Update app.py

app.py

@@ -252,6 +252,17 @@ class AudioFeatureExtractor:
                     Overall confidence: {confidence_metrics["overall_confidence"]:.2f}
                 """)
             
+            # Calculate pauses per minute
+            # Get silence frames using RMS energy
+            rms_db = librosa.amplitude_to_db(rms, ref=np.max)
+            silence_frames = rms_db < self.silence_threshold
+            frame_time = self.hop_length / sr
+            pause_analysis = self._analyze_pauses(silence_frames, frame_time)
+            
+            # Calculate pauses per minute
+            duration_minutes = len(audio) / sr / 60
+            pauses_per_minute = float(pause_analysis['total_pauses'] / duration_minutes if duration_minutes > 0 else 0)
+            
             # Add confidence metrics to return dictionary
             return {
                 "pitch_mean": pitch_mean,
@@ -262,7 +273,7 @@ class AudioFeatureExtractor:
                 "monotone_score": 0.0,
                 "mean_amplitude": mean_amplitude,
                 "amplitude_deviation": float(np.std(rms) / np.mean(rms)) if np.mean(rms) > 0 else 0,
-                "pauses_per_minute": float(pauses_per_minute),
+                "pauses_per_minute": pauses_per_minute,
                 "duration": float(len(audio) / sr),
                 "rising_patterns": int(np.sum(np.diff(valid_f0) > 0)) if len(valid_f0) > 1 else 0,
                 "falling_patterns": int(np.sum(np.diff(valid_f0) < 0)) if len(valid_f0) > 1 else 0,