batchalign 0.7.19.post5__tar.gz → 0.7.19.post7__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.

Potentially problematic release.

This version of batchalign might be problematic. Click here for more details.

Files changed (124) hide show
  1. {batchalign-0.7.19.post5/batchalign.egg-info → batchalign-0.7.19.post7}/PKG-INFO +1 -1
  2. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/cli/cli.py +3 -2
  3. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/retrace.py +1 -1
  4. batchalign-0.7.19.post7/batchalign/version +3 -0
  5. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7/batchalign.egg-info}/PKG-INFO +1 -1
  6. batchalign-0.7.19.post5/batchalign/version +0 -3
  7. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/LICENSE +0 -0
  8. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/MANIFEST.in +0 -0
  9. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/README.md +0 -0
  10. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/__init__.py +0 -0
  11. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/__main__.py +0 -0
  12. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/cli/__init__.py +0 -0
  13. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/cli/dispatch.py +0 -0
  14. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/constants.py +0 -0
  15. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/document.py +0 -0
  16. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/errors.py +0 -0
  17. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/__init__.py +0 -0
  18. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/base.py +0 -0
  19. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/__init__.py +0 -0
  20. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/file.py +0 -0
  21. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/generator.py +0 -0
  22. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/lexer.py +0 -0
  23. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/parser.py +0 -0
  24. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/chat/utils.py +0 -0
  25. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/textgrid/__init__.py +0 -0
  26. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/textgrid/file.py +0 -0
  27. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/textgrid/generator.py +0 -0
  28. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/formats/textgrid/parser.py +0 -0
  29. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/__init__.py +0 -0
  30. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/resolve.py +0 -0
  31. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/speaker/__init__.py +0 -0
  32. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/speaker/config.yaml +0 -0
  33. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/speaker/infer.py +0 -0
  34. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/speaker/utils.py +0 -0
  35. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/training/__init__.py +0 -0
  36. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/training/run.py +0 -0
  37. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/training/utils.py +0 -0
  38. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utils.py +0 -0
  39. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/__init__.py +0 -0
  40. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/cantonese_infer.py +0 -0
  41. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/dataset.py +0 -0
  42. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/execute.py +0 -0
  43. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/infer.py +0 -0
  44. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/prep.py +0 -0
  45. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/utterance/train.py +0 -0
  46. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/wave2vec/__init__.py +0 -0
  47. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/wave2vec/infer_fa.py +0 -0
  48. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/whisper/__init__.py +0 -0
  49. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/whisper/infer_asr.py +0 -0
  50. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/models/whisper/infer_fa.py +0 -0
  51. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/__init__.py +0 -0
  52. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/analysis/__init__.py +0 -0
  53. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/analysis/eval.py +0 -0
  54. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/__init__.py +0 -0
  55. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/num2chinese.py +0 -0
  56. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/oai_whisper.py +0 -0
  57. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/rev.py +0 -0
  58. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/utils.py +0 -0
  59. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/whisper.py +0 -0
  60. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/asr/whisperx.py +0 -0
  61. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/base.py +0 -0
  62. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/__init__.py +0 -0
  63. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/cleanup.py +0 -0
  64. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/disfluencies.py +0 -0
  65. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/parse_support.py +0 -0
  66. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/support/filled_pauses.eng +0 -0
  67. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/support/replacements.eng +0 -0
  68. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/support/test.test +0 -0
  69. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/dispatch.py +0 -0
  70. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/fa/__init__.py +0 -0
  71. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/fa/wave2vec_fa.py +0 -0
  72. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/fa/whisper_fa.py +0 -0
  73. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/__init__.py +0 -0
  74. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/coref.py +0 -0
  75. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/en/irr.py +0 -0
  76. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/fr/apm.py +0 -0
  77. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/fr/apmn.py +0 -0
  78. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/fr/case.py +0 -0
  79. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/ja/verbforms.py +0 -0
  80. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/morphosyntax/ud.py +0 -0
  81. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/pipeline.py +0 -0
  82. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/speaker/__init__.py +0 -0
  83. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/speaker/nemo_speaker.py +0 -0
  84. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/translate/__init__.py +0 -0
  85. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/translate/gtrans.py +0 -0
  86. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/translate/seamless.py +0 -0
  87. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/translate/utils.py +0 -0
  88. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utr/__init__.py +0 -0
  89. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utr/rev_utr.py +0 -0
  90. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utr/utils.py +0 -0
  91. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utr/whisper_utr.py +0 -0
  92. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utterance/__init__.py +0 -0
  93. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/utterance/ud_utterance.py +0 -0
  94. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/__init__.py +0 -0
  95. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/conftest.py +0 -0
  96. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/chat/test_chat_file.py +0 -0
  97. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/chat/test_chat_generator.py +0 -0
  98. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/chat/test_chat_lexer.py +0 -0
  99. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/chat/test_chat_parser.py +0 -0
  100. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/chat/test_chat_utils.py +0 -0
  101. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/formats/textgrid/test_textgrid.py +0 -0
  102. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/analysis/test_eval.py +0 -0
  103. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/asr/test_asr_pipeline.py +0 -0
  104. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/asr/test_asr_utils.py +0 -0
  105. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/cleanup/test_disfluency.py +0 -0
  106. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/cleanup/test_parse_support.py +0 -0
  107. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/fa/test_fa_pipeline.py +0 -0
  108. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/fixures.py +0 -0
  109. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/test_pipeline.py +0 -0
  110. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/pipelines/test_pipeline_models.py +0 -0
  111. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/tests/test_document.py +0 -0
  112. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/__init__.py +0 -0
  113. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/abbrev.py +0 -0
  114. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/config.py +0 -0
  115. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/dp.py +0 -0
  116. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/names.py +0 -0
  117. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/utils/utils.py +0 -0
  118. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign.egg-info/SOURCES.txt +0 -0
  119. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign.egg-info/dependency_links.txt +0 -0
  120. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign.egg-info/entry_points.txt +0 -0
  121. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign.egg-info/requires.txt +0 -0
  122. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign.egg-info/top_level.txt +0 -0
  123. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/setup.cfg +0 -0
  124. {batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/setup.py +0 -0
{batchalign-0.7.19.post5/batchalign.egg-info → batchalign-0.7.19.post7}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.2
2
2
  Name: batchalign
3
- Version: 0.7.19.post5
3
+ Version: 0.7.19.post7
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-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/cli/cli.py RENAMED
@@ -111,7 +111,8 @@ batchalign.add_command(train, "models")
111
111
  @click.option("--wav2vec/--whisper_fa",
112
112
  default=True, help="Use Whisper instead of Wav2Vec for English (defaults for Whisper for non-English)")
113
113
  @click.option("--pauses", type=bool, default=False, help="Should we try to bullet each word or should we try to add pauses in between words by grouping them? Default: no pauses.", is_flag=True)
114
-
114
+ @click.option("--wor/--nowor",
115
+ default=True, help="Should we write word level alignment line? Default to yes.")
115
116
  @click.pass_context
116
117
  def align(ctx, in_dir, out_dir, whisper, wav2vec, **kwargs):
117
118
  """Align transcripts against corresponding media files."""
@@ -122,7 +123,7 @@ def align(ctx, in_dir, out_dir, whisper, wav2vec, **kwargs):
122
123
  )
123
124
 
124
125
  def writer(doc, output):
125
- CHATFile(doc=doc).write(output)
126
+ CHATFile(doc=doc).write(output, write_wor=kwargs.get("wor", True))
126
127
 
127
128
  if not wav2vec:
128
129
  _dispatch("align", "eng", 1,
{batchalign-0.7.19.post5 → batchalign-0.7.19.post7}/batchalign/pipelines/cleanup/retrace.py RENAMED
@@ -22,7 +22,7 @@ class NgramRetraceEngine(BatchalignEngine):
22
22
  if i.type in [TokenType.REGULAR, TokenType.PUNCT, TokenType.FP]:
23
23
  content.append(i)
24
24
  # scan for n-gram retraces
25
- for n in range(1, len(content)):
25
+ for n in range(1 if "yue" not in doc.langs and "zho" not in doc.langs else 2, len(content)):
26
26
  begin = 0
27
27
  while begin < len(content)-(n):
28
28
  # get the n gram info; we convert it to
batchalign-0.7.19.post7/batchalign/version ADDED
@@ -0,0 +1,3 @@
1
+ 0.7.19-post.7
2
+ May 23th, 2025
3
+ don't write wor on certain cases
{batchalign-0.7.19.post5 → batchalign-0.7.19.post7/batchalign.egg-info}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.2
2
2
  Name: batchalign
3
- Version: 0.7.19.post5
3
+ Version: 0.7.19.post7
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-0.7.19.post5/batchalign/version DELETED
@@ -1,3 +0,0 @@
1
- 0.7.19-post.5
2
- May 23th, 2025
3
- abbreviations