BatchalignHK 0.7.20.post2__tar.gz → 0.7.20.post3__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 (162) hide show
  1. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/PKG-INFO +3 -2
  2. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/PKG-INFO +3 -2
  3. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/analysis/eval.py +5 -0
  4. batchalignhk-0.7.20.post3/batchalign/utils/abbrev.py +443 -0
  5. batchalignhk-0.7.20.post3/batchalign/version +3 -0
  6. batchalignhk-0.7.20.post2/batchalign/utils/abbrev.py +0 -182
  7. batchalignhk-0.7.20.post2/batchalign/version +0 -3
  8. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/SOURCES.txt +0 -0
  9. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/dependency_links.txt +0 -0
  10. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/entry_points.txt +0 -0
  11. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/requires.txt +0 -0
  12. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/BatchalignHK.egg-info/top_level.txt +0 -0
  13. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/LICENSE +0 -0
  14. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/MANIFEST.in +0 -0
  15. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/README.md +0 -0
  16. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/__init__.py +0 -0
  17. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/__main__.py +0 -0
  18. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/cli/__init__.py +0 -0
  19. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/cli/cli.py +0 -0
  20. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/cli/dispatch.py +0 -0
  21. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/constants.py +0 -0
  22. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/document.py +0 -0
  23. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/errors.py +0 -0
  24. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/__init__.py +0 -0
  25. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/core.py +0 -0
  26. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/exception.py +0 -0
  27. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/logging.py +0 -0
  28. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/realtime_meeting.py +0 -0
  29. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/speech_recognizer.py +0 -0
  30. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/speech_synthesizer.py +0 -0
  31. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/speech_transcriber.py +0 -0
  32. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/stream_input_tts.py +0 -0
  33. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/token.py +0 -0
  34. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/util.py +0 -0
  35. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/version.py +0 -0
  36. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/__init__.py +0 -0
  37. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_abnf.py +0 -0
  38. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_app.py +0 -0
  39. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_cookiejar.py +0 -0
  40. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_core.py +0 -0
  41. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_exceptions.py +0 -0
  42. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_handshake.py +0 -0
  43. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_http.py +0 -0
  44. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_logging.py +0 -0
  45. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_socket.py +0 -0
  46. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_ssl_compat.py +0 -0
  47. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_url.py +0 -0
  48. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/_utils.py +0 -0
  49. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/__init__.py +0 -0
  50. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/echo-server.py +0 -0
  51. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_abnf.py +0 -0
  52. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_app.py +0 -0
  53. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_cookiejar.py +0 -0
  54. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_http.py +0 -0
  55. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_url.py +0 -0
  56. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/extern/nls/websocket/tests/test_websocket.py +0 -0
  57. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/__init__.py +0 -0
  58. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/base.py +0 -0
  59. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/__init__.py +0 -0
  60. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/file.py +0 -0
  61. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/generator.py +0 -0
  62. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/lexer.py +0 -0
  63. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/parser.py +0 -0
  64. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/chat/utils.py +0 -0
  65. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/textgrid/__init__.py +0 -0
  66. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/textgrid/file.py +0 -0
  67. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/textgrid/generator.py +0 -0
  68. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/formats/textgrid/parser.py +0 -0
  69. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/__init__.py +0 -0
  70. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/resolve.py +0 -0
  71. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/speaker/__init__.py +0 -0
  72. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/speaker/config.yaml +0 -0
  73. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/speaker/infer.py +0 -0
  74. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/speaker/utils.py +0 -0
  75. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/training/__init__.py +0 -0
  76. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/training/run.py +0 -0
  77. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/training/utils.py +0 -0
  78. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utils.py +0 -0
  79. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/__init__.py +0 -0
  80. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/cantonese_infer.py +0 -0
  81. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/dataset.py +0 -0
  82. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/execute.py +0 -0
  83. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/infer.py +0 -0
  84. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/prep.py +0 -0
  85. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/utterance/train.py +0 -0
  86. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/wave2vec/__init__.py +0 -0
  87. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/wave2vec/infer_fa.py +0 -0
  88. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/whisper/__init__.py +0 -0
  89. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/whisper/infer_asr.py +0 -0
  90. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/models/whisper/infer_fa.py +0 -0
  91. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/__init__.py +0 -0
  92. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/analysis/__init__.py +0 -0
  93. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/__init__.py +0 -0
  94. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/aliyun.py +0 -0
  95. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/num2chinese.py +0 -0
  96. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/oai_whisper.py +0 -0
  97. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/rev.py +0 -0
  98. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/tencent.py +0 -0
  99. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/utils.py +0 -0
  100. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/whisper.py +0 -0
  101. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/asr/whisperx.py +0 -0
  102. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/base.py +0 -0
  103. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/__init__.py +0 -0
  104. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/cleanup.py +0 -0
  105. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/disfluencies.py +0 -0
  106. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/parse_support.py +0 -0
  107. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/retrace.py +0 -0
  108. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/support/filled_pauses.eng +0 -0
  109. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/support/replacements.eng +0 -0
  110. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/cleanup/support/test.test +0 -0
  111. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/dispatch.py +0 -0
  112. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/fa/__init__.py +0 -0
  113. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/fa/wave2vec_fa.py +0 -0
  114. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/fa/whisper_fa.py +0 -0
  115. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/__init__.py +0 -0
  116. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/coref.py +0 -0
  117. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/en/irr.py +0 -0
  118. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/fr/apm.py +0 -0
  119. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/fr/apmn.py +0 -0
  120. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/fr/case.py +0 -0
  121. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/ja/verbforms.py +0 -0
  122. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/morphosyntax/ud.py +0 -0
  123. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/pipeline.py +0 -0
  124. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/speaker/__init__.py +0 -0
  125. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/speaker/nemo_speaker.py +0 -0
  126. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/translate/__init__.py +0 -0
  127. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/translate/gtrans.py +0 -0
  128. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/translate/seamless.py +0 -0
  129. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/translate/utils.py +0 -0
  130. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utr/__init__.py +0 -0
  131. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utr/rev_utr.py +0 -0
  132. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utr/tencent_utr.py +0 -0
  133. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utr/utils.py +0 -0
  134. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utr/whisper_utr.py +0 -0
  135. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utterance/__init__.py +0 -0
  136. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/pipelines/utterance/ud_utterance.py +0 -0
  137. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/__init__.py +0 -0
  138. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/conftest.py +0 -0
  139. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/chat/test_chat_file.py +0 -0
  140. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/chat/test_chat_generator.py +0 -0
  141. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/chat/test_chat_lexer.py +0 -0
  142. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/chat/test_chat_parser.py +0 -0
  143. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/chat/test_chat_utils.py +0 -0
  144. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/formats/textgrid/test_textgrid.py +0 -0
  145. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/analysis/test_eval.py +0 -0
  146. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/asr/test_asr_pipeline.py +0 -0
  147. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/asr/test_asr_utils.py +0 -0
  148. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/cleanup/test_disfluency.py +0 -0
  149. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/cleanup/test_parse_support.py +0 -0
  150. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/fa/test_fa_pipeline.py +0 -0
  151. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/fixures.py +0 -0
  152. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/test_pipeline.py +0 -0
  153. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/pipelines/test_pipeline_models.py +0 -0
  154. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/tests/test_document.py +0 -0
  155. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/__init__.py +0 -0
  156. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/compounds.py +0 -0
  157. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/config.py +0 -0
  158. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/dp.py +0 -0
  159. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/names.py +0 -0
  160. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/batchalign/utils/utils.py +0 -0
  161. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/setup.cfg +0 -0
  162. {batchalignhk-0.7.20.post2 → batchalignhk-0.7.20.post3}/setup.py +0 -0
@@ -1,6 +1,6 @@
1
- Metadata-Version: 2.2
1
+ Metadata-Version: 2.4
2
2
  Name: BatchalignHK
3
- Version: 0.7.20.post2
3
+ Version: 0.7.20.post3
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
@@ -52,6 +52,7 @@ Dynamic: author-email
52
52
  Dynamic: classifier
53
53
  Dynamic: description
54
54
  Dynamic: description-content-type
55
+ Dynamic: license-file
55
56
  Dynamic: provides-extra
56
57
  Dynamic: requires-dist
57
58
  Dynamic: summary
@@ -1,6 +1,6 @@
1
- Metadata-Version: 2.2
1
+ Metadata-Version: 2.4
2
2
  Name: BatchalignHK
3
- Version: 0.7.20.post2
3
+ Version: 0.7.20.post3
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
@@ -52,6 +52,7 @@ Dynamic: author-email
52
52
  Dynamic: classifier
53
53
  Dynamic: description
54
54
  Dynamic: description-content-type
55
+ Dynamic: license-file
55
56
  Dynamic: provides-extra
56
57
  Dynamic: requires-dist
57
58
  Dynamic: summary
@@ -12,11 +12,13 @@ from batchalign.utils.config import config_read
12
12
  from batchalign.utils.dp import align, ExtraType, Extra, Match
13
13
  from batchalign.utils.names import names
14
14
  from batchalign.utils.compounds import compounds
15
+ from batchalign.utils.abbrev import abbrev
15
16
 
16
17
  import logging
17
18
  L = logging.getLogger("batchalign")
18
19
 
19
20
  joined_compounds = ["".join(k) for k in compounds]
21
+ lowered_abbrev = [k.lower() for k in abbrev]
20
22
 
21
23
  fillers = ["um", "uhm", "em", "mhm", "uhhm", "eh", "uh"]
22
24
  def conform(x):
@@ -25,6 +27,9 @@ def conform(x):
25
27
  if i.strip() in joined_compounds:
26
28
  for k in compounds[joined_compounds.index(i.strip())]:
27
29
  result.append(k)
30
+ elif i.strip() in lowered_abbrev:
31
+ for j in i.strip():
32
+ result.append(j)
28
33
  elif "'s" in i.strip():
29
34
  result.append(i.split("'")[0])
30
35
  result.append("is")
@@ -0,0 +1,443 @@
1
+ abbrev = [
2
+ "FBI",
3
+ "CIA",
4
+ "NSA",
5
+ "NATO",
6
+ "UN",
7
+ "WHO",
8
+ "NASA",
9
+ "CDC",
10
+ "IRS",
11
+ "EPA",
12
+ "HTTP",
13
+ "URL",
14
+ "HTML",
15
+ "CSS",
16
+ "API",
17
+ "IP",
18
+ "DNS",
19
+ "SQL",
20
+ "USB",
21
+ "VPN",
22
+ "ATT",
23
+ "AT&T",
24
+ "CEO",
25
+ "CFO",
26
+ "COO",
27
+ "IPO",
28
+ "ROI",
29
+ "GDP",
30
+ "LLC",
31
+ "HR",
32
+ "M&",
33
+ "KPI",
34
+ "GPA",
35
+ "SAT",
36
+ "ACT",
37
+ "MBA",
38
+ "PhD",
39
+ "BA",
40
+ "MA",
41
+ "STEM",
42
+ "ESL",
43
+ "GED",
44
+ "AWOL",
45
+ "MIA",
46
+ "POW",
47
+ "IED",
48
+ "UAV",
49
+ "RPG",
50
+ "NATO",
51
+ "SEAL",
52
+ "JAG",
53
+ "ROTC",
54
+ "CERN",
55
+ "GMO",
56
+ "H2O",
57
+ "CO2",
58
+ "UV",
59
+ "IR",
60
+ "AI",
61
+ "VR",
62
+ "AR",
63
+ "NPR",
64
+ "BBC",
65
+ "MTV",
66
+ "CNN",
67
+ "HBO",
68
+ "ESPN",
69
+ "TMZ",
70
+ "AMC",
71
+ "IMAX",
72
+ "WWE",
73
+ "ASAP",
74
+ "DIY",
75
+ "ETA",
76
+ "RSVP",
77
+ "FYI",
78
+ "LOL",
79
+ "BRB",
80
+ "IDK",
81
+ "BTW",
82
+ "TMI",
83
+ "PBJ",
84
+ "AIDS",
85
+ "HIV",
86
+ "ADHD",
87
+ "COPD",
88
+ "PTSD",
89
+ "CHF",
90
+ "CAD",
91
+ "TB",
92
+ "UTI",
93
+ "GERD",
94
+ "MRI",
95
+ "CT",
96
+ "ECG",
97
+ "EEG",
98
+ "CBC",
99
+ "BMP",
100
+ "ABG",
101
+ "PFT",
102
+ "FOBT",
103
+ "ENT",
104
+ "OB",
105
+ "PCP",
106
+ "ICU",
107
+ "NICU",
108
+ "ER",
109
+ "OR",
110
+ "PT",
111
+ "OT",
112
+ "EM",
113
+ "OTC",
114
+ "NSAID",
115
+ "IV",
116
+ "IM",
117
+ "SC",
118
+ "PRN",
119
+ "BID",
120
+ "TID",
121
+ "QID",
122
+ "NPO",
123
+ "CNS",
124
+ "PNS",
125
+ "GI",
126
+ "GU",
127
+ "CV",
128
+ "MSK",
129
+ "ENT",
130
+ "BMI",
131
+ "BMR",
132
+ "BP",
133
+ "WBC",
134
+ "RBC",
135
+ "HGB",
136
+ "HCT",
137
+ "PLT",
138
+ "ESR",
139
+ "CRP",
140
+ "LFT",
141
+ "TFT",
142
+ "INR",
143
+ "MMR",
144
+ "DPT",
145
+ "HPV",
146
+ "Tdap",
147
+ "BCG",
148
+ "IPV",
149
+ "HBV",
150
+ "HAV",
151
+ "HCV",
152
+ "RSV",
153
+ "SOAP",
154
+ "DNR",
155
+ "AMA",
156
+ "LOS",
157
+ "EHR",
158
+ "EMR",
159
+ "ICD",
160
+ "CPT",
161
+ "HIPAA",
162
+ "HR",
163
+ "RR",
164
+ "SpO2",
165
+ "MAP",
166
+ "GFR",
167
+ "A1C",
168
+ "LDL",
169
+ "HDL",
170
+ "TG",
171
+ "BUN",
172
+ "SIDS",
173
+ "DVT",
174
+ "PE",
175
+ "ARDS",
176
+ "SLE",
177
+ "RA",
178
+ "TIA",
179
+ "CVA",
180
+ "ALS",
181
+ "MS",
182
+ "AC",
183
+ "AM",
184
+ "BC",
185
+ "BFF",
186
+ "BO",
187
+ "COD",
188
+ "DIY",
189
+ "EG",
190
+ "ETA",
191
+ "FYI",
192
+ "ID",
193
+ "IE",
194
+ "JK",
195
+ "KO",
196
+ "MIA",
197
+ "NA",
198
+ "NB",
199
+ "OK",
200
+ "PS",
201
+ "RSVP",
202
+ "API",
203
+ "B2B",
204
+ "B2C",
205
+ "CPU",
206
+ "DNS",
207
+ "HTML",
208
+ "HTTP",
209
+ "HTTPS",
210
+ "IP",
211
+ "ISP",
212
+ "PDF",
213
+ "RAM",
214
+ "ROM",
215
+ "SaaS",
216
+ "URL",
217
+ "VPN",
218
+ "WLAN",
219
+ "XML",
220
+ "SSD",
221
+ "AR",
222
+ "AP",
223
+ "CEO",
224
+ "CFO",
225
+ "COO",
226
+ "EOY",
227
+ "GM",
228
+ "IPO",
229
+ "KPI",
230
+ "MA",
231
+ "PL",
232
+ "ROI",
233
+ "SOP",
234
+ "TBD",
235
+ "TBA",
236
+ "VC",
237
+ "WB",
238
+ "WIP",
239
+ "YOY",
240
+ "BOE",
241
+ "AF",
242
+ "CIA",
243
+ "DOD",
244
+ "DEA",
245
+ "FBI",
246
+ "GOP",
247
+ "NATO",
248
+ "NSA",
249
+ "UN",
250
+ "US",
251
+ "VA",
252
+ "FEMA",
253
+ "SWAT",
254
+ "ATF",
255
+ "IRS",
256
+ "DHS",
257
+ "CDC",
258
+ "MIT",
259
+ "NASA",
260
+ "FEC",
261
+ "AIDS",
262
+ "BPM",
263
+ "CPR",
264
+ "CT",
265
+ "DNA",
266
+ "EBV",
267
+ "HIV",
268
+ "ICU",
269
+ "MRI",
270
+ "NPO",
271
+ "OTC",
272
+ "Rx",
273
+ "SARS",
274
+ "STD",
275
+ "TBA",
276
+ "WBC",
277
+ "WHO",
278
+ "BP",
279
+ "CDC",
280
+ "EMT",
281
+ "AS",
282
+ "BA",
283
+ "BS",
284
+ "GPA",
285
+ "MA",
286
+ "MS",
287
+ "PhD",
288
+ "SAT",
289
+ "ACT",
290
+ "MOOC",
291
+ "TA",
292
+ "ESL",
293
+ "GED",
294
+ "NCLB",
295
+ "SOP",
296
+ "BEd",
297
+ "CPA",
298
+ "DO",
299
+ "HS",
300
+ "CO2",
301
+ "H2O",
302
+ "O2",
303
+ "UV",
304
+ "GMO",
305
+ "NPO",
306
+ "EPA",
307
+ "N2",
308
+ "HCl",
309
+ "NaCl",
310
+ "pH",
311
+ "DNA",
312
+ "RNA",
313
+ "NMR",
314
+ "PCR",
315
+ "FOSS",
316
+ "STEM",
317
+ "Toxic",
318
+ "VOCs",
319
+ "BOD",
320
+ "ASAP",
321
+ "CEO",
322
+ "CFO",
323
+ "DIY",
324
+ "FOMO",
325
+ "HOA",
326
+ "IOU",
327
+ "LOL",
328
+ "MOU",
329
+ "NGO",
330
+ "PO",
331
+ "RIP",
332
+ "SOS",
333
+ "TMI",
334
+ "VIP",
335
+ "WTF",
336
+ "YOLO",
337
+ "TLDR",
338
+ "SNAFU",
339
+ "BTW",
340
+ "AFK",
341
+ "BTO",
342
+ "CSF",
343
+ "DS",
344
+ "ER",
345
+ "FAQ",
346
+ "GOAT",
347
+ "HR",
348
+ "IT",
349
+ "KPI",
350
+ "LTD",
351
+ "MO",
352
+ "NDA",
353
+ "OS",
354
+ "PM",
355
+ "QED",
356
+ "RD",
357
+ "SWOT",
358
+ "TOS",
359
+ "USA",
360
+ "VO",
361
+ "WA",
362
+ "XO",
363
+ "YT",
364
+ "ZA",
365
+ "AAA",
366
+ "BBC",
367
+ "CBO",
368
+ "DBA",
369
+ "EU",
370
+ "FDA",
371
+ "GDP",
372
+ "HALO",
373
+ "ICU",
374
+ "JD",
375
+ "KISS",
376
+ "LOA",
377
+ "MOU",
378
+ "NHS",
379
+ "OP",
380
+ "PA",
381
+ "QA",
382
+ "ROE",
383
+ "SLA",
384
+ "TC",
385
+ "UK",
386
+ "VAT",
387
+ "WTO",
388
+ "XR",
389
+ "YMCA",
390
+ "ZONE",
391
+ "AC",
392
+ "BLM",
393
+ "CLI",
394
+ "DSL",
395
+ "ERP",
396
+ "FAA",
397
+ "GPS",
398
+ "HTML",
399
+ "IPA",
400
+ "JS",
401
+ "KNOW",
402
+ "LED",
403
+ "MAC",
404
+ "NTP",
405
+ "OT",
406
+ "PC",
407
+ "QR",
408
+ "RA",
409
+ "SAS",
410
+ "TRA",
411
+ "USP",
412
+ "VCR",
413
+ "WASP",
414
+ "XYZ",
415
+ "YOY",
416
+ "ZIP",
417
+ "AED",
418
+ "BAS",
419
+ "CAD",
420
+ "DA",
421
+ "ECG",
422
+ "FIFO",
423
+ "GOP",
424
+ "HIV",
425
+ "ICT",
426
+ "JAB",
427
+ "KAP",
428
+ "LEA",
429
+ "MAS",
430
+ "NED",
431
+ "OCD",
432
+ "PBR",
433
+ "QT",
434
+ "RAD",
435
+ "SAD",
436
+ "TOEFL",
437
+ "UNC",
438
+ "VAN",
439
+ "WC",
440
+ "XFS",
441
+ "YOP",
442
+ "ZAF"
443
+ ]
@@ -0,0 +1,3 @@
1
+ 0.7.20-post.3
2
+ July 21th, 2025
3
+ Some benchmarking abbreviations
@@ -1,182 +0,0 @@
1
- abbrev = [
2
- "FBI",
3
- "CIA",
4
- "NSA",
5
- "NATO",
6
- "UN",
7
- "WHO",
8
- "NASA",
9
- "CDC",
10
- "IRS",
11
- "EPA",
12
- "HTTP",
13
- "URL",
14
- "HTML",
15
- "CSS",
16
- "API",
17
- "IP",
18
- "DNS",
19
- "SQL",
20
- "USB",
21
- "VPN",
22
- "ATT",
23
- "AT&T",
24
- "CEO",
25
- "CFO",
26
- "COO",
27
- "IPO",
28
- "ROI",
29
- "GDP",
30
- "LLC",
31
- "HR",
32
- "M&",
33
- "KPI",
34
- "GPA",
35
- "SAT",
36
- "ACT",
37
- "MBA",
38
- "PhD",
39
- "BA",
40
- "MA",
41
- "STEM",
42
- "ESL",
43
- "GED",
44
- "AWOL",
45
- "MIA",
46
- "POW",
47
- "IED",
48
- "UAV",
49
- "RPG",
50
- "NATO",
51
- "SEAL",
52
- "JAG",
53
- "ROTC",
54
- "CERN",
55
- "GMO",
56
- "H2O",
57
- "CO2",
58
- "UV",
59
- "IR",
60
- "AI",
61
- "VR",
62
- "AR",
63
- "NPR",
64
- "BBC",
65
- "MTV",
66
- "CNN",
67
- "HBO",
68
- "ESPN",
69
- "TMZ",
70
- "AMC",
71
- "IMAX",
72
- "WWE",
73
- "ASAP",
74
- "DIY",
75
- "ETA",
76
- "RSVP",
77
- "FYI",
78
- "LOL",
79
- "BRB",
80
- "IDK",
81
- "BTW",
82
- "TMI",
83
- "PBJ",
84
- "AIDS",
85
- "HIV",
86
- "ADHD",
87
- "COPD",
88
- "PTSD",
89
- "CHF",
90
- "CAD",
91
- "TB",
92
- "UTI",
93
- "GERD",
94
- "MRI",
95
- "CT",
96
- "ECG",
97
- "EEG",
98
- "CBC",
99
- "BMP",
100
- "ABG",
101
- "PFT",
102
- "FOBT",
103
- "ENT",
104
- "OB",
105
- "PCP",
106
- "ICU",
107
- "NICU",
108
- "ER",
109
- "OR",
110
- "PT",
111
- "OT",
112
- "EM",
113
- "OTC",
114
- "NSAID",
115
- "IV",
116
- "IM",
117
- "SC",
118
- "PRN",
119
- "BID",
120
- "TID",
121
- "QID",
122
- "NPO",
123
- "CNS",
124
- "PNS",
125
- "GI",
126
- "GU",
127
- "CV",
128
- "MSK",
129
- "ENT",
130
- "BMI",
131
- "BMR",
132
- "BP",
133
- "WBC",
134
- "RBC",
135
- "HGB",
136
- "HCT",
137
- "PLT",
138
- "ESR",
139
- "CRP",
140
- "LFT",
141
- "TFT",
142
- "INR",
143
- "MMR",
144
- "DPT",
145
- "HPV",
146
- "Tdap",
147
- "BCG",
148
- "IPV",
149
- "HBV",
150
- "HAV",
151
- "HCV",
152
- "RSV",
153
- "SOAP",
154
- "DNR",
155
- "AMA",
156
- "LOS",
157
- "EHR",
158
- "EMR",
159
- "ICD",
160
- "CPT",
161
- "HIPAA",
162
- "HR",
163
- "RR",
164
- "SpO2",
165
- "MAP",
166
- "GFR",
167
- "A1C",
168
- "LDL",
169
- "HDL",
170
- "TG",
171
- "BUN",
172
- "SIDS",
173
- "DVT",
174
- "PE",
175
- "ARDS",
176
- "SLE",
177
- "RA",
178
- "TIA",
179
- "CVA",
180
- "ALS",
181
- "MS",
182
- ]
@@ -1,3 +0,0 @@
1
- 0.7.20-post.2
2
- July 21th, 2025
3
- Some benchmarking anonyms