Update diarization.py

diarization.py

@@ -51,14 +51,14 @@ def get_diarization_pipeline(hf_token: Optional[str] = None):
         return None
 
 
-def diarize_audio(audio_path: str, pipeline, num_speakers: int = 2) -> List[Tuple[float, float, str]]:
+def diarize_audio(audio_path: str, pipeline, num_speakers: int = None) -> List[Tuple[float, float, str]]:
     """
     Perform speaker diarization on audio file.
     
     Args:
         audio_path: Path to audio file
         pipeline: Pyannote diarization pipeline
-        num_speakers: Expected number of speakers (default: 2 for counselor-client)
+        num_speakers: Expected number of speakers (None for auto-detect)
         
     Returns:
         List of (start_time, end_time, speaker_label) tuples
@@ -67,13 +67,30 @@ def diarize_audio(audio_path: str, pipeline, num_speakers: int = 2) -> List[Tupl
         return []
     
     try:
-        # Run diarization with fixed number of speakers
-        diarization = pipeline(audio_path, num_speakers=num_speakers)
+        # Run diarization (auto-detect speakers or use specified count)
+        if num_speakers:
+            diarization = pipeline(audio_path, min_speakers=1, max_speakers=num_speakers)
+        else:
+            diarization = pipeline(audio_path)
         
-        # Extract segments
+        # Extract segments - handle both old and new pyannote API
         segments = []
-        for turn, _, speaker in diarization.itertracks(yield_label=True):
-            segments.append((turn.start, turn.end, speaker))
+        
+        # Try new API first (pyannote 3.x)
+        if hasattr(diarization, 'itertracks'):
+            for turn, _, speaker in diarization.itertracks(yield_label=True):
+                segments.append((turn.start, turn.end, speaker))
+        # Fallback for different output format
+        elif hasattr(diarization, 'get_timeline'):
+            for segment, _, speaker in diarization.get_timeline().itertracks(yield_label=True):
+                segments.append((segment.start, segment.end, speaker))
+        # Direct iteration if it's a list-like object
+        else:
+            # Try to iterate over the diarization result
+            for item in diarization:
+                if hasattr(item, 'start') and hasattr(item, 'end'):
+                    speaker = getattr(item, 'speaker', 'SPEAKER_00')
+                    segments.append((item.start, item.end, speaker))
         
         print(f"✅ Diarization tamamlandı: {len(segments)} segment bulundu")
         return segments