ok

app.py

@@ -68,8 +68,8 @@ def process_audio(audio_path):
     model = WhisperModel("large-v3", device="cpu", compute_type="int8")
 
     # Ejecutar la diarización y la transcripción
-    diarization = pipeline(audio_path, generate_kwargs={"language": "spanish"})
-    segments, info = model.transcribe(audio_path)
+    diarization = pipeline(audio_path)
+    segments, info = model.transcribe(audio_path, vad_filter=True, language="Spanish")
 
     # Asignar hablantes a segmentos de transcripción
     return assign_speakers_to_segments(diarization, segments)