Merge branch 'development' of gitea.centx.de:lelo/bethaus-app into development

requirements_transcription.txt

@@ -1,5 +1,8 @@
+pyaml
+librosa
 openai-whisper
 #https://pytorch.org/get-started/locally/
-torch==2.5.1
-torchvision==0.20.1
-torchaudio==2.5.1 --index-url https://download.pytorch.org/whl/cu124
+--extra-index-url https://download.pytorch.org/whl/cu128
+torch
+torchvision
+torchaudio

transcribe_all.py

@@ -19,6 +19,23 @@ with open("transcription_config.yml", "r", encoding="utf-8") as file:
     settings = yaml.safe_load(file)
     folder_list = settings.get("folder_list")
     model_name = settings.get("model_name")
+    gpu_only = settings.get("gpu_only", False)
+
+print("PyTorch version:", torch.__version__)
+print("CUDA available?", torch.cuda.is_available())
+print("CUDA version:", torch.version.cuda)
+print("GPU count:", torch.cuda.device_count())
+if torch.cuda.is_available():
+    for i in range(torch.cuda.device_count()):
+        print(f"  Device {i}:", torch.cuda.get_device_name(i))
+
+if not folder_list or not model_name:
+    print("Error: Please check the transcription_config.yml file. It should contain 'folder_list' and 'model_name'.")
+    sys.exit(1)
+
+if gpu_only and not torch.cuda.is_available():
+    print("Error: You requested to only use GPU but it is not available. Please check your PyTorch installation.")
+    sys.exit(1)
 
 def load_audio_librosa(path: str, sr: int = 16_000) -> np.ndarray:
     audio, orig_sr = librosa.load(path, sr=sr)      # load + resample to 16 kHz