1292 fix speech dependencies (#1302)

.gitignore

@@ -11,6 +11,7 @@ coverage.xml
 .idea/
 .venv/
 venv/
+.claude/
 
 # whisper_cpp
 whisper_cpp

Buzz.spec

@@ -22,6 +22,19 @@ datas += copy_metadata("tokenizers")
 datas += copy_metadata("huggingface-hub")
 datas += copy_metadata("safetensors")
 datas += copy_metadata("pyyaml")
+datas += copy_metadata("julius")
+datas += copy_metadata("openunmix")
+datas += copy_metadata("lameenc")
+datas += copy_metadata("diffq")
+datas += copy_metadata("einops")
+datas += copy_metadata("hydra-core")
+datas += copy_metadata("hydra-colorlog")
+datas += copy_metadata("museval")
+datas += copy_metadata("submitit")
+datas += copy_metadata("treetable")
+datas += copy_metadata("soundfile")
+datas += copy_metadata("dora-search")
+datas += copy_metadata("lhotse")
 
 # Allow transformers package to load __init__.py file dynamically:
 # https://github.com/chidiwilliams/buzz/issues/272
@@ -92,7 +105,22 @@ a = Analysis(
     pathex=[],
     binaries=binaries,
     datas=datas,
-    hiddenimports=[],
+    hiddenimports=[
+        "dora", "dora.log",
+        "julius", "julius.core", "julius.resample",
+        "openunmix", "openunmix.filtering",
+        "lameenc",
+        "diffq",
+        "einops",
+        "hydra", "hydra.core", "hydra.core.global_hydra",
+        "hydra_colorlog",
+        "museval",
+        "submitit",
+        "treetable",
+        "soundfile",
+        "_soundfile_data",
+        "lhotse",
+    ],
     hookspath=[],
     hooksconfig={},
     runtime_hooks=[],

buzz/file_transcriber_queue_worker.py

@@ -1,12 +1,42 @@
 import logging
 import multiprocessing
 import queue
+import sys
 from pathlib import Path
 from typing import Optional, Tuple, List, Set
 from uuid import UUID
 
 from PyQt6.QtCore import QObject, QThread, pyqtSignal, pyqtSlot
 
+# Patch subprocess for demucs to prevent console windows on Windows
+if sys.platform == "win32":
+    import subprocess
+    _original_run = subprocess.run
+    _original_check_output = subprocess.check_output
+
+    def _patched_run(*args, **kwargs):
+        if 'startupinfo' not in kwargs:
+            si = subprocess.STARTUPINFO()
+            si.dwFlags |= subprocess.STARTF_USESHOWWINDOW
+            si.wShowWindow = subprocess.SW_HIDE
+            kwargs['startupinfo'] = si
+        if 'creationflags' not in kwargs:
+            kwargs['creationflags'] = subprocess.CREATE_NO_WINDOW
+        return _original_run(*args, **kwargs)
+
+    def _patched_check_output(*args, **kwargs):
+        if 'startupinfo' not in kwargs:
+            si = subprocess.STARTUPINFO()
+            si.dwFlags |= subprocess.STARTF_USESHOWWINDOW
+            si.wShowWindow = subprocess.SW_HIDE
+            kwargs['startupinfo'] = si
+        if 'creationflags' not in kwargs:
+            kwargs['creationflags'] = subprocess.CREATE_NO_WINDOW
+        return _original_check_output(*args, **kwargs)
+
+    subprocess.run = _patched_run
+    subprocess.check_output = _patched_check_output
+
 from demucs import api as demucsApi
 
 from buzz.model_loader import ModelType
@@ -95,6 +125,7 @@ class FileTranscriberQueueWorker(QObject):
                 logging.error(f"Error during speech extraction: {e}", exc_info=True)
 
         logging.debug("Starting next transcription task")
+        self.task_progress.emit(self.current_task, 0)
 
         model_type = self.current_task.transcription_options.model.model_type
         if model_type == ModelType.OPEN_AI_WHISPER_API: