batchalign 0.8.2.post13__tar.gz → 0.8.2.post14__tar.gz

{batchalign-0.8.2.post13/batchalign.egg-info → batchalign-0.8.2.post14}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: batchalign
-Version: 0.8.2.post13
+Version: 0.8.2.post14
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu
@@ -19,15 +19,15 @@ Requires-Dist: platformdirs>=4.3.0
 Requires-Dist: filelock>=3.0.0
 Requires-Dist: nltk>=3.8
 Requires-Dist: praatio<6.1.0,>=6.0.0
-Requires-Dist: torch>=2.10.0
-Requires-Dist: torchaudio>=2.10.0
+Requires-Dist: torch<2.9,>=2.7
+Requires-Dist: torchaudio<2.9,>=2.7
 Requires-Dist: soundfile>=0.12.0
 Requires-Dist: pydub
 Requires-Dist: plotly>=5.3.0
-Requires-Dist: transformers>=4.38.2
+Requires-Dist: transformers<5,>=4.38.2
 Requires-Dist: tokenizers>=0.14.1
 Requires-Dist: numba>=0.61.0
-Requires-Dist: numpy<=2.2
+Requires-Dist: numpy<2.3,>=1.24
 Requires-Dist: pycountry>=22.3
 Requires-Dist: stanza[transformers]>=1.10.1
 Requires-Dist: scipy~=1.11
@@ -50,7 +50,7 @@ Requires-Dist: openai-whisper
 Requires-Dist: llvmlite>=0.44.0
 Requires-Dist: praat-parselmouth==0.4.6
 Requires-Dist: opensmile>=2.5.0
-Requires-Dist: pyannote.audio
+Requires-Dist: pyannote.audio<4,>=3.4.0
 Requires-Dist: onnxruntime
 Requires-Dist: certifi>=2025.10.5
 Provides-Extra: dev
@@ -63,6 +63,16 @@ Requires-Dist: accelerate~=0.27; extra == "train"
 Provides-Extra: docs
 Requires-Dist: mkdocs-material; extra == "docs"
 Requires-Dist: mkdocs-click; extra == "docs"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # TalkBank | Batchalign2
 

{batchalign-0.8.2.post13 → batchalign-0.8.2.post14}/batchalign/pipelines/diarization/pyannote.py

@@ -47,9 +47,25 @@ class PyannoteEngine(BatchalignEngine):
 
     def process(self, doc: Document, **kwargs) -> Document:
         if self.pipe is None:
-            from pyannote.audio import Pipeline
-            self.pipe = Pipeline.from_pretrained("talkbank/dia-fork")
+            # torch>=2.6 defaults torch.load(weights_only=True), which rejects
+            # pyannote 3.x checkpoints that pickle TorchVersion/Specifications/etc.
+            # The pyannote checkpoint comes from a trusted HF repo, so force
+            # weights_only=False for the duration of the pipeline load.
+            import torch
+            _orig_torch_load = torch.load
+            def _patched_load(*args, **kwargs):
+                kwargs["weights_only"] = False
+                return _orig_torch_load(*args, **kwargs)
+            torch.load = _patched_load
+            try:
+                from pyannote.audio import Pipeline
+                self.pipe = Pipeline.from_pretrained("talkbank/dia-fork")
+            finally:
+                torch.load = _orig_torch_load
         assert doc.media != None and doc.media.url != None, f"We cannot diarize something that doesn't have a media path! Provided media tier='{doc.media}'"
+        if not doc.tiers:
+            L.warning("Skipping diarization: upstream ASR produced no tiers/utterances.")
+            return doc
         res = self.pipe(doc.media.url, num_speakers=self.num_speakers)
 
         speakers = list(set([int(i[-1].split("_")[-1])

batchalign-0.8.2.post14/batchalign/version

@@ -0,0 +1,3 @@
+0.8.2-post.14
+April 24 2026
+pyannote.audio/torchaudio version pin fix for diarization

{batchalign-0.8.2.post13 → batchalign-0.8.2.post14/batchalign.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: batchalign
-Version: 0.8.2.post13
+Version: 0.8.2.post14
 Summary: Python Speech Language Sample Analysis
 Author: Brian MacWhinney, Houjun Liu
 Author-email: macw@cmu.edu, houjun@cmu.edu
@@ -19,15 +19,15 @@ Requires-Dist: platformdirs>=4.3.0
 Requires-Dist: filelock>=3.0.0
 Requires-Dist: nltk>=3.8
 Requires-Dist: praatio<6.1.0,>=6.0.0
-Requires-Dist: torch>=2.10.0
-Requires-Dist: torchaudio>=2.10.0
+Requires-Dist: torch<2.9,>=2.7
+Requires-Dist: torchaudio<2.9,>=2.7
 Requires-Dist: soundfile>=0.12.0
 Requires-Dist: pydub
 Requires-Dist: plotly>=5.3.0
-Requires-Dist: transformers>=4.38.2
+Requires-Dist: transformers<5,>=4.38.2
 Requires-Dist: tokenizers>=0.14.1
 Requires-Dist: numba>=0.61.0
-Requires-Dist: numpy<=2.2
+Requires-Dist: numpy<2.3,>=1.24
 Requires-Dist: pycountry>=22.3
 Requires-Dist: stanza[transformers]>=1.10.1
 Requires-Dist: scipy~=1.11
@@ -50,7 +50,7 @@ Requires-Dist: openai-whisper
 Requires-Dist: llvmlite>=0.44.0
 Requires-Dist: praat-parselmouth==0.4.6
 Requires-Dist: opensmile>=2.5.0
-Requires-Dist: pyannote.audio
+Requires-Dist: pyannote.audio<4,>=3.4.0
 Requires-Dist: onnxruntime
 Requires-Dist: certifi>=2025.10.5
 Provides-Extra: dev
@@ -63,6 +63,16 @@ Requires-Dist: accelerate~=0.27; extra == "train"
 Provides-Extra: docs
 Requires-Dist: mkdocs-material; extra == "docs"
 Requires-Dist: mkdocs-click; extra == "docs"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 # TalkBank | Batchalign2
 

{batchalign-0.8.2.post13 → batchalign-0.8.2.post14}/batchalign.egg-info/requires.txt

@@ -3,15 +3,15 @@ platformdirs>=4.3.0
 filelock>=3.0.0
 nltk>=3.8
 praatio<6.1.0,>=6.0.0
-torch>=2.10.0
-torchaudio>=2.10.0
+torch<2.9,>=2.7
+torchaudio<2.9,>=2.7
 soundfile>=0.12.0
 pydub
 plotly>=5.3.0
-transformers>=4.38.2
+transformers<5,>=4.38.2
 tokenizers>=0.14.1
 numba>=0.61.0
-numpy<=2.2
+numpy<2.3,>=1.24
 pycountry>=22.3
 stanza[transformers]>=1.10.1
 scipy~=1.11
@@ -34,7 +34,7 @@ openai-whisper
 llvmlite>=0.44.0
 praat-parselmouth==0.4.6
 opensmile>=2.5.0
-pyannote.audio
+pyannote.audio<4,>=3.4.0
 onnxruntime
 certifi>=2025.10.5
 

{batchalign-0.8.2.post13 → batchalign-0.8.2.post14}/setup.py

@@ -31,8 +31,8 @@ setup(
         "filelock>=3.0.0",
         "nltk>=3.8",
         "praatio>=6.0.0,<6.1.0",
-        "torch>=2.10.0",
-        "torchaudio>=2.10.0",  # Only uses transforms/functional/pipelines (I/O via soundfile)
+        "torch>=2.7,<2.9",
+        "torchaudio>=2.7,<2.9",  # <2.9: keep list_audio_backends; <2.10: keep AudioMetaData — needed by pyannote.audio 3.x (torch.load weights_only patched in batchalign.pipelines.diarization.pyannote)
         "soundfile>=0.12.0",  # Audio I/O backend (replaces torchaudio.load/save/info)
         # "pyAudioAnalysis",
         # "hmmlearn==0.3.0",
@@ -40,10 +40,10 @@ setup(
         "pydub",
         # "imblearn",
         "plotly>=5.3.0",
-        "transformers>=4.38.2",
+        "transformers>=4.38.2,<5",
         "tokenizers>=0.14.1",
         "numba>=0.61.0",
-        "numpy<=2.2",
+        "numpy>=1.24,<2.3",
         "pycountry>=22.3",
         "stanza[transformers]>=1.10.1",
         "scipy~=1.11",
@@ -66,7 +66,7 @@ setup(
         "llvmlite>=0.44.0",
         "praat-parselmouth==0.4.6",  
         "opensmile>=2.5.0",
-        "pyannote.audio",
+        "pyannote.audio>=3.4.0,<4",
         "onnxruntime",
         "certifi>=2025.10.5"
     ],

batchalign-0.8.2.post13/batchalign/version

@@ -1,3 +0,0 @@
-0.8.2-post.13
-April 17 2026
-compare update, again