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch4-cnn.zip"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch4-cnn \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
+#--label_plan 2-voicemail \
+#--config_file "yaml/conv2d-classifier-2-l3-ch4.yaml" \
+#--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch4-cnn.zip"
 #
 #sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-l3-ch4-cnn \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-PE/wav_finished/*/*.wav" \
@@ -250,11 +250,11 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 #--config_file "yaml/conv2d-classifier-2-l3-ch32.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch32-cnn.zip"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch32-cnn \
---filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
---label_plan 2-voicemail \
---config_file "yaml/conv2d-classifier-2-l3-ch32.yaml" \
---pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch32-cnn.zip"
+#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch32-cnn \
+#--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
+#--label_plan 2-voicemail \
+#--config_file "yaml/conv2d-classifier-2-l3-ch32.yaml" \
+#--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-ch32-cnn.zip"
 
 #sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-l3-ch32-cnn \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-PE/wav_finished/*/*.wav" \

examples/sound_classification_by_lstm/run.sh

@@ -12,17 +12,17 @@ sh run.sh --stage 2 --stop_stage 2 --system_version windows --file_folder_name f
 E:/Users/tianx/HuggingDatasets/cc_audio_8/data/wav_finished/id-ID/wav_finished/*/*.wav" \
 --label_plan 4
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch32-lstm \
+sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name sound-3-ch32-lstm \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 --label_plan 3 \
 --config_file "yaml/lstm_classifier-3-ch64.yaml"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch32-lstm \
+sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch32-lstm \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
 --config_file "yaml/lstm_classifier-2-ch64.yaml"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch64-lstm \
+sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-ch64-lstm \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
 --config_file "yaml/lstm-classifier-2-ch64.yaml"
@@ -90,6 +90,7 @@ dataset="${file_dir}/dataset.xlsx"
 train_dataset="${file_dir}/train.xlsx"
 valid_dataset="${file_dir}/valid.xlsx"
 evaluation_file="${file_dir}/evaluation.xlsx"
+evaluation_onnx_file="${file_dir}/evaluation-onnx.xlsx"
 vocabulary_dir="${file_dir}/vocabulary"
 
 $verbose && echo "system_version: ${system_version}"
@@ -142,7 +143,7 @@ fi
 
 
 if [ ${stage} -le 3 ] && [ ${stop_stage} -ge 3 ]; then
-  $verbose && echo "stage 3: test model"
+  $verbose && echo "stage 3: evaluate model"
   cd "${work_dir}" || exit 1
   python3 step_4_evaluation_model.py \
   --dataset "${dataset}" \
@@ -170,7 +171,22 @@ fi
 
 
 if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
-  $verbose && echo "stage 5: collect files"
+  $verbose && echo "stage 5: evaluate onnx model"
+  cd "${work_dir}" || exit 1
+
+  python3 step_9_evaluation_onnx_model.py \
+  --model_file "null" \
+  --dataset "${evaluation_file}" \
+  --vocabulary_dir "${vocabulary_dir}" \
+  --model_dir "${file_dir}/best" \
+  --onnx_model_file "${file_dir}/model.onnx" \
+  --output_file "${evaluation_onnx_file}" \
+
+fi
+
+
+if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
+  $verbose && echo "stage 6: collect files"
   cd "${work_dir}" || exit 1
 
   mkdir -p ${final_model_dir}
@@ -178,7 +194,8 @@ if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
   cp "${file_dir}/best"/* "${final_model_dir}"
   cp -r "${file_dir}/vocabulary" "${final_model_dir}"
 
-  cp "${file_dir}/evaluation.xlsx" "${final_model_dir}/evaluation.xlsx"
+  cp "${evaluation_file}" "${final_model_dir}/evaluation.xlsx"
+  cp "${evaluation_onnx_file}" "${final_model_dir}/evaluation-onnx.xlsx"
 
   cp "${file_dir}/model.onnx" "${final_model_dir}/model.onnx"
 
@@ -200,8 +217,8 @@ if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
 fi
 
 
-if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
-  $verbose && echo "stage 6: clear file_dir"
+if [ ${stage} -le 7 ] && [ ${stop_stage} -ge 7 ]; then
+  $verbose && echo "stage 7: clear file_dir"
   cd "${work_dir}" || exit 1
 
   rm -rf "${file_dir}";

examples/sound_classification_by_lstm/run_batch.sh

@@ -3,22 +3,22 @@
 
 # sound ch64
 
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name sound-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 #--label_plan 2 \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name sound-3-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 #--label_plan 3 \
 #--config_file "yaml/lstm-classifier-3-l3-ch64.yaml"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-4-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name sound-4-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 #--label_plan 4 \
 #--config_file "yaml/lstm-classifier-4-l3-ch64.yaml"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name sound-8-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name sound-8-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 #--label_plan 8 \
 #--config_file "yaml/lstm-classifier-8-l3-ch64.yaml"
@@ -26,7 +26,7 @@
 
 # pretrained voicemail
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-l3-ch64-lstm \
+sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-2-l3-ch64-lstm \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/*/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
 --config_file "yaml/lstm-classifier-2-l3-ch64.yaml"
@@ -34,79 +34,79 @@ sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name fi
 
 # voicemail ch64
 
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-ph-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-ph-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-PH/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-sg-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-sg-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-SG/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-us-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-en-us-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/en-US/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 
-sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch64-lstm \
+sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-mx-2-l3-ch64-lstm \
 --filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-MX/wav_finished/*/*.wav" \
 --label_plan 2-voicemail \
 --config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 --pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-es-pe-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/es-PE/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-fi-fi-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-fi-fi-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/fi-FI/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-id-id-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-id-id-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/id-ID/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-jp-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ja-jp-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ja-JP/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ko-kr-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ko-kr-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ko-KR/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-ms-my-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/ms-MY/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-pt-br-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-pt-br-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/pt-BR/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-th-th-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-th-th-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/th-TH/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \
 #--pretrained_model "/data/tianxing/PycharmProjects/cc_audio_8/trained_models/voicemail-2-l3-ch64-lstm.zip"
 #
-#sh run.sh --stage 0 --stop_stage 6 --system_version centos --file_folder_name file_dir --final_model_name voicemail-zh-tw-2-l3-ch64-lstm \
+#sh run.sh --stage 0 --stop_stage 7 --system_version centos --file_folder_name file_dir --final_model_name voicemail-zh-tw-2-l3-ch64-lstm \
 #--filename_patterns "/data/tianxing/PycharmProjects/datasets/voicemail/zh-TW/wav_finished/*/*.wav" \
 #--label_plan 2-voicemail \
 #--config_file "yaml/lstm-classifier-2-l3-ch64.yaml" \

examples/sound_classification_by_lstm/step_9_evaluation_onnx_model.py

@@ -33,13 +33,19 @@ def get_args():
     parser = argparse.ArgumentParser()
     parser.add_argument(
         "--model_file",
-        # default=(project_path / "trained_models/cc_audio_8.zip").as_posix(),
-        default=(project_path / "trained_models/voicemail-ms-my-2-ch64-lstm.zip").as_posix(),
+        default="null",
+        # default=(project_path / "trained_models/voicemail-ms-my-2-ch64-lstm.zip").as_posix(),
         type=str
     )
+    parser.add_argument("--dataset", default="evaluation.xlsx", type=str)
+    parser.add_argument("--vocabulary_dir", default="vocabulary", type=str)
+    parser.add_argument("--model_dir", default="best", type=str)
+    parser.add_argument("--onnx_model_file", default="model.onnx", type=str)
     parser.add_argument("--output_file", default="evaluation_onnx.xlsx", type=str)
     parser.add_argument("--device", default="cpu", type=str)
 
+    parser.add_argument("--max_count", default=10000, type=int)
+
     args = parser.parse_args()
     return args
 
@@ -47,23 +53,29 @@ def get_args():
 def main():
     args = get_args()
 
-    model_file = Path(args.model_file)
+    model_dir = Path(args.model_dir)
 
     device = torch.device(args.device)
 
-    with zipfile.ZipFile(model_file, "r") as f_zip:
-        out_root = Path(tempfile.gettempdir()) / "cc_audio_8"
-        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 / model_file.stem
-    config_file = tgt_path / "config.yaml"
-    onnx_model_file = tgt_path / "model.onnx"
-    vocab_path = tgt_path / "vocabulary"
-    evaluation_file = tgt_path / "evaluation.xlsx"
+    if args.model_file != "null":
+        model_file = Path(args.model_file)
+        with zipfile.ZipFile(model_file.as_posix(), "r") as f_zip:
+            out_root = Path(tempfile.gettempdir()) / "cc_audio_8"
+            # 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 / model_file.stem
+        config_file = tgt_path / "config.yaml"
+        onnx_model_file = tgt_path / "model.onnx"
+        vocab_path = tgt_path / "vocabulary"
+        evaluation_file = tgt_path / "evaluation.xlsx"
+    else:
+        config_file = model_dir / "config.yaml"
+        onnx_model_file = args.onnx_model_file
+        vocab_path = args.vocabulary_dir
+        evaluation_file = args.dataset
 
     config = WaveClassifierConfig.from_pretrained(config_file.as_posix())
     ort_session = ort.InferenceSession(onnx_model_file.as_posix())
@@ -87,8 +99,12 @@ def main():
     total_correct = 0
     total_examples = 0
 
+    count = 0
     progress_bar = tqdm(total=len(df), desc="Evaluation ONNX")
     for i, row in df.iterrows():
+        if count >= args.max_count:
+            break
+
         row = dict(row)
         filename = row["filename"]
         ground_true = row["labels"]
@@ -158,6 +174,7 @@ def main():
             row_[f"onnx_correct_{idx}"] = correct
 
         result.append(row_)
+        count += 1
 
         total_examples += 1
         total_correct += row_["onnx_correct_3"]