Update pipeline

51f72cdf · Marcin Wątroba · 428b93f6 · 51f72cdf · 51f72cdf · 428b93f6
Commit 51f72cdf authored Apr 29, 2022 by Marcin Wątroba
--- a/dvc.lock
+++ b/dvc.lock
@@ -55,15 +55,15 @@ stages:
    cmd: PYTHONPATH=. python experiment/voicelab/import_data.py
    deps:
    - path: experiment/voicelab/import_data.py
-      md5: 0cf7cf604b202489ce3b0cb51bb47fa2
-      size: 2264
+      md5: 41acb98a1517e66c052182fe0a1403ba
+      size: 2108
    - path: experiment_data/dataset/voicelab_cbiz_testset_20220322
      md5: 3c2b18e1f1f89e4c5ad7b254e472b25e.dir
      size: 4803739404
      nfiles: 1600
    outs:
    - path: experiment_data/dataset_relation_manager_data/voicelab_cbiz_testset_20220322
-      md5: 6d56f24b0ff78c0d44ade2114158150d.dir
+      md5: 926ef9bab4ce41b9de95f2f3d5ab67a0.dir
      size: 110711470
      nfiles: 1600
  luna_gold_transcript_processing:
@@ -179,3 +179,73 @@ stages:
      md5: 4cfbb2830b280084ece14b1ef815b92a.dir
      size: 17298
      nfiles: 500
+  voicelab_gold_transcript_processing:
+    cmd: "PYTHONPATH=. python -u experiment/voicelab/voicelab_pipeline_gold_transcript.py\n"
+    deps:
+    - path: experiment/voicelab/voicelab_pipeline_gold_transcript.py
+      md5: 4ba38fdfac616f8a0818cedabf66b94d
+      size: 2312
+    - path: experiment_data/dataset/voicelab_cbiz_testset_20220322
+      md5: 3c2b18e1f1f89e4c5ad7b254e472b25e.dir
+      size: 4803739404
+      nfiles: 1600
+    - path: experiment_data/dataset_relation_manager_data/voicelab_cbiz_testset_20220322
+      md5: 926ef9bab4ce41b9de95f2f3d5ab67a0.dir
+      size: 110711470
+      nfiles: 1600
+    outs:
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript
+      md5: fb6812b2f3044c0285ee6ee2b21d0523.dir
+      size: 21846798
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript_spacy
+      md5: f2e68dcc8842a15e417ae6f5221a802a.dir
+      size: 26643278
+      nfiles: 800
+  voicelab_techmo_processing:
+    cmd: "PYTHONPATH=. python -u experiment/voicelab/voicelab_pipeline_techmo.py\n"
+    deps:
+    - path: experiment/voicelab/voicelab_pipeline_techmo.py
+      md5: 23c0869d7cc9f0088870362d669ab82e
+      size: 2685
+    - path: experiment_data/cached_asr/voicelab_cbiz_testset_20220322_techmo
+      md5: 6c3b356723d562c978f84e733b91f5d0.dir
+      size: 17539259
+      nfiles: 800
+    - path: experiment_data/dataset/voicelab_cbiz_testset_20220322
+      md5: 3c2b18e1f1f89e4c5ad7b254e472b25e.dir
+      size: 4803739404
+      nfiles: 1600
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript
+      md5: fb6812b2f3044c0285ee6ee2b21d0523.dir
+      size: 21846798
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript_spacy
+      md5: f2e68dcc8842a15e417ae6f5221a802a.dir
+      size: 26643278
+      nfiles: 800
+    outs:
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_techmo_alignment_wer
+      md5: 8c5f0380ba2891b3e726d647c2863c60.dir
+      size: 81650836
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_techmo_metrics_wer
+      md5: b1a674826142a44095a4c6439ac49024.dir
+      size: 27934
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/techmo_polish_asr
+      md5: c45e29b08af7bb13cdf54da9655bd96c.dir
+      size: 39158267
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/techmo_spacy
+      md5: a39c82666419c2b7791952a1fa116d61.dir
+      size: 24482297
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_techmo_alignment_wer
+      md5: 72ff86c7cb2e89ac7e04677f532255b2.dir
+      size: 83756423
+      nfiles: 800
+    - path: experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_techmo_metrics_wer
+      md5: 2fe3288abe85e4a385e2aefa0e8cad7e.dir
+      size: 27780
+      nfiles: 800
--- a/dvc.yaml
+++ b/dvc.yaml
@@ -62,9 +62,9 @@ stages:

    voicelab_gold_transcript_processing:
        cmd: |
-            PYTHONPATH=. python experiment/voicelab/voicelab_gold_transcript_processor.py
+            PYTHONPATH=. python -u experiment/voicelab/voicelab_pipeline_gold_transcript.py
        deps:
-            - experiment/voicelab/voicelab_gold_transcript_processor.py
+            - experiment/voicelab/voicelab_pipeline_gold_transcript.py
            - experiment_data/dataset_relation_manager_data/voicelab_cbiz_testset_20220322
            - experiment_data/dataset/voicelab_cbiz_testset_20220322
        outs:
@@ -73,7 +73,7 @@ stages:

    voicelab_techmo_processing:
        cmd: |
-            PYTHONPATH=. python experiment/voicelab/voicelab_pipeline_techmo.py
+            PYTHONPATH=. python -u experiment/voicelab/voicelab_pipeline_techmo.py
        deps:
            - experiment/voicelab/voicelab_pipeline_techmo.py
            - experiment_data/dataset/voicelab_cbiz_testset_20220322
@@ -88,20 +88,20 @@ stages:
            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_techmo_alignment_wer
            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_techmo_metrics_wer

-    voicelab_ajn_processing:
-        cmd: |
-            PYTHONPATH=. python experiment/voicelab/voicelab_pipeline_ajn_asr.py
-        deps:
-            - experiment/voicelab/voicelab_pipeline_ajn_asr.py
-            - experiment_data/dataset/voicelab_cbiz_testset_20220322
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript_spacy
-        outs:
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/ajn_polish_asr
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_ajn_metrics_wer
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_ajn_alignment_wer
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/ajn_spacy
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_ajn_alignment_wer
-            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_ajn_metrics_wer
+#    voicelab_ajn_processing:
+#        cmd: |
+#            PYTHONPATH=. python -u experiment/voicelab/voicelab_pipeline_ajn_asr.py
+#        deps:
+#            - experiment/voicelab/voicelab_pipeline_ajn_asr.py
+#            - experiment_data/dataset/voicelab_cbiz_testset_20220322
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/gold_transcript_spacy
+#        outs:
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/ajn_polish_asr
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_ajn_metrics_wer
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/word_ajn_alignment_wer
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/ajn_spacy
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_ajn_alignment_wer
+#            - experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/pos_ajn_metrics_wer

 # concurrent features, multiprocessing
--- a/experiment/luna/pos.ipynb
+++ b/experiment/luna/pos.ipynb
--- a/experiment/voicelab/voicelab_pipeline_ajn_asr.py
+++ b/experiment/voicelab/voicelab_pipeline_ajn_asr.py
@@ -64,4 +64,6 @@ def example_run():


 if __name__ == '__main__':
-    example_run()
+    # example_run()
+    path = '/home/marcinwatroba/PWR_ASR/asr-benchmarks/experiment_data/dataset/voicelab_cbiz_testset_20220322/bankowe/cbiz_tc_2.agnt.wav'
+    print(AsrWebClient('http://localhost:5431/process_asr', '__example_token__').call_recognise(path))
--- a/experiment_data/cached_asr/voicelab_cbiz_testset_20220322_techmo.dvc
+++ b/experiment_data/cached_asr/voicelab_cbiz_testset_20220322_techmo.dvc
 outs:
- md5: 94b1709c05bd09b77c5a6850e2f2f373.dir
-  size: 34654307
+- md5: 6c3b356723d562c978f84e733b91f5d0.dir
+  size: 17539259
  nfiles: 800
  path: voicelab_cbiz_testset_20220322_techmo
--- a/experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/.gitignore
+++ b/experiment_data/pipeline/asr_benchmark_voicelab_cbiz_testset_20220322/.gitignore
+/gold_transcript
+/gold_transcript_spacy
+/techmo_polish_asr
+/word_techmo_metrics_wer
+/word_techmo_alignment_wer
+/techmo_spacy
+/pos_techmo_alignment_wer
+/pos_techmo_metrics_wer
--- a/sziszapangma/integration/repository/multi_files_experiment_repository.py
+++ b/sziszapangma/integration/repository/multi_files_experiment_repository.py
 import json
+import os.path
 from pathlib import Path
 from typing import Any, List, Optional, Set

@@ -48,7 +49,7 @@ class MultiFilesExperimentRepository(ExperimentRepository):

    def get_all_properties(self) -> Set[str]:
        experiment_path = Path(self._root_directory).joinpath(self._experiment_name)
-        return set([it.name for it in experiment_path.iterdir()])
+        return set([it.name for it in experiment_path.iterdir() if os.path.isdir(it)])

    def _get_file_path(self, property_name: str, record_id: str) -> Path:
        return (