batchalign 0.7.3b3__tar.gz → 0.7.3b5__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (107) hide show
  1. {batchalign-0.7.3b3/batchalign.egg-info → batchalign-0.7.3b5}/PKG-INFO +1 -1
  2. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/document.py +5 -0
  3. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/utils.py +1 -0
  4. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/dispatch.py +1 -1
  5. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/morphosyntax/ud.py +1 -1
  6. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utterance/ud_utterance.py +1 -1
  7. batchalign-0.7.3b5/batchalign/version +3 -0
  8. {batchalign-0.7.3b3 → batchalign-0.7.3b5/batchalign.egg-info}/PKG-INFO +1 -1
  9. batchalign-0.7.3b3/batchalign/version +0 -3
  10. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/LICENSE +0 -0
  11. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/MANIFEST.in +0 -0
  12. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/README.md +0 -0
  13. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/__init__.py +0 -0
  14. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/__main__.py +0 -0
  15. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/cli/__init__.py +0 -0
  16. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/cli/cli.py +0 -0
  17. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/cli/dispatch.py +0 -0
  18. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/constants.py +0 -0
  19. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/errors.py +0 -0
  20. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/__init__.py +0 -0
  21. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/base.py +0 -0
  22. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/__init__.py +0 -0
  23. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/file.py +0 -0
  24. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/generator.py +0 -0
  25. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/lexer.py +0 -0
  26. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/parser.py +0 -0
  27. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/chat/utils.py +0 -0
  28. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/textgrid/__init__.py +0 -0
  29. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/textgrid/file.py +0 -0
  30. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/textgrid/generator.py +0 -0
  31. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/formats/textgrid/parser.py +0 -0
  32. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/__init__.py +0 -0
  33. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/resolve.py +0 -0
  34. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/speaker/__init__.py +0 -0
  35. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/speaker/config.yaml +0 -0
  36. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/speaker/infer.py +0 -0
  37. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/speaker/utils.py +0 -0
  38. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/training/__init__.py +0 -0
  39. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/training/run.py +0 -0
  40. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/training/utils.py +0 -0
  41. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utils.py +0 -0
  42. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/__init__.py +0 -0
  43. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/dataset.py +0 -0
  44. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/execute.py +0 -0
  45. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/infer.py +0 -0
  46. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/prep.py +0 -0
  47. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/utterance/train.py +0 -0
  48. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/whisper/__init__.py +0 -0
  49. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/whisper/infer_asr.py +0 -0
  50. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/models/whisper/infer_fa.py +0 -0
  51. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/__init__.py +0 -0
  52. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/analysis/__init__.py +0 -0
  53. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/analysis/eval.py +0 -0
  54. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/__init__.py +0 -0
  55. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/rev.py +0 -0
  56. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/whisper.py +0 -0
  57. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/whisperx.py +0 -0
  58. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/base.py +0 -0
  59. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/__init__.py +0 -0
  60. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/cleanup.py +0 -0
  61. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/disfluencies.py +0 -0
  62. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/parse_support.py +0 -0
  63. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/retrace.py +0 -0
  64. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/support/filled_pauses.eng +0 -0
  65. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/support/replacements.eng +0 -0
  66. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/cleanup/support/test.test +0 -0
  67. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/fa/__init__.py +0 -0
  68. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/fa/whisper_fa.py +0 -0
  69. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/morphosyntax/__init__.py +0 -0
  70. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/morphosyntax/fr/case.py +0 -0
  71. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/pipeline.py +0 -0
  72. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/speaker/__init__.py +0 -0
  73. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/speaker/nemo_speaker.py +0 -0
  74. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utr/__init__.py +0 -0
  75. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utr/rev_utr.py +0 -0
  76. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utr/utils.py +0 -0
  77. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utr/whisper_utr.py +0 -0
  78. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utterance/__init__.py +0 -0
  79. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/__init__.py +0 -0
  80. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/conftest.py +0 -0
  81. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/chat/test_chat_file.py +0 -0
  82. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/chat/test_chat_generator.py +0 -0
  83. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/chat/test_chat_lexer.py +0 -0
  84. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/chat/test_chat_parser.py +0 -0
  85. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/chat/test_chat_utils.py +0 -0
  86. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/formats/textgrid/test_textgrid.py +0 -0
  87. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/analysis/test_eval.py +0 -0
  88. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/asr/test_asr_pipeline.py +0 -0
  89. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/asr/test_asr_utils.py +0 -0
  90. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/cleanup/test_disfluency.py +0 -0
  91. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/cleanup/test_parse_support.py +0 -0
  92. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/fa/test_fa_pipeline.py +0 -0
  93. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/fixures.py +0 -0
  94. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/test_pipeline.py +0 -0
  95. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/pipelines/test_pipeline_models.py +0 -0
  96. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/tests/test_document.py +0 -0
  97. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/utils/__init__.py +0 -0
  98. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/utils/config.py +0 -0
  99. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/utils/dp.py +0 -0
  100. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/utils/utils.py +0 -0
  101. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign.egg-info/SOURCES.txt +0 -0
  102. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign.egg-info/dependency_links.txt +0 -0
  103. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign.egg-info/entry_points.txt +0 -0
  104. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign.egg-info/requires.txt +0 -0
  105. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign.egg-info/top_level.txt +0 -0
  106. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/setup.cfg +0 -0
  107. {batchalign-0.7.3b3 → batchalign-0.7.3b5}/setup.py +0 -0
{batchalign-0.7.3b3/batchalign.egg-info → batchalign-0.7.3b5}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: batchalign
3
- Version: 0.7.3b3
3
+ Version: 0.7.3b5
4
4
  Summary: Python Speech Language Sample Analysis
5
5
  Author: Brian MacWhinney, Houjun Liu
6
6
  Author-email: macw@cmu.edu, houjun@cmu.edu
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/document.py RENAMED
@@ -263,6 +263,11 @@ class Utterance(BaseModel):
263
263
  if last_tok in ENDING_PUNCT + MOR_PUNCT:
264
264
  detokenized = detokenized.replace(last_tok, f" {last_tok}")
265
265
  detokenized = detokenized.replace(" ", " ")
266
+ detokenized = re.sub(r",(\w)", r", \1", detokenized)
267
+ detokenized = re.sub(r",.", r",", detokenized)
268
+ detokenized = re.sub(r".,", r",", detokenized)
269
+ detokenized = re.sub(r"\? !", r"!", detokenized)
270
+ detokenized = re.sub(r"! \?", r"?", detokenized)
266
271
 
267
272
  ## TODO deal with angle brackets for retraces
268
273
  # NOTE: we don't use detokenize here to put spaces
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/asr/utils.py RENAMED
@@ -75,6 +75,7 @@ def retokenize_with_engine(intermediate_output, engine):
75
75
  utterance = [i for i in utterance if i[0].strip() != ""]
76
76
 
77
77
  joined = " ".join([i[0] for i in utterance])
78
+ joined = joined.replace("。", ".")
78
79
  split = engine(joined)
79
80
 
80
81
  # align the utterance against original splits and generate final outputs
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/dispatch.py RENAMED
@@ -73,7 +73,7 @@ def dispatch_pipeline(pkg_str, lang, num_speakers=None, **arg_overrides):
73
73
  packages.append("disfluency")
74
74
  if "retracing" not in packages:
75
75
  packages.append("retracing")
76
- if "utterance" not in packages and resolve("utterance", lang) == None and lang in ["cho", "dan", "eng", "ind", "ita",
76
+ if "utterance" not in packages and resolve("utterance", lang) == None and lang in ["cho", "eng", "ind", "ita",
77
77
  "jpn", "por", "spa", "tur", "vie"]:
78
78
  packages.append("utterance")
79
79
  if "fa" in packages:
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/morphosyntax/ud.py RENAMED
@@ -686,7 +686,7 @@ def morphoanalyze(doc: Document, retokenize:bool, status_hook:callable = None, *
686
686
 
687
687
  elif not any([i in ["hr", "zh", "zh-hans", "zh-hant", "ja", "ko",
688
688
  "sl", "sr", "bg", "ru", "et", "hu",
689
- "eu", "el", "he", "af", "ga"] for i in lang]):
689
+ "eu", "el", "he", "af", "ga", "da"] for i in lang]):
690
690
  if "en" in lang:
691
691
  config["processors"]["mwt"] = "gum"
692
692
  else:
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/batchalign/pipelines/utterance/ud_utterance.py RENAMED
@@ -258,7 +258,7 @@ class StanzaUtteranceEngine(BatchalignEngine):
258
258
 
259
259
  elif not any([i in ["hr", "zh", "zh-hans", "zh-hant", "ja", "ko",
260
260
  "sl", "sr", "bg", "ru", "et", "hu",
261
- "eu", "el", "he", "af", "ga"] for i in lang]):
261
+ "eu", "el", "he", "af", "ga", "da"] for i in lang]):
262
262
  if "en" in lang:
263
263
  config["processors"]["mwt"] = "gum"
264
264
  else:
batchalign-0.7.3b5/batchalign/version ADDED
@@ -0,0 +1,3 @@
1
+ 0.7.3-beta.5
2
+ June 7, 2024
3
+ bug patches
{batchalign-0.7.3b3 → batchalign-0.7.3b5/batchalign.egg-info}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: batchalign
3
- Version: 0.7.3b3
3
+ Version: 0.7.3b5
4
4
  Summary: Python Speech Language Sample Analysis
5
5
  Author: Brian MacWhinney, Houjun Liu
6
6
  Author-email: macw@cmu.edu, houjun@cmu.edu
batchalign-0.7.3b3/batchalign/version DELETED
@@ -1,3 +0,0 @@
1
- 0.7.3-beta.3
2
- June 7, 2024
3
- more languages
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/LICENSE RENAMED
File without changes
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/MANIFEST.in RENAMED
File without changes
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/README.md RENAMED
File without changes
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/setup.cfg RENAMED
File without changes
{batchalign-0.7.3b3 → batchalign-0.7.3b5}/setup.py RENAMED
File without changes